A long conversation about SpaceX’s valuation, launch economics, Starlink, xAI, data centres, robotics, autonomous driving, capital intensity, and the execution premium behind Elon Musk’s companies.
Henry and Pritam try to price SpaceX by splitting it into three businesses (launch, Starlink, and AI/data-center compute) and stress-testing each against public reviews, live back-of-envelope math, and basic physics, rather than treating the company as a single story.
This page is a synthesis of a machine-generated transcript of a private conversation between Henry and Pritam. Automatic speech recognition introduces errors, including garbled names and numbers (some corrected here only where context made the intended term clear, and flagged where it did not). Figures on revenue, cost, valuation, timelines, and technical performance reflect claims made in conversation, not verified financial or engineering disclosures, and should be checked against primary sources such as company filings, official product documentation, and independent reporting before being relied upon.
The conversation repeatedly separates launch (Falcon, Starship), Starlink (consumer, enterprise, mobile), and a newer AI/data-center compute line tied to contracts with Anthropic and Google, arguing each has a different risk and margin profile.
Falcon 9 is described as cheap, reliable, and already profitable. Starship is framed as the riskier bet: recent satellite launches are cited as encouraging, but its cost and cadence targets are not yet demonstrated at scale.
Numbers for quarterly revenue, run-rate, and year-over-year growth are cited during the discussion (including internally inconsistent versions across the conversation) but are not sourced to a primary filing or press release.
Data-center and GPU-hosting revenue linked to Anthropic and Google is described as high margin and a major growth driver, but the speakers also note hyperscaler contracts can reportedly be cancelled on short notice, which cuts against treating it as durable recurring revenue.
Pulling up UK reviews during the call surfaces recurring complaints: unreliable connections, poor customer service, unexpected price increases, and billing after cancellation, with no dedicated official review page found.
Both speakers converge on Starlink's strongest case being rural, disaster-response, military, and remote/maritime connectivity, a narrow segment estimated conversationally at a small share of the market, rather than a fiber/mobile replacement for most users.
One thread frames launch, data centers, GPU compute, and Starlink connectivity as feeding into robotics and autonomy across Musk's companies, treated as a structural bet rather than a proven business line.
A detailed exchange concludes that heat dissipation in vacuum requires radiator mass roughly ten times the compute payload being cooled, and that space-based data centers only pencil out if launch cost per kilogram falls far below where it reportedly sits today.
Because Musk's companies are treated as one interlinked bet, the speakers scrutinize Tesla's Full Self-Driving and Robotaxi claims, including skepticism about how a headline driverless-miles figure translates into actual cars operating without a safety driver.
The people who got here, even if I are, actually, in Prade, they went to 220, I think. So... It's, it's, it's coming somewhat back down good. So everything you look at with SpaceX has to be in relation to this value, what... It's all for each trillion. So it's like. Which is still a lot. Yeah, what makes this someone how trillion the company? What business is the name? And all of that stuff. So, when I send you, you have WhatsApp on your computer? Yeah, yeah. So send you a girl. So, I mean.
'Cause a lot of these egg crews are, like, to just say, anything evil, you know, this, and that do equities. It's, uh... I'm also an investment test.
Oh, gee, I just thought I could use it, just got any employees, and those guys, or you said? Sorry. You see how's Tesla stock?
Oh, yeah, not at the moment, but...
Did you need money when you exit it?
Yeah, yeah, yeah. So I invested it before the Model 3 model one. So it went off. 20 times. Even, this report from the New York Times, he, he, like, he wrote an article, coaches, who share on, as he quoted me.
Oh, really? That's good, yeah. That's good, like, so if I search your name now, you might show up in a Tesla article.
If you search for my name in Tesla, maybe the article is still there. He, um... He was, I forget what he wrote for. I think Robin Hood's newspaper or something. Interesting. Oh, interesting. But, uh, that is because I calculated all the business dreams. At that time, that was what, 30 billion.
And I was extremely confident that would be worth a lot more than 3000000000 because I understood being engineering, understood how batteries work, understand the shortcomings of the, you know, the car manufacturers.
I understand being part of the game. So, for SpaceX, when we invested, like, me and some friends in the Valley, they basically had large services. Okay.
Which is, you know, and...
So, what they call consumer government here is basically stalling. These two are stalling. So, these businesses are actually, so, so, load services are actually small now. It's like, if you see 648 million, plus 314 for the launch and their stuff. This is, like, under a billion. So it's pretty small.
Shit, that's the whole business.
It used to be the whole business.
So the scaling thing was a huge unlock for the features of them. Well, if you look at Scott, one thing is, like, 4 billion.
Yeah, the starting is 4.3 billion, but consumer, which is something like 12 million subscribers, is 2.5 billion, and enterprising government is 1.8 billion. That's probably, like, military and all sorts of contracts. That's certainly interesting. But that's 4.3 billion, this quarter.
It's very interesting that this business, yeah, this is the cost, and then this is the name, because this is to send the thing to space. It doesn't cost that much, but then it makes so much. manufacturer.
But this is how much they make. So this is what people pay them to launch this.
But it can be a loss.
No, they do it on a healthy profit.
Yeah, that's what I'm saying. Like, it takes just this investment to make this much. Yeah.
No, well, what do you mean? This is stalling. Yeah, yeah.
They need to launch the space. So this is the this is the key component. This is the key cost item for this business.
Yeah, for styling. Well, for the starting launch part of it, uh,
They, I don't know how they do it. don't know if loan services charge is starting. Or if they just say this is SpaceX. And so.
And they just put it in there as kind of cost interesting.
But they must be, they can't like, they can't like subtract those costs. So I assume they were coming here. But still, they are charging themselves cost price, basically. Yeah.
Yeah. Because, I mean, they're noticing that they charge everybody that they charge themselves, and then they charge everybody the same thing, they charge themselves, setting on the court. that they're offering them much services at the base price, that they used, or that competition is going to go to space for lunch. I mean, it was great when I was talking about it.
That's weird. Yeah, that's what you're saying.
That's The point is, um, the point is actually making is, um, I mean, they, they, they're predicting that we start sheep, that the kind of like production volume will be 10 X, what, what, um, Falcon is right now. Yeah. And then he was like, yeah, like, I mean, they opened, they just, I know a movie, he's, it was like, they basically almost opened sauce everything about like their rocket business, they keep the schematics, they charge the same reads.
No, I don't believe that. I mean, they can't charge people cost price, right?
I mean, not... Just say it's fair market reads.
Unless for accounting purposes, he's charging them. He's charging, sculling fair market rates, but it doesn't matter because SpaceXO itself is just, yeah. Which means that this doesn't make much money at all. if so much of their business comes to space size.
Yeah. And this is the, this didn't exist.
I mean, appetising is like X, basically.
Yeah, it's like, as we know, people...
But nobody cares about it. I mean, he's pissed off with a lot of people, so... Unlikely, exile diet is gonna do much. But this is really where...
This is where this is the madness.
A lot of the action is. But there are lots of questions. So I think the key to value, you know what the current price of the stock is, right?
valuation? No, what is that?
What do you mean? So basically, let's see.
Yeah, these guys don't know about that. This movie, he's just a bit straight.
This is much better. This yours? No. Yeah, I think you're just connected to the Wi Fi, and then...
You know, with clothes, yeah. No codes, like, innovation is the only more.
Oh, yeah, yeah, yeah. I mean, companies always do those things, right? I mean, this, I think it's just about that. Yeah, hey. There was a one on the... 22nd London position... I think it must, like, a billion dollars or something. Easter, yes.
Why did we publish this?
Or is it someone else?
No, this is that much about this. This takes because I'm funded by one time, they just take a list of to my PTA. Oh, wow, this is a, this is to my family, this is how. Yeah, normally years year, I already had one billion in a sense that the mandating speaker was done many, so. When he was there, it was Sky, everybody.
It is German.
It's at the end of here. Oh, the leverage, same old.
So they eat leverage, I just fucked up.
It's the same situation, yeah. Man, I didn't read that thing. I'm just reading it now. Yeah, a friend of mine and me, he would, we were like, we should start a phone, because these guys are like rookies.
I started to understand. The 3 things that kill men is like... and then...
But did you read institutional awareness back then? Did you do that for juicy?
So I have friends who've been done me for ages or phone number of years, displays great, this and that, and I was like, not too great. He's like, so what he does have is because he worked for opening and I... Yeah, he...
No, no, his wife was phantropic. Yeah, the fiancé.
So he understands very well what part of AI came from will be in demand. And that knowledge is super valuable. So...
So that means all those employees that those companies that have done some crazy sort of treating those equities.
If they treated themselves. I don't know if they have like restrictions though, you know? If you were to open it, I don't know if you can just buy people. You think so? I don't know. So in the banks there were, like, in when I was trading, you weren't allowed to buy certain things without approval and stuff.
Yeah, but that's because you work in the back. Yeah, yeah, yeah. But like if you work in a company and you know that they're going to be supplied. You know what the supply chain looks like your industry and your abolition industry and the outside. I don't think there's anything like...
You might be right. I don't know for sure. That what I'm saying. Yeah, I'm gonna check. I don't know if there's like...
Because it's possible that... Because he was fighting at too, right? It's not like he left opening, I had he was fired.
Yeah, I don't think there's anything wrong with the guy. I just think he made some very good errors on risk management.
Yeah, it's just a leverage. everybody says, yeah. And if you just waited 2 more days, you would have been fine.
But that's like neither, you know, there. Basically his public book, God wiped those. It's no more Go work fine.
So anything left is just the anthropic. And then as hopefully, speaking of anthropic doesn't appreciate from where they are now.
Yeah. But I would say that the scale that he has, which is understanding which parts of AI, will do well is a good one. I don't think that necessarily makes me more in buffet. So the thing is, you become a good investor when you can trade through all kinds of markets.
Over time. Correct. Although. Yeah, he's very... I mean, so I'm reading the thing. I read like 2 of the 2 of the essays. It's PD. I mean, for something between 2024, it was pretty good. I mean, accurately predicted like agents, almost predicted everything well. I mean, thing you couldn't obviously didn't get was open source and open source is like a tsunami for the whole industry. Open source society.
So if I had 4 times leverage. My 6 year return, 6.5 year return would be 2400%. So hold on.
So the leverage defense, our leverage means is like, you borrow money?
It means it's simple. That means you have a $100
You use that $100 to buy, let's say, Nvidia. That's like an unlevered trade, but leverage is uh, 4 X leverage is when you have a $100 and you borrow $300 and you buy $400 worth of Nvidia.
Yeah, so you borrow you borrow. Yeah, we were like, it's like 200% of what your capital. And so, and then when you burrow, I mean, I guess they give you a clause that sees your, and they know you're using it to trade. It's for trading. So the leverage is especially for trading.
Specifically for trading is given by your prime brokers, the guys who execute traits for you and give you the backup, those admin systems in the back end. And those brokers have authority because they're giving you leverage. to close your position before you lose money. They handle it. They can they can affirm that cash if they don't want to lose money, right? So they're like, uh, if you, if you have $400 of Nvidia and Nvidia is down 25%, your capital is wiped out, right? Because you had $100. So these... 25% is your...
So $300, they land to you. So they have authority to close the rest of the book, which is what they did, I heard.
Ah, they closed it. Yeah, that's what now I understand all import because he said it's such a matter in all import too. So basically these guys have the right when you have leverage, you just go into all your drinks and like close it. Yeah. It's basically that.
Yeah, because they don't want to lose money. So it's like you take leverage, but then you better make sure you just make money.
Oh, God. I do want to connect to that. Yeah. And the thing is, room could still have to connect to Wi-Fi.
Do you have to connect to the TV somewhere? That's is there anything like...
Have you connected before? No, actually I haven't done it. I think we woke up a time in the city.
Yeah, we work is easy because they give you a lake of sneakers.
Okay. Yeah, he said, let me see the last one.
Yeah, connected to first or large meeting, but it's asking for a code.
And he's saying that I tried to do this very well.
Absolutely. HDMI, right? Okay, it's the same. So, I mean, Zoom. Then what? Go to shadow, Zoom, WS. Share screen. Interesting.
So do you, I guess you can do it. I don't know. You can do a remove share from it. I guess like, is it working? Share screen. No, that's the setting. Use the one here.
Where?
Oh, cool.
So...
What's the school? you can do it, um, you don't even have to be in the network, because one I'm used to is it has to be on the same network. This one, I mean, you can do it on the internet. You can have someone's shit. That's what the hero is really still not here.
Oh, sorry, it's not.
Oh, it's James.
Where did you do that?
Your work?
Are you still connected to, um... Are you still connected to matcha partners or the laundry?
Uh, I'm connected to guest.
Yeah, yeah, that's... Anyways. Um
So why is it, I mean...
It's coming home. I'm thinking quite a good time. Just margin 55%. I'm pretty imagine, around quite a year and year, so they're pretty managing this. But I mean, I guess, to be honest, this business, I actually don't think that this is doing too bad, but the fact that you don't keep selling that it's the long term.
But when you say it's not doing bad, you mean relative to work.
I mean, just like, I mean, so to compare it, like, in the context here is, because he long keeps selling the business of like a long-term thing and he's not like optimizing for, like, quarter unquarter like metrics.
So the 1st thing is, these numbers are only important relative to the market gap of the company, right? Yeah, which is what you say, yeah. Yeah. So everything you see here, like you might see +66 and all the... If the if the company's worth, say, 50 trillion and you have all these numbers, they grab. They might be 66% greater than last year. But with respect to 5000000000 there, they're not. not clear on that. But it's one. 5 trillion. I don't know.
less than think of. I mean, how many companies are on, uh, trade companies are on the $1000000 that are like, cutoffs now.
You can't, actually, this is the same thing, man, basically. I guess we can kind of like...
They actually crash my computer. So we're doing for a few of the purses.
I crashed a computer. That's crazy. This is. Yeah, they, uh, see me broken.
Try again. You do fast clues.
Yeah, yeah. So, no, actually, hopefully, do something. Well, I've closed mine.
crazy, honestly.
J, I, yes.
What does it say? T, B, A, T, H. Man, you know, um, this is the last time we know, talking about, um, talking about the GPs, and this thing came out, there you are. So Miami is working.
What's that? Well, he was working. I guess mine should maybe just.
Let's try again. Yeah, it's a good one. So let's.
Especially on music, I wish I was screaming.
Maybe if you hang out here a lot. Uh, tomorrow, man. But I have my feelings about the guy. You don't like a guy? I don't necessarily dislike him. What the fuck, guys? Yeah. Oh, that's crazy.
Well, he's very Asian, from where are you from? Indian. Are you Indian? Yeah. I guess when I was born here. But anyway, I like that, just the name. I mean, I guess Italian, India, no, I'm not.
Boston, it is the... has a joke, I tell, my friends. The most prominent news, Sonya, my brother. So, actually, the original form of bossu.
This is Boston State.
Technical is DOFC. So the British changed it to both, uh, which is the USC, and my granddad, so he was both, but he changed it back to Boston. interesting, but some of the well-known poses were the guys started both speakers. Really? Yeah, Amaran Bose is then you have...
That's a European name. Yeah. That's an Indian name? I didn't know that.
No, so Indian, you have the Bozai Sanctuary.
Bosons, you have, I forgot all of them now.
I'm only up on the same, I mean, not just that they have a name.
Yeah, actually. Same wider family. Yeah, a particular region.
Then, uh, there's another famous, uh, scientist, uh, who was famous in the radio space. Uh, I think his name was Jagdish Chandra Bulls. He was there as well. So, yeah, there's a bunch of them.
His favorite is your default, right? You're not an Oaky fan. And then you're not my father. I travel $20 stuff, but I really easy. Okay, so... Okay.
Yeah, I was hoping it gives like a good one.
I assume you'll put everything on the last. Zoom the. I mean, the maximizes concern, I mean, cultures.
Oh, this is like super excellent. Uh, yeah. I'll, I'll just go into YouTube.
I'm very DCF one, because it used to pass it on, you started.
Yeah, so it's basically a DCF, right? It's a discounted cash flow. So basically is of the value of stock. is the present value of all the future cash work. of the company, right? So all you have to do is figure out how much money is the company gonna make each year in profit and you just come back back to the present. And that's that should give you the price of the stock. So whenever I'm buying. Well, not all the time. Because sometimes it doesn't matter.
You don't need a precise phone, but you just need a rough, like an endless case, 30 billion, but I'm pretty confidently going to sell over a 1000000 model wise, and I'm sure this margin will hold and this and that, and so it's going to be well, it's already been.
Because that gives them easy better, when you do the math on the on the Y, and then you kind of take what a, like a PU ratio of like a 2X, 5X, whatever.
Yeah, so I don't use P, P, and all, P, like when you have a value. So if you have the price of the stock, like space X now, Firstly, a lot of stuff don't have to be ratios, right? Because you're being lost. You don't have big ratios just black. Yes. But if you have a profit, that's going to be very good. comes into into effect. And that's why like if you look at companies like Tesla. But it's just a very simple thing to realize whenever you're looking at stock. It's the value of all the cash flows.
Yeah, yeah, that's interesting. So listen, you see this kind of model. The Dennis one is, you need to do it. And the system is common. So this is if you're profitable and then I guess this is the best comment.
PCM. It's good enough. That is what it is. I don't know why Flores giving us other stuff. It's not... It's so relevant. It's like, it's the present value of all future cash flow, right? So, basically, when you're one thing you can do is if you start a spreadsheet, and you write down all of SpaceX business lines, and then you write
down how much revenue you think they'll make for the next 10 years, and you discount that, you get the stock price.
If we do that now with that mood for this kind of, for our business, I guess you would, right?
Yeah, it could. It would be a good exercise.
You just don't ask that your that line graphic. He already has that idea, right?
Yeah.
Oh, if I ask people. I think it's people from very good stock analysis.
No, I mean, stop. Just ask the dumpo there of kind of the business line, just cheat and then do like a 10-year projection bit on that recently.
Have you done all of the, no, we still have to project it forward. Did you maybe do that, you know?
Just tell it to project that based on like the innings, like the recent enemies they did, like, what was there? I mean, they didn't project that much forward. I feel on, um, space sex.
You don't know anything about SpaceX.
So what is 700000000 ARR? That is analyzed, right?
Yeah. So I'm just saying is 100 billion ARR is they're gonna make like just over 8000000000 in a month and that's gonna mean a 1000000000 AR.
But then you wouldn't do that into the other month. Right now, they're making like, you need 4000000000 in a, in the whole part. So they made like, so this was $1000000000, um... So yes, is that 8 billion. So they make 8 billion this quarter. Yeah, the city was Italy.
They made 8000000000 this quarter, right? Which is roughly 2.7 billion. Yeah. So like maybe they're on a run rate right now of 32000000000 or something. And then you've been projecting that. Okay, we grew up. I realized that is like +92.one% year over year. Yeah. So, um, actually, let me bring this draft up. uh, Let's see, right?
So, I mean, I guess, okay, some of the sheet, right? I mean, I don't know that we can, um, so you remember kind of what we wanted to do is be able to kind of like, hold, go down even like a fundamental part of like, what's the call business? Yeah. I think it was lunch. It was lunch. I think I want to do this, right? Because we're kind of like one of these, you see bomb shows, like, it was like, because we wanted to see if they could, what would happen in the future? Like, is there like a really, is there like a robust business there?
Um, in launch. Oh, launches the place of that. Yeah, I mean, if they lose money on launch, that is a competitive advantage, but they won't lose money on launch. So that's the thing. So let's open let's do SpaceX valuation, right?
Good order. You pay Gemini for this, you pay them, you pay for the workspace. No, no, no, this is a private account. For what? Because they give you the Gemini as a fashion. Do they use everything?
Oh, this is actually, this is my account. No, but it's fine. No, we have a company on as well.
Yeah, you would get that. I'm talking on lesser things, right?
It takes something, you know, it's not much, though.
You did for AI company.
So the thing is not about email runs on Gmail. Oh, where are you running? You don't want to drive with me, just in case... Where did you use work space? So we use like a, I forgot the name now, but a private ISB that's not very massive. It's like small to mid-size. So that, so every time like Gmail or something goes down, we're still like operating. So all these are like packages or whatever.
So you don't have to pay for workspace because for Gmail, use it for something else. Yeah. interesting. Okay.
Because I mean, generally not here. I mean, generally, let me ask you that, but it is, you can use it to ask any of these questions, and, you know, it's a special area line.
I can put all in Gemini.
Oh, okay, yeah, so like Gemini, right? This is Gemini, right? So there you are. You can ask the Gemini, anything? So we can busy.
That's interesting. I don't know if personal, you know, personal account of pay for an event. Should I try it?
Yeah, try it. I think I think I give it to you free on the personal account. You don't get it.
No, actually, this is better, so let's go. Let's put this head in so that he knows what we're doing. This is even better actually. So we do business line.
It's not typing, what's going on?
But the general, I think you click, do anything.
Oh, no, it's the same thing.
disconnected.
What the hell? That is great.
This is very high. I get, like, how they get, like, 20% of this, huh?
I go one giga home. Shit. But I know why. Oh, it says reconnecting. Is it because I'm on the guest network?
No, but that's what I'm on. I think he opens up. Okay, this is...
disconnecting, share this, share this, I guess you, this thing comes on.
such bad life But you should still be on the type, right?
She needs type. I think it's, um, is Google what you are, basically, you are, um, you don't have a flying name. So when you're online, you just, uh, like...
Anyway, let's go back to this graph.
Can you see it? Yeah, I can see. Yeah, so this is a quarter revenue.
This is like gross about it. So the question is where is this going from, right? Yeah. And most of this is coming from consumer enterprise, which is stalling. Yeah, right? Some is coming from launch services, I reckon. This is probably losing money. This is losing money. Uh, you think the AI data center business is a the data center is a super high margin, but they have a contract with anthropic that only kicks in, I think, after June or July?
Yeah, pissing next month.
Yeah, so then it'll turn. But even then, even the preliminary agreement they have is generated to most of that 2000000000 is coming from the, you know, business. And that is...
They've got like a Google Choo and then that's kind of like, I mean, yeah.
So here my very high level issues now, do we get into the detail? Like launch business, great. um
The Falcon business is great. They, they, one of the cheapest launches in the world, most cost-effective, highly reliable, highly good. Starship is yet to be proven, though everything is promising so far. And they even, you know, they launched 20 G, uh, B3 satellites, uh, last week. The caveat with starship is, um, they have to be able to launch it at a frequency and rein at a frequency, but I think the, the figures they have to get
the cost on the $200 a ton or kilo or something like that. I think, okay, look.
But I think that's basically what translates like one month a day. It has to do 11 starship lunch a day. And we, um, you know?
I think we have to do more of that to realize the data center plans, but they have like 15, 20, one launch a day from each base, but they have lots of bases. Right. Right. The question, so the uncertainty over the launch services is, is carship going to work, will it be able to launch whatever number it needs to perdain to get to their data center, their data center plans.
And will it be able to do so at a cost? Because even if Starship works and launch it, If they can't do it with that magic cost of $200 a kilo, it's not going to work. So that's the 1st bit of uncertainty with this part. I'm not entirely sure what they classify as launch in development, but I guess you can consider these together.
Lunch and development, that already makes money. Yeah, because I was going to say maybe some sort of R&D around going to Mars game space, but yeah, that's the revenue generating.
Yeah, generous revenues. There must be some, but we can just put it on the launch. Um, the, so each of these areas are huge deep dive. huge.
But if we stuck today, then we get back home. Yeah, because I don't think you can do it. I know if you want to.
How long did we go? Sorry? No, we go. Oh yeah, I got, well, you can go through it. I got time. Yeah.
Oh, shit. Tell me, Mrs. Cole.
Stand by.
Is it working? Is that working? Um, he's asking you, do you have your own mother days?
It's pretty bad.
So 1st year I gave very, I said, just populated with the current info.
Are we Gemini? Alone this, yeah, is very stupid.
What is this? Which one is this? Do you know which one you're using? Is this 3.6? No idea. Can you pick it? What is this? I think it existed basically, so that's what's in there. This is weird, man, that's not it. So they don't tell you, but you're easy to change, I know. It doesn't take away easy. Microsoft, then, Microsoft, So announced a new set of models, like, I mean, it was 2 weeks ago, last week, that was pretty good.
these models, like, seriously.
I think it's pink games.
Well, firstly, I told it not to do projections. I said, just put the actual data interesting number.
Yeah, actually, never East, um... I mean, Germany is pretty backward now.
They're all done. We tested all of them. For finance, for finance.
Yeah, I remember you said finance.
So, I mean, I'll do it for Florida. Let's see if Florida is the only thing.
I'm sure it's much better than Gemini, but like, you might not meet like an actual like, like an actual Xbox standard.
But anyway, while I was doing that, like, all of these business areas are huge, huge areas. Uh, to deep dive into.
Yeah, but we say we start with them, that's why we launch it. I start with the corn dessert from the most exciting. It depends, right? It depends, depends on what's styling is, from the way you protect the styling. If everybody, because they're going to go African humans too, right? I mean, mobile, staling mobile.
Yeah, so Staling, right? The business is divided into, uh, in simple terms, I just came from it as starting, uh, enterprise, uh, like consumer, uh, and maybe you can have Stalin mobile there.
Okay. That's the new, the one they bought, the telco, that they put, that I got to start buying finish.
Yes. So, yeah, so they bought Echistoft. So if you want to understand starting, I feel you need to have some understanding of the engineering behind these communication networks. So I have a friend who's like a age team in like satellites and antennas.
So he designs like antennas for, uh, you know, uh, uh, car companies, uh, like Ten Audi, a few hours. I dont know if his game, but what the groups... the antennas that connect into the internet or they connect to satellites and they connect to mobile base stations and stuff like that. For one.
So, you know, what in the car internet?
For the car. Well, every call you buy now is connected to the internet, right? 1st needs to track your location. So it needs to plug into satellites when you go into McDonald's and it needs to plug into, you know, a mobile network of GPS when you go through all this stuff.
These all different antennas or they have one antenna?
Oh, I don't know. He would know better. Right. But basically, last time he and his police have described, you know, bentally around London, but they have to do it at night so that people don't know. Um, so I met them, but uh, a lot of them, there was a big into the back screen and stuff like that, and to the back screen. So you know when you see these weird lines in the back windows? Those are tennis? Those are I'm done.
Yeah, because what else would they be, right?
Because the gods doesn't need to have those lines.
Yeah, yeah. Well, I think sounding are also for keeping.
Oh, yeah, yeah. Heating is money. Heating is what I like about the way in the past.
But they hide it in all these things. Like, even the act, like, you know, the famous the antenna. There is no antenna, but things like the frame and all that. That's why they had antenna game when you have the phone is starting away. Stop working. So, um...
You're talking about, you're talking about, uh... Okay, so 1st education is the way that this starts calling. Yeah, it's not speaking.
Yeah, the 1st thing about stalling is what they're saying is that this year they have pregnant Indian subscribers. We've got many subscribers, which is double last years. Sounds great on paper, right? Um, But like, if you actually Google a few things, like, let's say, let's do, okay, let's do study for this.
Oh, okay, back. Oh, shit. Yeah.
2.one? But this is UK. There's one from the US as well. So she actually read reviews for Stalling, yeah? Uh, let's see. As we are important, see all of them. So basically you see the top, what are they said? Let me say, most reviews are unhappy with the experience overall, many customers express significant dissatisfaction with the customer service, serving a non-existent and helpful, difficulty reaching in person, AI products, numerous issues with the service
and product, including culture equipment, unreliable connections, unexpected price increasing the jobs, for service aggregation. So it's what like any product I do, I look at the reviews.
Is this universal when you cook the American one is the same thing?
Yeah, so you go to the American one as well? So this is the UK one. So the only reason you have this is because there's a stalling installer, um, that, you know, comes and places your dish and by proxy of that installer, people have reviewed stalling, but otherwise stalling doesn't have an official page or anything to review it, which that's almost interesting.
It doesn't have enough one.
Sorry, doesn't have enough one. Starling doesn't have its own official review page.
Ah, like don't they install it themselves? So they always use a 3rd class between them.
No, you could you could go to Stalin, and they send you the dish and you saw it yourself. Oh, you can pay 75 quid or 100%.
So this is for the distributors, the defenders is fine. This is actually the review to Staling directly.
This is actually a review of Stirling. But it's the vendor's page for the installation. But it doesn't have a page.
Why can't people, this review says, don't let you just do that company and start reviewing the company. The company has to create the patient themselves. Anyways.
Yeah. I mean, the U.S. wine.stalling.com, right? But the UK has this week, guys. So maybe some 40s, I guess, 6 days ago. Yeah, so I wanted to study many of my home was on a European view. So before this, Arcadia is stalling with the most likely early customers are stalling, but people in like rural areas or areas with sketchy research, because or no broadband coverage.
Yeah. Because, I mean, if you already have like for like one terrible, one gigabyte or whatever, it's the, like you don't need to buy styling, but you're fine.
Yeah, yeah. I mean, I have one HD. It's a big 35 quid a month of home. Why would I change to anything? Um, so...
But that's so good, man. Who is that? budget. Shit. What do I have? I think I have sky. I have like 150 and I think I paid probably like 35 or 40 years old. It's, uh, you need to find it. And I used to have a virgin. And when I didn't live in, like, Bromly, I used to have virging. I think I have like, shit, man, I'm gonna go get that.
Yeah, but I have to do a dance every year because I've been a Virgin customer for like 10 years now. And they always increase it, like, ultimately we just 45, 15, like, I'm leaving. I'm going with these guys. And then they say, no, no, no, we bring it down to 30 for you. So every, every, every 2 years of my contract renew, this happens. But if you look at the reviews, right? So one thing to realize is you want to understand the theory behind how does the satellite provide the internet, right?
So you want to look into Obviously, you have a huge advantage. They have wrong history on societal life. And traditionally, satellites were used as a fill-in for no what they call dead zones, right? So when you don't have like the mobile networks have already built like beautiful internet coverage. It's your show.
Yeah, like in north, stock the cover, like most common areas, and Virgin and all these guys have been like beautiful, cheap underground cable fiber networks, which are very high speed and pretty cheap, right? So the question is, what's the idea behind Staling? And the idea is like, if you've seen stuff from the dog, then everywhere on the planet, we have access. And that's why when you have wars or natural disasters, starling, starling will still work. If you're stuck in the ocean or some panic towards starting to work.
So that's a that's a huge case. But it is a niche case, right? Because most people won't care if for 90%, they just cases, uh, they don't need starting. Yeah, that's my theory. But I think there's a 10% to 15% premium market where people would be like, especially and there's a huge enterprise market, government, military and all that, where it works well.
Because particularly, you can see the, you can the American government and like your party. You're jumping into like Russia, just some like recognizance mission. You don't have internet, right? Like you need like something for me, in like sometimes when structuralized, they kind of like be maintenance to you when you're under it.
Yeah. So the 1st question for any product revaluating is, is the product actually good, right? So it's really weird. Like, John was a lot of founders, so the engine investments I made or whatever, uh, a couple of founders told me. I the only person who actually downloaded and checked the product.
Because we would just write the check without knowing like...
Yeah, yeah, just like theoretical. Yeah, this sounds good.
Theoretical. I mean, I do that, though. My product founders. I mean, I mean, that's like very, I don't have that much money. So I invest in my friends. Yeah, yeah, yeah. And so I always look at the product, I need to look at the product, I need to understand it, I give you real fit back. Yeah. So, So, for me, does the product work?
Well, let's see what they say. At no point during the order, you can service it by the KAV of the northern sky. The website repeatedly refers to checking for obstructions, which naturally suggests trees, buildings, physical obstacles, not that an entire direction, the sky is unusable. This is hilarious, man. Have this fundamental requirement being made here before the purchase. I know we have ordered. The main matter is worse after canceling the service and opening a former complaints, stalling still debit did a further 40 points from a bank account.
Technology may be impressive with the custom experience being disappointing.
So people are not, they're not complaining about the product. They're just complaining about the customer service.
No, that's complaining about the product. He's saying that because they say on the web, obviously satellite only works if you are outdoors facing the sky, right? It's not going to work inside here. So he's saying he checked. and his, wherever his whatever house is, it's facing the whole northern the sky. But it requires the northern sky or something like that, which he's saying is weird. I am facing this guy, but nobody told me. I don't even know what that means. So he has a fair point, but it doesn't work. They didn't inform properly.
So this is terrible. You expect a good connection. Nope.
It doesn't matter a lot.
But the choosing came and did, but choosing, you know, only, if it was coming up before my baby, every month, you know, physically changes. So, okay, so that's why...
This is the 2nd one I'm seeing, giving charge ahead of the payment date.
Payment data, he's also made another point. He said you'd expect a good connection. No, so he's training, even the connection is not good. What a bit of kid, literally plug it in, pointed at the story for Wi-Fi.
It's going to give it 5 or 5, yeah.
You know, no passwords, no portals, 5 bus straight away.
Is that how easy it is to use disconnected? and that's it.
Well, he's like an exception. If you look at the single?
Yeah, he's 95, I think. 32%, 57%? Oh, there's lots of 5 actually.
But 57% one starts like 15.
Yeah. It's pretty low for...
Maybe it was for certain things. Great product of a customer service, like all the stallings. So he looked at stuff and obviously. Probably the worst customer service experience. Talking to brok is the only customer service option.
I mean, I have this experience. I'm having this experience now with WhatsApp, because I'm trying to set up a business account for my agent. I've searched up a WhatsApp account, but I want to use it for my agents, like I'm an agent with WhatsApp groups to me. Um, what is this one?
Seeing 60 MB download, but one MBO download.
Yeah, that's pretty bad. Stan V is not even that impressive. Um, so yeah, so with Metcox, I only have, I can only talk to an agent. Like, there's no, I mean, like, hey, I want to talk to a human, and it can't solve my problem, and I still can't talk to him. But I get it. I get it for WhatsApp. I get it. They can't give every new customer a human, so, um, Okay, let's check up the American one. Yes, it's feedback. So, yeah, that's styling. I feel like the your face styling is the business.
Starling is very comfortable business, yeah. Like, it's the business.
When you look at space, like, salad is the business.
Um, No, the data center is the business now. But it's very
Yes, this one, I think. Sterling development.
It's cost stalling, man. 600 reviews. Yeah. How many percent, let's look at the essentially stats again. He's screwed down a little bit. Yeah. Yeah, you're going. Shit, it's the same 60%. So this is this is 5 to 70%.
This is the US, right?
So, you see, it's very, it's very bad though, right? It's like you can do like the bulk of it is a negative, but it's almost like the same metrics. like 60%, 20% on like 5 spots. So people are like extreme. Either I really hit it or they really like it. And then they're vegged in the middle.
Yeah, it's what trying to make sense of this. Uh, so... What is the story's been you, like, want to make social C or see, you know? I guess that's some other thing. Cannot change. That's good, man. Check my mood as missing a free mood device.
So this is still a green? This is ops.
Quite simply the worst internet provider in the world. Constant speed drops, terbo customer service. It only works outdoors. They charge for a service that was supposed to be a 30 day trial. I'm sending my back.
whatever. This is this is last week.
Yeah. I got the starting mini with the wrong packaging country. This is a good one. Always a cheese, excellent speeds, an office, frustrate stability. The odd thing is, although it's slower than my one GB home drawdown, Starling is offering 250 and BPS, it doesn't seem slower and works well for running the business and also streaming the voiceover.
You know, one thing, I wasn't used to me until I read the, so I listened to the thing. It's like if robots and agents become the bulk of how we do things, like internet uses gonna like an X.
So that's true. So that is why the Tesla SpaceX merger looks interesting. So the one argument, 1st stalling is, we go through, we go to the consumer vision, stalling, mobile, everything. But if it becomes a vertically integrated company, where they can launch, they can run data centers, GPN space, and they can provide the internet for their robots to operate. So that is an interesting, right?
So the big question is, are we ever going to see a lot of those things? Or does it make sense? Or when will we see it?
Yeah, I mean, I mean, I think every 2 thing is, it's really interesting, why not? I think especially... It's just fine.
I think it's a question of when I don't know if there's an advantage. Because so many people are competing for it now for which part? For robots, especially.
Man, I think China is the only one that is like value. I mean, they're like, what, two, 3 or that big American companies like the bread guy.
I don't think he's going to reach right on. Red guy? Yeah, my friends an investor. I don't know about it.
I mean, they're the only American business that I'm not from Tesla and them. I don't know any other ones.
There's some British companies as well.
Yeah, I know that British is like some European one, like the guy who wears a cat, new, you guys are making new, right? Yeah. Are they American or they're Dirokian, right? Because I think they're having Europeans too. Even though they're positively in the sound of sound consistent. You know you, right? No. They were the ones that did the pre-order last year, new robots. Um, So they actually, they're the only 3 I know, this guy's. One X. You don't know what else? So this guy's did a Pyota last year. Um, I mean, I almost thought that. They had a Fyota.
lets say, this one. So my big problem with AI.
So this guys were, I charge him like 23 for this. Yeah, yeah. And then they had like a tour.
And then you could use...
So what is super, I think I'm using it this year. Um, but I think we still see that bulk of it doesn't work, they can let the tellen to it. And then you can do a few things, autonomously, but logically they have a, someone sits somewhere and then like control it till it gets good enough. So their thesis is that using Tesla is got good as South revenues, so they drove on the rules for a few years and collected a lot of data. But they need the robots in the house.
But that's what's very far, so... When I say very far, let me correct myself. They've been trying to practice for 15 years now. last year. It became better and it is an amazing assistant, especially the new version in the US. Like, I have friends who say, yeah, just steering wheels, close to work, comes back. That's amazing, but they have to crack the lost one% if they're going to have over that soon, right? Yeah. And...
The one percent of these always has problems, right?
Yeah, well, especially AI. So if you look at Tesla's stock price and the earnings score, so what do you use this? This is our names, right? Default stock app? I use Apple stock. I have another one as well. Main Street data. That's that's even better. But it fell off a cliff, right?
So the truth is, I think partly what the market is, firstly, my guess was started when I was an investor, et cetera, uh, was like, and I was going to buy EVs, going to go 50% a year for the indefinite future, you made a... It did that for 2 years, was amazing. That stopped. It like stopped growing and even reduced. But the Robotaxi, which they took the balance.
But the phone is just recent, though. Like, what happened?
That's not the earnings. So what they said on the earnings call was, the guy was trying to mess it up as a great entertainment. It's like we've done 380,000 miles of, and this is the lowest they've been in the field in the... But in 5 years, that's what I haven't been going anyway.
So anyway, something low has been in the field. They're really bad in, um, no, not really. I mean I got that. This is my thing, yeah. They're really bad. So if you go on, this is 5 years, it's 5 right now. Yeah, yes.
To play, yeah, it's true. It went even lower. a couple of years ago. But it's been the lowest for a while, I guess, since, what is that, 2024 or? Yeah. So the thing is, uh... That's I said, gonna go 15, the reason I got out of most of my Tesco stock was because the reasons I originally invested was sustainable energy company, uh, EDs make sense.
It's going to go 50% a year of confident, uh, deep crack, you know, mass production of the model 3 model Y. But it grew 3 years, 50s, and then I stopped, and I was like, maybe this will happen on a longer timeline. Maybe all the early adopters of both their English. I mean...
So, so the, the, the, any of those adults. Yeah, you're not going to keep great because you're not...
But the biggest thing was, Elon said, we're not going to manufacture $25,000 car, we're going to, he shifted the goalpost for us to Robotaxi, which is a bold path, and I'm cool with it, but I cannot...
So he's betting that people are not going to buy, buy cheap cars that they would drive. It's really...
He said that so they had a design for a $25,000 car. It was an amazing design in the process for manufacturing it was a lot cheaper. It was like this unboxed manufacturing technique they talked about, which is way better than the original manufacturers. But he ditched all of that and spent on Google Taxi, which is a 2C the car with no driver, right? Mostly anyway.
And he's gonna, and is he selling it or is he gonna rent it?
So his idea was he's going to sell it. plus they're gonna have their own network.
Yeah, and then you put it up on the network.
Yeah, how much is this? How much did he take? He said it was $30,000. But forget back, to do that, you 1st start to crack self-driving, completely, to level five, right?
Yeah. So you're going to wish that if you put that out at least one or 2 generations before the river taxi.
So the reason I sold my most of my Tesla stock was not because I didn't think it won't work or whatever, it was just too hard for me to figure out, right? And if something's too hard for me to figure out, I'll take a pause, like have a look and come back later. So I would address if they crack there. absolutely. But the reason the stock fell off a cliff here. is because the, I don't know if it was a CFO, the head of AI. He was trying to dress up. He said we have we've given 380,000 miles. Let's see if I can find this for you.
So, I wanted to, like, the, you know, Robotaxis here. You can check by Bay Area Dallas. Is that the number of banquets? Yeah, but keep in mind, these are not real over taxis. They have people in the car.
Let's see. So basically, what he said on that call was... Yeah, so he said we've taken 385. So the Robotaxis, I know that Robotaxis is a lot of more and more wise, right?
The only miles that are important to me are miles driven without a driver or a chaser car, right?
Which wiggle house, right?
Limo house, yeah. The label uses a different approach. for doing it, which is more root force and having a city rather than other stuff. ?
Hello?
Yeah. Yeah, yeah, not bad, not bad. Uh, listen, I'm just, uh, discussing some things so funny, but yeah.
So... He said that we've drew 380,000 miles with no driver in the car for Robo taxi. Um, And if you actually break that number down, keep in mind, they started with over taxi trials last year, June. It was being over a year already, with them promising road I've seen.
And if you actually break down 380,000 miles for the time period, you said, whatever months, it works out to 26 to 36 cars roughly. So what he's telling everyone is over this whole year, uh, we've only had 26 to twice because playing trips without a driver.
And even then, people say that Jason...
But I mean, that doesn't make sense because you just shouldn't be, like, earned it or something.
Yeah, so the distinction is, HSD, unsupervised is different from robot taxes. No. There should be no driver in the car. A lot of these cars, like Bay Area is not allowed.
So the 300, the 23 cars are seen, the ones that don't have any drive. Any driver, right?
So some of them have a driver in the passenger seat, and some of them have a driver in the privacy. So in the Bay Area, regulation doesn't allow you. So I think there's a driver in the driver's seat. I believe in Austin you are a lot. But yeah, 16, 16 unsupervised. So, And then you can see the cars as well. These are many model wise. So, yeah, I'll never go taxis. I mean, it's a small number.
It's a control number. They, you know, they have some, like Dallas and Houston also allowed. So they have some. But it's like 26 cars, right? So it's not cracked. So they have a manufacturing process. So, I don't know how many 1000000 global taxis. So, and you know, I think maybe she has a bad value. Oh, really?
He said if he's, yeah, he said if he sell a Robotax, he'd do a customer before 2027, he'll shave his hair.
Does he have any hair? Oh, yeah, okay.
Yeah, yeah, yeah. So, yeah, that's that's the best story. So there are all these promises. When you when you evaluate space, you need to consider that not everything comes true.
not taking anything away from Elon because you have to have big goals, right? Yeah, yeah. But there's a difference between aiming for huge stuff and it actually has any reality, right?
Yeah. So I guess like when you look at, so if you look at this, the coffee is in the business is this one. Um, so yes, this is where we kind of excited. I mean, I can like project out. And then this is a nice big one and then this one would always be there and it's kind of like proven that they can make it cuter. This one is just, but there's just not a lot.
This one, this one will also increase, launch business will increase as they send rockets for data centers because they will charge large business.
They charge. They're charge, what is today?
Yeah. Yeah. But that's internal, so it doesn't matter so much.
And then if the move, they, people who actually opens, then you're kind of starting to see the distance kind of the business in the building.
Yeah, but that's a huge thing as well. Yeah.
So, but the biggest, the biggest deal win is, is the Estonan is space. So that's going to be, the biggest, huge case, launch of this. So this 2 are very complimentary, right?
According to what, Elon's 1000000000 that he's talking about, from my rough estimate, 60 to 7000000000 of that is data center. But the problem with data center, right? So the thing is the launch business is a unique business. It's hard to copy. It's hard for anyone to launch rockets with that cost. Japan's been trying for longer than you know.
Just blunt the rocket now, huh?
Yeah. He has a he's he's lagging behind. That is a unique business that is hard to copy with a mode. But it's only like a billion. But they can use that for their own advantage to do other businesses. Staling. Staling has starting enterprise, staling, consumer, which is broadband, basically, and starting mobile. Correct. We look through the consumer things.
And then the mobile is, um, the mobile is and come out yet. Yeah, we saw the cars, you might... People are unhappy, but if you want to, you know, really making a case for enterprise, for example, the cola, it was like, oh, you know, they're really going to ramp up on enterprise, have like a backlog, uh, because the ground is coming very well.
So one thing I would say about this is, uh, I spoke to my friend who's who is the, my ex-boy, I'm friends with satellites, and he said that, uh, so 1st he, they, they launched B3 satellites, right? Yes. So they still have their beat. They're not going to, that was just a test, but in a few years, they start shit gets up and running at the cost they want. It will be launching V3 satellites. And Elon said they plan to launch a thousand. Yes. But a 1000 satellites, how many connections kind of support, right? So actually requiring us plot this.
How many how many satellites do they have right now? Okay, I've wanted you in space.
But the thing is, the satellites they have now, you can see the feedback you get it, right? It's slow. It's not very fast on the internet. It's fluggy. So let's ask... Please...
If they launch a 1000 B3 subdivides. They pick a look, post.
Oh.
And how many broadband? and mobile connections scan of support.
Can you check the, can you check the one out? Yeah. The DCF I try to do, let's see if it's any different from the... Okay, this year. Did he get it?
So give him 47, 477 billion. Applied share price, $40.
I should. That's right.
Yeah, I don't think it's wrong, man.
going on? So this is 2026. What does this eat? Okay, this is launch. But it's giving it less time. They made 4 million, right? No, they take this much. Oh, this is projected. It's revenue for the end of the year. I mean, this is for this year. Okay.
What's the same for starting, 17 in the year? That is, well, it's not actually pricing any growth. I think that would be closer to 22 to 25 million.
It's pricing. So pricing for like 40% root month per month.
So it's taking right now it's 4.2, which is roughly 17 billion, right? So it's saying there'll be no growth. Thank you, 3 and four, which doesn't make sense.
So this should be good. I think it gives you a slight out, so very slightly a little bit.
Let's see Oh, there's a slayer. Yeah, yeah, yeah, you're right. Oh, that's cool. So that's 25. Okay. Launch to each.
We're doing one thing again now. So I mean, yeah, so it's Actually launch, we could put the 4 billion, okay? No, but everything is doing 4000000000 now. So you're saying they wouldn't make any more? Didn't they have like a good bank contract, they're announcing the earnings like 6 billions or something like that?
No idea. They announced big contracts, so...
But that may not turn on yet, right? So, um, AI. And uh, no, this is way low.
AI is on 9 billion.
What are they on right now? is like, um, So, 18, how much does what are they getting?
So they get 900000000 a month from Google and one.. from Anthropic, right? So surely they get 2 billion. From now on. But then I guess you're seeing the entropy from now. And when does Google K came from? Because that's 900 million. So they should be making $2000000000 a month from...
I remember, yeah, that was them. I don't remember to start with, yeah. So it's just stupid in the month, and that's this video, and then I guess like if you have like 20%. 24 billion. Yeah, this is like, yeah, that's a good.
So now what's the same? So whatever, it said 76. All right, doubles.
It's like, this is what I should get. It's a useful tool, right? Yeah, and then there's an Elon, um, there's an Elon, uh, kind of like browning thing there as well, right? So, see, I say, buy some, just one trillion right now. I mean, at the end of the year, right? I mean, this is Can I share this with you? Yeah, where's the thing you have to go ahead? Here, yeah, I don't know. Yeah. Copy, club, or... Do this or that one? Yeah? This one? Yeah.
Oh, yeah, yeah, yeah. . Okay. So, it should...
Yeah, to do, yeah, for the player.
Let's see.
Okay, so if the launch are going to be 3 satellites, which is the best of the best satellite. It's still testing it. The reason they launch 20 now is just to see if it works and stuff. Great question. Can you call me between his lunches and cost? The physical setup, each week 3 is designed for one team, GPS of downloading, and 160 GBPS of Apple. That is each satellite door, right?
At a mass of 2000, KG per unit, more than 3 times would be too many. So 3 kinds of size of its car, you know, which is why Starship North Falcon 9 is the only vehicle that can depoil our economically. Each starship, E3 launch, is planned to carry 60 satellites, adding 60 PVPS of network capacity. TDPS is a 1000 GDPS, right? Yeah. So... A 1000 satellites is equal to 17 Starship launches, $2000. Makes sense.
Cost scenarios, rumored internal transfer price, 15 to 25000000 for $200.
No. Yeah.
So everyone's just custom calendar, maybe. 500 million? I mean, I'm just picking approximates, I don't know, I mean, you know, 300 million, you know?
So, yeah. Yeah, some people are saying they won't reach 200 K per KG or whatever. Long-round SpaceX, dog, 100, 200, 200,
Okay, so the thing I'm more interested in is this. How many connections can the housing be free support? Because that's what, you know, it's saying. So role capacity. down link, roughly triple games that I can remember. But to support lots of users, Stali needs to have the bandwidth, right? Yeah
Okay, so the usable capacity will be 300 to 400 DVPS. So raw capacity is not equal to usable capacity. So they lose like 50%. Fixed broadband subscriber consumes maybe 5 to 10 and VPS on average if he goes, even on a gigabit plan. That's called roughly 30 to 80000000 growth game subscribers.
Yeah, so it's basically saying you need to get a thousand. Um, to be able to maintain nine, 3 times the size of what they have today.
You know, I don't know, because my friend said, um, He said, like, you have to go into the nuance. He said right now, starting satellite software on burst mode. So what that means is the data comes out in first. So let's say you are streaming something, which is more constant download needs. It wouldn't support it properly. So... Yeah. Oh, interesting.
I also find this number quite rosy because it's framing a 1000 satellites in sports 30 to 80 million. And it claims that a fixed broad can subscribe.
But also, the thing is also that the, like it's also saying that you just need 16, 17 launches to do that, all that. So you just need to do that 20 times and then that's it.
What's 17 times 60? Because it's a 1000 satellites, right? Yeah. So it's a thousand.
Can a fixed button, subscribe, confuse 5 to 10 and GPS on average, a week ago, is that after?
Because I mean, even though you have one gigabit, but you're not really always using that, like, Pastani, um, You know? Uh, PT hours, I reckon this is for me to work out, so this isn't like, uh, we can use treatment movies, right? Or do they need peak hours of internet? Is it because of the business or pecans of the consumer?
I'm guessing pecos is evenings for consumers. Yeah, right. But then what about business?
You have a business, you have a business. Like unless like you're a robot company or like you're streaming, like, I mean, listeners, you need to be documented, right? I need you in documents, right? And I'm like, they don't always stream me like HDV, right?
Even you goes on, like, high quality data, man, like, how, like, mobile direct is a very different profile, full news on satellite links, I reach, tiny, three, put, through, puts in.
What's the messaging going if I be there? I'm looking at 10% of the screens, mobile, and some other...
I just don't trust it's numbers. I think it's bullshit. It's not going to support 150000000 phone connections.
I want to see them again so far from 800 minutes. I like to dive in. That's too big.
Yeah. That is a fine...
These things are psychophantic, right? It's gonna work it backwards. going to work it back aggressive.
Okay good point. Let me see. Are you sure all of you, those, broad, the mobile connections? So the thing is,
Mobile also needs to work in those. And there is Skype, a limit to, and...
needed to make it work indoor.
They talk about an anime. Someone was, I came out, these people were going to play the dish on it, was very classic, complex, and back into the video.
Well, they claim that the satellite dish will act like a base station. I asked my friend and he said Nokia tried this a while ago and it didn't work for some reason.
But, um, So when, like, Vikash seems like, you need to have lots of people for the mobile to work where you need to have lots of people around you that already have the, the, the, the, the dish.
So they said they're going to have corporate dishes and styling customer dishes. So if you're a stalling customer, they're going to use your dish to do them all. Send signals, uh, and receive signals from mobile phones. So I don't know how happy customers would be about that, but, um, listen, something of Russia. Yeah, yeah.
I I have a lot of questions on the mobile network, because satellite, you know, as you know, it only works, if you point out at the sky, then we read all these reviews where people are like you, and so I'm happy with it now, because some skies don't even work.
They're gonna have to build a replacement for base stations. So all the networks we have now, they have like huge base stations to, you know how mobiles work, right? You have a lot of different case stations and your phone is connecting to the nearest base station right now and being in bad stuff. They claim, you know, like you said, that we're going to have dishes. Firstly, everything that penetration of stalling is that why. It's spelled many globally. globally, of which some figure I saw said 2000000 in the US. So I don't know if that's enough.
Maybe it is. But one of my friends saying it's something that needs to be really tested and scolded out. Spectrum is the binding constraint on the satellite binding, deep C, we uses slices of form.
The thing is, my question on spelling mobile is, I pay mobile is already cheap. I pay £13 50, 173B data. Good roaming, free calls of text, I believe the US is more expensive. But how much better can it be? How will they eat into the DVC market? Obviously, does have great incentives with package, broadband, with mobile.
I mean, I guess you can do that, and for this, you can do some sort of comparison between, like, how they competed in a broadband sector to see, like, is it, is this, I think, competitive? Are they competitors? Not really, right? Specific like they are. Because you're saying you get like your, I mean, you're so slightly discounted, but you say you get for, like, 40 feet, you get, like, basically get, like, 500 and maybe get, like, 300 and maybe, yes.
One game, one game, right? Yeah. One year. And for 2050, I go to 5 GB, £20, 50 for a mobile, 170 gig, uh, GB data.
Yeah, so that's where you went. So I guess I guess if someone is like, okay, how much did it cost today? and then how much does the brand bags cost? Um, I mean, I guess there's always some some premium you put on the fact that they are, And you don't know if AT&T and and TL, um, all these telco companies, it rain a bit as fast. Like, do they improve their technology quickly?
So what would they improve too? Because it's already very good.
Yeah, but I'm saying, like, if you do trust, like, is what I'm saying is like, because you're kind of wondering, um, see, you take Verizon, right?
The Mon pickup, I think, is 200 billion. I actually still coagnosed that. So this one? Oh, Visa, here. CC, market cap on a 9 billion, 12B. It's a low-E business. If you look at the 10 year stock price, nothing's happening. It's like down 15%. It's a stable utility side space mass.
It just, it has customers, has, I think, like, check the, I guess, one, T, AT&T. No, no, no, T, uh, T, uh, T, uh. The team while I was reading big. I think it's the biggest.
Teamoba, NASDA. Is that one? Oh, that's yours. That's team, uh, the scope.
What's the ticket? I think it's TN.
Is it not this one?
What's the senior note? I'm looking for the stuff. Team is, yeah, this thing. So, same thing, same thing. Slightly high PE. That's interesting. Uh, stock price. Whoa, it's not better. Oh, so you said this come from behind.
Yeah, you came from behind.
So yes, it did, it kind of like...
But you couldn't, they're like huge companies, uh, that like Proctor and Gamble. The stock doesn't move much. They low-key, they make money, they make profit, the prices are set. It's not a growth industry. They already have most of the customers. So I don't see how, even if there's no mobile works, it can add more than, not as well, stalling. It can add more than a 100000000000 to their market cap, a 100 billion. Yeah. At best.
Yeah, but last thing you would say, I wonder if we need to market by 50% of the biggest companies. ridiculous.
Yeah, yeah. So again, the thing is it's like one. 10000000000, right? And they're saying it's going to be worth 10 trillion. Like, why would you invest if you're going to go from one Australian to?
But it's crazy, right? Because like... Because he's looking to say, if you think that Tesla is worth one.5000000000 and then you divided across those 3 businesses, like... specific, sorry. Okay. And then you divided like across like those 3 businesses are like, what's falling gets most of it, like, right? So it's one, like, one, one, one, one, one, like, so one, video?
So already gone down. It's one. 45 now.
ing. Do they, yeah, they, it's just kind of joking. So, so it's like, so this guy is worth, um, what, like, 600 million? Who? Staling, communication.
Oh, starting on its own. What will it be worth? Yeah, inside, yeah, if you take out, like, if you use the revenue and then take the, well, the thing is stomach is right now, let's let's look at this chart. Yeah, Sonic is like, Sonic, it's 50% of the revenue, right?
That's what I'm saying. So...
This is what's 750 billion, yeah.
And that's like bigger than normal feet, like couple companies in America altogether. I guess this is a big deal with Tesla for good tonight.
But then it's also broadband. So which is the biggest broadband provider in America? Actually, don't know. It's not team at all?
One thing is, these are like commodity businesses, broadband and mobile.
I guess we're about to buy that in the world. That is, nobody. Comcast. In the US. Yeah, in the US.
Because they're doing US only for the more personal.
I mean, it's fine, it's still good. And I think competing in China? Everything is single because of a ranch for everybody, the U.S. convicted, usually talking about energy. But I'm seeing broadcasts. Yeah, sleep part fun, and I'm in your podcast.
Yes.
The mother is broken. Even though I, even that character, the prompts, it didn't change that out. What the fuck?
It's connecting, correcting you.
I was saying broadband. Why is he correcting me?
Say no, I'm saying broadband. Listen to the questions, stupid.
Google, you idiot, sorry. It's kind of like a thing. I mean, I barely use Google, right? But like, what the hell, this is still uncharging.
You see, use Google for lots of such?
Interesting I mean, because... stuff that you... course.
that.
Then, now, that is some compass for me.
Was it take the same book?
Yeah, let's see. The Comcast says it's Comcast, especially, ATMC, right? Actually, T-Mobile is actually mine.
.
Oh, CMC is it? Okay.
Shit, they're even up to 100 million. Any customers?
Okay, we probably mentioned podcast. They own more than they own NBC. That's why I mean, they actually also still immediate. Um, maybe...
Asia didn't give you broadcasting as well.
What are you going to do today?
But just for broadband. If you go show more.
It doesn't allow me, it doesn't work on checking.
Like, here's the thing.
Oh, yeah, yeah. 29. Okay, that's the one. Yeah. So I'm guessing a household has two, 3 people, right?
So maybe they're like... this is the one we're talking about. We're trying to talk about it. What is domestic wireless line? So it's like, what is that?
Domestic wireless, right? I don't know, man.
Well, why doesn't mind shit with German? I think, especially in Germany, right? Yeah. Because this is EIA release, right? He not showing me, yeah.
Okay, so one thing we're overlooking is, like you said, Stalin can be global. But they still need to buy specs from in every country because of themselves.
For broadband, you know? So one advantage they have to broadband is they can just beam it down if the country allows them. Like, for example, India bands starting, right? because of country, judging people use it.
This is also trying to protect that one telco, right? Because when you do that, like you, you take business from the existence that they're, I mean, you're not a local company, so you're not being in that position, right?
possibly possible. Actually, uh, the biggest dump in India, uh, one of the biggest mobile nights, uh, at, uh, my friends on COVID started, uh, like pure love, because the, the richest guy's companies took the best out the right one that took that, that invested in matter and stuff. Oh, reliance for that. Yeah, yeah, yeah. But they're like a huge conglomer. Real something, right, yeah. Yeah, pleasure.
His on cooking started at, uh, which, um, and he wasn't rich on anything in the 90s, when they were giving out mobile licenses cheap, uh, he bought them, uh, bought some and, uh, became like super wealthy basically, but, right, right time, right place.
Yeah. So, when you went in different countries, I mean, sometimes in Nigeria. Yeah, I didn't get back to it though, but it became Zen.
Oh yeah, it's very big. in Africa as well, right? Yeah, yeah, yeah.
It's big, but like Venetal is a number of four. So it's like the furniture, right?
Yeah. So his work, like, Mayor, this some guy. And my friend said that as a gift, the guy bought him, the, you know, the hotel, the hotel, Jane, the hotel, and shortage, he bought that, and now he's opening different outside hotels and different parts of... Some of the Indian guy? Yeah. And he also bought him the 10 Eagles estate. you know, their eagles. It's like huge. They have zeros. Yeah, it was Scotland.
It's amazing. should go there.
Was it was that? Is that a place?
So they had the G8 summit there. It's like a huge famous golf course. Ah, it's where? In Scotland. The Eagles.
I've had a name.
It's amazing. Do they have better games too?
So it's a result sort of thing?
Yeah. We'll see. We should have better images.
Yeah, it's like amazing.
So you bought him that as well. So now he runs that, so it's pretty good.
Interesting. Um, But I guess for me, like the one of the shocking things is like, um, This business was like so, like, over, like, so valuable, and I guess, like, it's literally the same thing Tesla needs to manufacture. It didn't come manufacturer. Because that guy is like, as valuable as like a Dr. American car companies all together.
Yeah, but we realize one thing. There's nowhere for, in the long term, for Elon and Tesla to hide, because they have to develop deliver on these promises, all the stock prices will go down. There's no like, it's fine for now, but... As you saw in the reaction to Tesla, the design is the robotaxi. Ultimately, people will price it down. So he has to deliver or pay this more regard. So that's one thing he needs to.
What does he need market value for? Does he need it in a big time?
Well,
No, he has market value because He makes audacious promises.
And the beginning, it came true, right? Like reusable rockets is amazing. Like, Tesla. But launch is like not a very huge business. It's not a huge business. Well, Australia is like... Same with Tesla. They deliver, they make, they even better than the Chinese. They're more efficient in the Chinese, amazing cause, in their margins for making cause. Oh, I didn't even know that. Yeah. Like, this was, like, one of the only profitable comedy manufactures.
This VY did not probably get it.
I think there might be.
Yeah. Yeah.
I mean, I guess it's... But they have a lot of new Christmas, right?
They have a lot of hybrid cars as well. 4.73. So, so not Tesco level. But they sell a normal cost, remember?
What do you mean? How much is this funding?
I don't know now what it is.
It's like one billion.
Oh, that's the 2nd quarter though.
Is this Q4? Q2? Ah, then income. Yeah, I ask you too. Yeah, yeah. So, let me, let me, let me see, before I sit, and then, then, like, a 5 billion. All right, so what was VID's perfect? one.4
Oh no, 4.7, sorry.
And they said a lot more post, though.
They said a lot more cars. I haven't learned the high rates.
Oh, what's it called on the flying?
They're really making like Tesla is a lot of SpaceX. How much is Tesla like making now?
Oh, I'll get on my data thing because I have my own dashboard. So let's see Tesla.
So if we look at revenue, uh, can we do this by you?
I hate it when it does my quarter.
Yeah, it's giving me, like, find the, I think it's giving you animal and China.
Yeah, Mr. Annual and Tanya. Probably subscriber, we're going to love it.
So what does fly? Is that a what is it?
It's like a paid, so this is a data.
Is that what a crunch-based public focus?
So basically you can dissect all the figures of a company, if you're investing, like, for company companies. Yeah, for public companies. Oh, sure, you know, and you can choose your time horizon, look at all sorts of things, you know.
Because I mean, if you if someone is starting to use a chance to beat up strategies, you like AIs to all these things.
Yeah. Oh, their API is all this anyway. Yeah, exactly. I mean, there's an API. These guys are using standard API just presenting it.
Is there like free public theaters for some of these things?
Um, I don't know now there's, there used to be Yahoo Yahoo finance.
Yeah, which is where Apple is spinning those. But I guess they have an enterprise license, I don't know, the stocks. I'd be getting this people.
Is Apple still using Apple?
Yeah, streaming Apple finals. Interesting. Yeah, yeah, yeah.
But is that for the news or is it for the data?
I don't know, but it's the same.
Oh, no, it's actually nice. So using a real-time price.
Okay, but what is your food fan? You have a financial NASTA, or why is that here?
I think that's a good question. I don't know.
Yeah, that's it. Gives you straight to the stock page for Yahoo finance. So they're still getting the theaters from having finance to setting. They are blending 2 of them.
It was very surprising. anyways. But, um, Tesla 2026, so we made 100 million.
So they put almost the same amount of money.
Yeah, so they gross their gross profit is 20 billion. I told you why it's where it's like...
I was in that. I think I was in that. Wow, food, they make the same another...
So can I add BYU? Yeah.
You are too, yeah? Yeah. A BYE company.
Where is the actual one list? Isn't this coming now? What is the main one list? Let me check first.
Where is the hosting? All the tickles and move.
your ID, just uh...
The hurrying company is Hong Kong. So S-E-H-E-K. So that's the Hong Kong company, so that. I don't know if this is H-H-E-K? S-C-H-K. S-C-H-K.
Oh, bone sucks. Financial Apple, uh, uh, but it won't have the earnings.
But yeah, it would be interesting. Like, they were just like, DY, that's like the one. I'd be waiting here because you should have. Yeah, yeah, yeah. You should up then. You're right. True, something true. Yeah, list, list, this one.
Yeah, like nice, this is some congress, is it? Should we try? Yeah, that one. Oh, it's just the news of the financial is always a superstition.
All right, there's something here. So, I mean, see, so you see slightly less ghost profit, right? Yeah, no. Well, actually, let's let's do like for like yesterday. It is annual, right? Okay, fine. So Gustafa is less, 16.3. And these guys are the best of the best, right? One is the best, though.
This is that. But this guy costs way more skill.
So they sell a lot more cars than Tesla. Their modules are lower. Yes. They make less gross profit than Tesla, selling a lot more cars. So Tesla is like really a special company, manufacturing cost.
But it's also because America, because it's also like because like America also protects like the US for them, for them, right? And I guess like the news from there, I mean, he comes from the US. So I wonder, like, how much money they share, they have, if you kind of like compare these 2 companies in the markets where they can fairly compete, because they can't fairly compete in the US, because UIDs is blocked, right? Do you know what I mean?
I don't know. If they're blocked in the US.
Yeah, north of the US, non Chinese UVs a lot in America. So, um, so yes, I mean, I guess if we ask a message to see how much, uh, let's see, uh, what's the...
Come on, catch, yeah. Let's test that, uh, S, E, U, I, D, E, N, Y.
Uh... Okay, that sounds more, on point. And your ideas a lot more, of course. Yeah. But yeah, but this is, um, this is impressive, um, because at least it has that thing, uh...
Yeah, it does go with the Mayfair. The DID? Yeah, they don't do it, yeah. The shit? Yeah, you can go just call them up, schedule a test, right? Well, compared to Tesla, if you drive a Tesla all the time, then you drive DYD. Yeah, the Tesla, right? Yeah. And the difference is what's 4K between that and the model three. So 4K difference. It's worse. I was quite surprised actually. Because, you know, when you look at DYD, it seems like good, decent. Yeah. And you like that, totally different.
So it feels like this feels like thing, right? It feels like you're driving.
Okay, so like software is like taming software.
It was all surprised. And then when I gave it back, I asked the guy, I said, the most Tesla, I want to study the same, but the snow is very news, like, yeah.
Okay, so this is the market share. But I said you should do EVs. I'm surprised. Mostly you can have the biggest EV market share in Europe.
What does that mean? Not sure. Oh, no, that's bad for you, but that is the right one, yeah. That doesn't include hybrids, right? Yeah, that's... No, I thought Tesla was good. Yeah, the only team I do.
Yeah, because this is why UI is, like, yeah, it's like, are you sure it's Volkswagen or not Tesla? Let me make you actually think, shit, I'm not lucky. I'm getting like, oh, my ChatGPT, doesn't you like, do free chats? Yeah. I didn't know that. Because that was free. I wasn't locked in.
I don't even use ChatGPT.
And I use Kodex, but I mean, I don't use ChatGPT at all. I use my I use my agents.
How are you finding code that?
Critics is the bestie.
Betterly close. Because you said the goal.
No, like that has goals now. Like, here's the sweet. This tweet. are going to send. Are you an ex? Yeah, I'm an X. Yeah, same.
like BAS, you... underscore.
Yeah, just... I'm going to send you this. I'm gonna send you this tweet. But I know these tweets, yes, because I was telling a few friends about output access, undefeated, um, Let me just find it, uh...
I know the guy who's, like, um, book decks in London is, like, everywhere. That is what? Like, DEX guy, go, like, and on them. There's always like what's his name? Web of V-A-I-B-H-A-B. Uh, yeah, the, the study, yes.
Okay. Yeah, I mean, I'm not. I'm opening it. I definitely open it as time. Okay, cool, um, what's the market? uh, I read...
get the shots.
Um, we don't need to think indeed, because you're gonna take forever.
Open it as a traveler.
Why are you using my money?
No, it's like half the revenue of anthropic.
No, no, no, no, no, no. They call... I think on topic is like, I have to remove the topic is like 80 billion. Yeah. Yeah, 80 billion now, so that kind of, they're going to be a 100 billion for the end of K3. Yeah. Opening. I said they made the July revenue was the whole of Q2 revenue. Um, it's cute to me. People are saying cute to revenue experience. So it needs to begin in July, which is already, like, 17 of it. I like. So they cut off. So the, so when you look at the charts, Paul already, Paul already started declining, open source and open AI to decide the human market chef and plot.
This is a 2nd a bit longer.
So, BC, given where we are now, how much value would you ascribe to the goal of Staling? Yeah.
Yeah, I mean, I think it would be surprising, right? I mean, like when we did the, based on the numbers that projected. So if they make 100 billion, like they pretty much worth like that trillion.
But from my estimates, for the 100 billion. The whole of SpaceX's revenue outlook changed because Elon was GPUs, the data center, and he's letting it out at huge margins. Yes. So my question to you is, do you think...
Because that's sustaining your business.
GPU, data center. revenue is sticky. Yeah, exactly. You think it's, does it have a moment? Can...
I don't think James can have a node, which is why I think he's doing the traditional one. He's the only one who can function his face. But that's different. Yeah, yeah, yeah. So I think I think I'm bullish on data center. As a matter of fact, I want to start a data center business. I want to actually want to start one now. Yeah, yeah. Um, because you should look at a little church, you see snapshots that, that food's like way beyond, like, didn't need curve. Like, we only, like, there's like 2000 or something blocks and each block is, like, totally only. Have you seen that that thing? Yeah, that shows them. Yeah, yeah, yeah.
I've got this thing. I'm always recon... Yeah, I recommend... Yeah, I send you the news. Yeah. I've been my own, um, I mean, it's still very heavy, but I've been my own, um, recording stuff so I can pull. I can pull all my voice from all my phones, pull fireflies, this central lights. This is like my new time. Amazing. I really summarized me. Yeah, I mean, eventually, I mean, I'm still, it's my product. I've always done improving it, but I will summarize it and then I hype my agents to it all the time. Um, but anyways, um, where do we?
So yeah, how, how, how do we manage to study? It's profitable.
It's a profitable business. I think that... It doesn't look a value, actually, because then you need to look at if they make a 100 billion.
No, no, right now I'm making... Let's say they make 22000000000 this year, revenue. Staling. Because they're making 4.2, 4.3 in a Q2.
Yeah, six. So yeah, my last thing I'm doing is not the writing TV, right? 8 points. Right, the reason why he's saying a 100000000 TRI on the enemy. So... Coming online, right?
Because they have another thing. Yeah, yeah, yeah. And then because the public companies are not... A lot of that is just coming from the cater center deals because like, yeah, eventually they go to space, but right now they're on Earth, right? So, I'm dropping in Google or paying, and these are contracts which can be canceled within 90 days. Yeah. But I don't think you'll be canceled. Yeah, so you know what I'm saying that was their preference? Maybe that's right, maybe not, I don't know.
Because he asked, someone asked him, Miami, like, how much of, because initially, right, he was building the compute for Brock? Yeah. And we started used to it. And now in the 90s, you're saying that the maximum he thinks, who I'm looking to, bro, is 10%. So 90% of that is always going to be, bro. Yes. Absolutely, right. Yeah, yeah. So which is what he's saying. Yeah. But the maximum for someone else for the energy, how much of the computes that you're building here will be grot that you need to portray, trade models. So you set the bulk of it would be inference and compete use and cloud.
Yeah. So what he's doing is he's buying infrastructure. He finds you to use. He's buying very, very, very ruby.
Yeah, he's, uh, he's, uh, he's, uh, he's just his, he's just his guest customer. Justin is like receipt for images. He has this facility now to do it.
Jenson might be financing him as well. Yeah.
Just in fans in everyone.
You do 10, 10 eager ones of GPUs. Um, you need 500 billion. and SpaceX doesn't have that.
I think it's much more expensive now. Used to be interested together for 11 gig a lot. It's close up 1000000000 now.
No, 50 million sounds good. So, yeah, he was just 500 billion.
No, no, no, but it's more expensive. It's not like your c billion. That what I'm saying. It's cost 1000000000 now. Unless maybe because of his ball buy, he gets a discount.
I don't know. I thought the Jensen...
Giving the GV to the acache thing. I was super against that article. Guakesh was an article that computer could go up the next. Because, um...
So most must have a game, right?
I'm sure you will have a deal, right? Because he just announced in the endings as well that I, so this was the... So the guy is making a case. I, I, I don't agree with it, but you don't still set it on the surf setting on the area, so you can see. This is the revenues of the labs going on. What computer is going on because of RAM? Yeah. and computer centers.
I know. We, we, we had to fire on 14 views, uh, this is, it's one of the most expensive things.
Yeah, it's growing up. So which is why I think, let's see if I can find that chat chat of, chat of data and uh, in the US, Spain.
So, you said AI.
I'm sure if this is funny, but I treated about it, but I may not like, look, today, you want to share things, but I found it. I need to put this coming in, wait for something here. .
So each dog is 3.2 million people, like 2,500 dogs, which is if I wanted to get people, um, columnists advanced to me and you, I hit. So this is all the inference that everybody's dying, you know, they're selling everything out. Google is trying to buy. All of us are here. So the rest of the world happens. So everybody's like, this is kind of like charging with his feet, you see what me and you just did. And the rest of the war, this is like 7000000000 people like Navid, yeah.
It's like the early days of the PC.
It was not even like a sore egg.
I mean, even PC, yeah.
So this, I, I mean, I had to be, say, remember, like, pentium machines from the 90s. Like, like when the lunchtable memory was like...
Yeah, we had like the city room.
Yeah, when floppy disc was like a 4.02.7 kilobyde. Yeah. In 2001, when I was still like in, was it primary school now or like a 1st game, 2ndary school? I remember like I used to like thought computer in my computer teacher, right? I grew up in Nigeria. It's like, um, and that was like 2.4. The populace was 2.4.
So this is powerful, but like all great inventions, right? This infrastructure, essentially, right now what Elon scoring is, what are their advantages, hugely talented executor.
For halfware? I don't have any leverage in software for hardware for sure.
They know how to move fast. But this is not even building. It's just like, the, the, the data samples, like colossus. Yeah. Yeah, and he's throwing in and he has the connections to make the deals and people are playing a price where I think he said that we make the money up within a year, the Westpend, that's it. Yeah. So that's how... That's how crazy is. And this is a powerful growth that plugs into this. Yeah. There is like a lost one% of 5%.
Well, last one percent where is Robotaxi, self-driving, but with respect to AGI is a loss, quite a lot. Yes. So the question is how useful is it?
So I think, because your question, you can go back to your question, is this sticky? Man, if you compare this with this crap, It's like so fucking sticky. No, but other people can come in. No, but other people can come in. But remember what color people will come in and then it becomes them and then it becomes, I think, it becomes value per particular one, right? He's like, let's say building a data standard business becomes a commodity. like 500000000 is still like very expensive. Like there's still very few companies that you just like.
Okay, do you say Caplex is a mode? The tennis like an AWS kind of business. It's not like a, yeah, from this, it's payback here. So this is so efficient. So it's efficient now, but it will definitely go down, but that is where space. This is the show. This is generally...
This is terracial. It's space business. So again, let's say we have 12 months. You say they're going to space next year. They going to put the 1st week. What it cost? It's going to take years. But you said next year, I need.
It's not 10 years.
It depends on if you believe, what do you believe?
So because again, if it's basically works, if Starship from launch services, a launch ship to $100 to $200 location, if they don't get to that.
No, no, no, but you're still calculating it only, it's like a can only staling, right? But then they watch business now, doesn't it? Because the math for, so I think it has to make space data centers profitable.
They need to be able to send it to space, $200 or under for paging. Otherwise, it's not profitable.
No, no, but that map was for stalling. That wasn't from the data center. Which one? The Mac. The macro, like 200 like KG. Wasn't that for selling?
No, it's the same for the data center. Ah, Paton. Yeah, potato samples, it's even more important than bad cats.
So, finally, the nail color is much higher than the body for profile. I don't think it's the same.
So the thing about the space data centers is...
The value of cookies for, because if you see, if you do the value of a pawn, You have the data center.
And actually, you know what? I'll play your video.
We should actually do the map on the data stander. We should do the map on, like, daily standard space. What is what is possible today, physics. What is that?
There's this like one good Indian YouTube channel I found. For everything of Space Talk. I think ever. For things too. Thanks, cool, space.
What's wrong with 36 story? Is this guy? Yeah.
Well, it's not. It's not. Base 24. Gosh, it's just built enough.
Little company that started in a warehouse, we also go, um, is now going public for the... There's a lot of idea of that... I
Yeah, I can do too actually.
I do too normally.
They have no point for 80 years and they have $20 of proven points. So super space is basically $700 in one mine with $5000000000 in loses. It's the home operation, and how do you initially economy? That's P, C, X, on, that's that tomorrow. which just record. The world's part is IPL is 75, but not.
It could also make Elon Musk, the world's 1st trillion.
That was amazing. This is what they're going to be like, who says, on one side, by the sea, that is the biggest public computer street. On the other side, some street, there is the greatest promise ever made by an NSBCX is the Amazon of 1997. It doesn't make sense of people now, but one day, it will do this case, you know?
Imagine being able to build any, just by thinking... These guys are so addressed as 44. Okay, I have, like, it was a grocery...
and we found nothing cooking. So then all this is, if you're vegan in a book, it's not mutually, it's not setting, it's a tiny number, we say 200 dollars. Another new channels are busy showing you, okay? they don't tell you. Is there this entire tool to lead all the right scans of all this one tiny, but 200 push. It's besides this number, it's the episode of 1987. Hey, this is SpaceX, doing the biggest common cameras with the audio. So it is absolutely very deep to extent, why does a company that uses $500 a year, deserve a $2000000 addition? What is I think? This is $200 per. And how do you do ligal, the mic, the scanners, they show tiny.
And one of the biggest big classes, which is, I don't need it. Do we want, I want to hand question. You understand, yeah. interesting is not what you like. is about baring games. Because mostly it's, in any stores, in, for the real way, all the world, easy, there's just beautiful. Okay, and that's all... Imagine being able to build an app for your name. You know, sometimes you just... They like what I would do.
Before 2002, when you lost 200 year dollars, I started okay, company, every month after day, because back then, getting a single sign, I don't want it, it cost anywhere between $5000000 and $400 million. Now, most people because you, that rokies cost a lot, because the road is a legal state, all, the few is super strict. This is where, you know, this is where, yeah. These one, right? even though the market launch cost 400 million dollars, the total cost of the produce is just 2% of the cost. It was just 8 million. Look at this, the cost of the market. Can you look it down into 3 months? The 1st stage, it makes ages a few texts, then the Spinofair, which protects the paper, stylize, and cargo, uni.
And then you have to see these falls. So it's possible to be a computer, it's fine. This page contains a small engine, and in your system for navigation control. And if you look at the complaint, now, okay, really for planning, okay, cost 16, 18 million dollars. I think it goes into those stage, 10 to $30000000 goes with the stage. I think for $10 million, because with bags, and it goes, no, I think this PCX, all of the time companies, like the 1st piece of, you know, one of the, it was your, one dashing solution, and then it was just discarded as a piece of cash. And because this, there's when there's a course to make 1st stage, but yeah, but this is it, it's like, basically. For Friday night, doesn't just how much it actually comes back. After living the same speech. You know, that's taking you, that's doing.
The 1st stage, okay, nice, going to the conversation, we'll be ready to fly, okay? And my other, these pillow fries used to be kind of reusual. Space is very good back to you, you know, again. So it will be the same stage. The species is not used, because it's obviously stationed in the opening. This is the reason why. Species. And 85% from $8,500 per KG to $2,700 per each. And this is the reason why, okay, NASA and Betterman, NASA payload, uses BSX, okay, because BSX is not one of three. Today, we buy my chef species and one, because we, they need to are completely wrong. You're like 3 companies fascinating as one company. And this is where the red lights are. Let me skip.
The 1st business is, okay, so people are pretty, like, this, this, before the reversion, when you instantly... Yeah, actually, loses money on the kids. Why? Because I think labors, I mean, state.
That should be a big time she ever played like this.
Imagine being able to build an app with your name on it just by thinking about it. Meet base 44. So stupid. I thought it was the same psychological, new American pocket, built, not by NASA. my million Mus. Sorry. The most powerful rocket is introduction right now. Three, two, one. No, dude, look at that. I can lay math. I think we need to go to Google station. SpaceX, lag it, and reuse it similarly.
Okay, because it was 6 million dollars, they also caught and used for the 2nd stage, which is another $1 million, that might lose a, a season you operate for any, you could try. So even the most reasonable market. Still, there's a 20% for itself, discussion, especially, there's 2 crazy mixture. If this will blow your mind, start it, is better, like, every, every, to come back with 100% use, really. You mean this, it's just as small. Really? This way in Bustle, and then she went out, the Bustle doesn't even land on less anymore. It was back on the sky and the last hour, so they play, we try and play. This is what space is built off in October pretty, pretty good. But until now, the 2nd stage was usually thrown away.
But now it's actually, even this point, it's designed to come back and get caught. So the new stuff should be doing that game last month. It's been for full and rapid usability to refuel and realize. Yeah. And now we're going to, this is where it gets crazy. Okay, see, guys, go down over a half a day. And because nothing is going away. The only real deal in cost is fuel. It's not sure, it doesn't make, it's cost one or 2 million dollars. Sure. $100 million, to just end 1000000 dollars. And the cost when you know that can come down from $2,700 to $200. That is a 93% to replace.
This is real life.
This is not what we're studying. This is a catchy. Staling has more than 8000 points. I guys, we need to get down every content. I do this deal. They order it. I have 60% every time. Now, do you realize, here of US, those celebrity promotion, every street, okay, I don't think people are, it's not like, it's 10,60, 3%. So, hello, this is I did not even exist 5 years ago. He's now by magic, the most profitable elcome business on Earth, and then comes business Google 3, XAI plus X. and this is the surprise. In February, space is swallows SAI, right? And for some reason, why are you prearing the new predictable billion dollars? I mean, $3000000000 is very, like, pulled into limos. It's V6, but SAI.
This has a new $30000000 at $2150000000. So, give me guys, my only eyes, right? I think I guess. I know it is very nice. People are already deleting for the collections of opening, I don't know, SEI is, but 78X, which is more than what it was, or 2, 3 is players in the market. Now, one of these 3 is, you know, The rocky business uses greenly to this. The AI is this one, 6 points per billion, this. And the one satellite, right, business is slightly different, all of it. The question is, why are the most animal computer, earth structure, etc.? Like this, the other like, that is, like, 6 months ago.
If you remember, when you're watching IBL National Muscle, right now, in these days, when you get playing, your phone's in the US, you leave the internet to Oscars because they're in, but this facility has 1000s of servers running 20% and I need to do it. I mean, the request is a low time. And then they say, well, this is good. This is a, then you buy a subscription. You use a, you use like the new quality and place the video into data case. So data cycles are essentially, massive commute houses, and store, process, and deliver all our digital content is really, really discussed. And in 2010. So we're created or evaluated 2 ceramites of it has only 2 trillion gigabytes. This is all magnets you can hear. Like, I don't want to raise something they like.
The word is relatively generated. 2, 2, 3, is a 100 is more... So the world is producing more than a month, then it did in all of history until 2010. and the next 4 years, the worst, you know, it was in the game. Now this makes no process for the people. Number of us is power and number of us is true. If you can only perform, this is actually quite, quite good. I think is... is... I understand. Like, it's still starting to be profitable, but all you have to be discussed, but all you have to do is the satellites, the reviews. That's right.
I think you read welling up being, okay, doing it. That's not laziness. That's a specific activation.
We're supposed to show how energy gives by AI guys, or something we're talking about, are especially causing legacy prices to launch in some Americans. I think they send a thing. So, do you understand this? These days that are things. Do you understand it? Like, why Americans are having so much problem with data science, is it? Is it overdone in America or is it is it like Chinese for propaganda?
When you say problems? What do you mean? Because I mean, the same one?
There's a lot of, there's a lot of pushback. Yes, they descend as well.
Well, even in particular, they're pushback. They cancel one. So the reason this pushback is the environmental lobby, the like, how's it going to affect great electricity, 1st of all, is...
And is this a is this a pre- is this a pussy air problem or did you just happen during the crowd there? Actually, actually.
But the energy needs were a lot less in the cloud era. So it's like if they open a data center next to London, is price bar energy going to go up? Is there going to be pollution, to the water?
But some of this type of skillers are going to do their own energy, right? They're going to do their own plans. They have to.
There's no option. They can't cause the local areas to plug in, right?
So but I think people know that. Why are we still pushing against the designers?
For all these environment movies, the same reason they oppose factories and all that, right? Right.
Because it's like a huge country, right? It needs to like don't waste sound.
Yeah. For, like, the city, you know, you need to break out. So a friend of mine who's big in power, he has North America has for some of the large energy companies. He was looking at an investment engine investment. He's definitely bought. So these guys, when he sent me the deck, They, um, And he understands the business well. What they do is they look for new sites, but data scientists can be built.
They make sure they figure out how to get power to those sites. and they get all the regulations, all the permissions, everything, which they tend to sell on to the hyper scalers. So it might take them one to 3 years to get everything done. But then what might cost them, I don't know, 10000000 or 15 million, hyperscalers will then buy from them for like a lot more interesting. So he was telling me about it.
So then the business of doing the leg work for the for the happy skimmer. So you just buy an already-made product.
kind of like that. They're not in the business of buying GPUs and building the center, but all the stuff that delays. They mean around there?
Yeah, like, interesting, that's good ways. So it's like, so there's already an industry where people kind of like, they're broken it down into different companies and people are already kind of like kind of specialized and fast tracked, but those interesting... Yeah.
This use, as an ACV, as a 100,000 uses, and you don't have to use code. So, US exercise, like, you said, it's in Italian, but, you know, and you've been in a cycle, some days jump. Oh, really? Is this true? is the fastest jump. So this is the question.
No, like a city question. Yeah.
every single day. No, but this is water one, so I had this a meat. Yeah, no idea. Yeah, no, better than me. The dinner centers don't use that much water. I mean, like... The problem is, the I was dedicated to the 90s. So we came up with this because we were done. We cannot do realizing this online and we cannot stop talking. So the question is, what the hell do you, when you learn in Jack, have the same answer. We must put data centers in space. I think this idea, I think, but in one way, as well as this season, like 40% more effectively, that one count.
And this one, this one, you take spaces, Ios, and it's cool, it must be really easy. Right. It is true that this is cool. But the problem is, space goes so vacuum and vacuum mass is a problem. And as you bring it on these, like a sizebook, you are writing, you're gay and go to, automatically, there are no marines. A vacuum so the fat is, you know, because there are no molecules to master heat. So what do you heat that it doesn't generate?
The only human doubt in space is to use a really, you know, the challenge. These materials need to be so big. the entire space station produces just 70 kilos of heat, but it's really resisted with 7000 kilos. So if you get one, make a word. You really doesn't weigh 100 tons, suicide, at any time, for you. So, currently, the total cost of the government is very continually, but it's space. It is $3000000000 in value9, right? Because now the loss cost is $2,70, but if you put it in, it also lost $200 per KG, the total loss of consuming 20s and litres.
It's more like that, due to what I now say, will be because it's Bible's based as somebody. So this entire billion dollars. No, no, no, this. Currently, no, no, no, no, no, no, no, no, no, no. So unless you make the radiator on space, Okay, Kunas makes sense.
Okay. So basically for every one kilo, 10 kilos of radiator.
Yeah, yeah, 10 X. Yeah. Okay. 300 more, $10000 in 59, which is it? Mike, is now?
Yeah, that's 2700 a KG, right? Yeah. So that's that's the main thing, problem. So 300000000 per data center, it would not work. So it only works if it's 200 or on the, but it's, it's $302000000, $899.
So, my use now, the last cost, $70, but it's cheap. But if you bring up the cost of cost, you, the total cost of consuming $27 million, it's more like I, dude, what are they going to say? It'll be, it's in those microspace, has a better work. So this entire $1000000000 you should, I'm not just one of them, when can you just cost for $2,0000 which is $200 per each? Or, can mandize somehow, come with an insane idea, to reduce a 100 times radio, because something cheaper, not. So there should be like, there should be probably like a bunch of startups now, doing like hardcore research around like radiator.
Oh, if you're looking for ideas, I have some. Yeah. First, if you could come up with brand, an alternative to gram that can be made differently. That's a short 1000000000 dollars. Alternative to dissipating heat in space, that radiators, huge problem. These are like problems people should work on actually. These are like very elemental.
big, big on lock problems. And this is what you used to be taking a little bit. Like, we have to build a vehicle, and not only does the mission, but does it any expensive, like, 600s of million dollars, even just get started with something interesting. What I'm gonna do is reduce that price of admission. With reusable space vehicles, we will be able to beat the cost of terrestical cases.
This guy understands all the problems, but he can't should be communicated.
It's a business guy, it's not as technical a helo.
Yeah, that's so interesting. like a couple decades.
So do you realize something? Ready to leave the future of AI?
So one thing is this guy skims over. Join the doctorate?
Because, because, because, because, I guess there's plugins, I also know that technical, right? So you can't do it as people say, yeah.
You can do it, okay?
This is what they are, like. because I'm with my Tesla species, like, excellent, like, quickly too, where is dream? See, that's generally, you know, made all these announcements that are generally huge. If you build their fab and you compete with Nvidia, an ASML, sure, you'll be huge. But when you, that's so...
But when? Because he just announced in the thing that he's not doing...
How easy is it? Because the things he's not good at doing, like XCI is not as good as Anthropic and opening.
They're not good in software. I'm good in software. I say this are nice all the time. I mean, Elon is great. You know, it's good at hand, we really had hardware. But like with model, his story is not really that good.
Have you read his biography?
Yeah, but I read the one that came across there.
So you know how work Jackson says he's like a visual thinker, um, intuitive visual thinker, like, uh, the engineers will come and say, okay, we need to use this goals or whatever, and he's like, no, actually we could use this cheaper one, it will work as well. Try it and let me know. Yeah. So that is like gold. When you're building the rocket, all the aerospace spots are so expensive. Remember, you were saying, but Elon was like, no, we actually have this part in this industry, which is a fraction of the cost. We can use that.
So that is where it is a lot of his talent seems to be, right?
You sort of not to match like things that they can do from like other things. Yeah. Like 1st principles, like the...
But in the data center business, show for space, he could, but it's the stuff like the reusability bringing down the stages.
No, the hot, that was the... That is real idea. So I think I think we did center, right? They're using what he might be.
He has connection, but he's not building the gym.
He doesn't know, but I think he goes to space. That's what his move is now, right? Because now the truth is, because he's the only one who's got like half a long vehicles, I mean, you, it completes around like the heat, the heat, the dissipation, and all, how does this flute?
And I think police launching for other people as well.
The data center.
37 in space?
Yeah. Like the British guy from Gillingham, what's his name? The Stackfall, guys? Start loud. Yeah, yeah, yeah, yeah.
Yeah, because I'm thinking, like, would anybody else have enough expertise to beauty cluster in space? Or would you not just beauty? And then...
Look, the fact is, Eno's already build scholarship, and once people build things, other people build it as well.
That's true. So, China already, you know, like China just, there was some sort of new musicability like this, yeah.
See, you know how I don't know who I was discussing with, but you know, when the Reich brothers did their 1st guest for the airplane, right? with my plane, there were no there were no TV cameras there, there were no, there was no video. I don't even know if photography was our widespread. But within years, Europe had a version of what we played, and the military had one as well.
So how did they do that without any pictures, videos, tons of articles, or people discussing on X? And it's because a lot of ideas were around that nobody knew if they worked or not, but then when someone does... They're like, okay, this works. Yeah. And then this hears it, people go, you know, I was dead, Kitty Hawk, and this is what I saw. And that's how it is.
I agree. So your biggest, like your biggest kind of pushback is what's the, what's the retention on that like this and a remedies, right? Is this like...
So if you go back to the job...
My questions, this generally game changing, but so many people will come in, but it's gonna be tough. It's gonna be tough, like, building a profitable EV company and stuff.
Starling, I have all those open questions, right?
Like on the mobile, on the, on like how much you can launch its name?
Yeah, where I see stalling is very valuable, is as a, firstly, as a politically integrated thing, part of this type. So for robots, data centers, you need to not unique.
You need like you need like high bandwidth.
Yeah, well, all the stuff that this thing in space is doing, when you started to calm down, that's cool. Also, starting is genuinely, genuinely a game changer in dead zones and war natural disaster situations. It's like completely full proof, right? And that is huge.
But I mean, that's the like in this market, right? Like, um, yeah, but I guess like, yeah, the biggest use case for style, like, is far in, like, travel S cars and robots.
So my rough rule of thumb is, I'd say, Staling put on 15% of the premium market for broadband and mobile.
Yeah. Because like 5 hour proies. For like for an electric for like a sub-driving car robots, you can't use fiber, right? Because 5R requires you to plug in that one. And fiber right now is the biggest, like, when it stops you feel at home, Yeah, it's 5 out. Yeah. So now you should take 5 around. Fiber can buy your robots because you can have like you can have like maybe some broadcaster in a house, you can use the same type of your robots in the house. But if the robot is moving around like from the house to...
It'll be Wi-Fi, right? The fiber gives you Wi-Fi. That what I'm saying.
the robot sees in the house. You can't work with a car. And they can't work if the robot is like working outside the house, you've got a robot, who's out to buy something from the shops for me. So that one would.
Um, I'm glad I know fish your mobile network. So maybe, you know, like that shot, you showed me a insurance and compute. It'll be similar for mobile. The number of connections needed will increase dramatically. So that is a different angle.
Yeah. And so that is the stepscape, it keeps getting bigger. Yeah, but I don't see a word where if this keeps getting bigger. that this doesn't get bigger. So your argument is...
If this gets bigger, then this should get bigger. That's what I mean.
Yeah. So this obviously will get bigger. I don't think we're going to slow down. I mean, what is in the Chinese company? You seen the Chinese robots, the one from UB tech, you seen the robots that look like humans with like real skin and like real... No, I don't think people will like, oh, it's too long. Like, the regulation in China. already mainstreaming in China. Yeah, maybe the West will have all this like freedom and issue.
You saw those robots on the street now, you probably give a smack or something. It looks so annoying. Like, I'm used to a lot of normal people.
Yeah, of course. I think in the worst people have a problem with it, but not in China. Like in China, like UBT is already like ministry. The robot dogs are in there, like lots of people are already buying those things.
I think, you know, watch like self-driving would take even longer to become useful. Self-driving or? Like actual cabs? Self-driving is not even here yet. It's been 15 years. But what we got out of it so far is a very useful due for human drivers. Because if you only have to drive 10% of the time and just keep an eye on the road, that's much better. Yeah. Robots. And I've been speaking to some guys, researchers who feel like their own role in world models and stuff.
I read this long, long article where essentially the models we have now in the world, you'll have to teach the robot for every use case to some extent. So if you have a robot to clean your house, completely do every single thing. You... I feel you need to have like 100s of 1000000s of robots in houses collecting real data on how to do this stuff. Which is what people are doing now.
I'm listening, I'm trying to find...
You are generating synthetic data now.
I'm trying to find, this is the, this is the growth rate. I mean, I'm going to say this is the current data. Um, This is the most recent, like, in the industry sense. So this is what I was trying to, because another conversation with a friend who's in, who's your group, and it's kind of disagreement. You can see like the bulk of self-driving, it's kind of the growth, like, and it's like, it's like, it's 100% here and here.
But what does that number mean?
This is number of sound grabbing, the number of cities. But it is level five. And that's a good question. I don't know.
As in is it without a driver?
No, this isn't enough. I think it's just full because it's still got a few cities. So you can see Wemo, doesn't have Tesla, so it's really validating, this is.
Oh, so it's police software. So it's 15, that's like...
So it's 50 cities, but you see out of this 50 cities. Half of them are in China. Yeah. Half of the cities are in China, right? So we know he's got Atlanta, so but that was half of them in China. China is ahead, right? much, much, much, much, much ahead.
I don't know, you know, like every time I've tried a Chinese product, I'll be disappointed.
Yeah, they're not. The experience is not as high quality, but it works. It is functional, but it doesn't have like that, like, experience that, like, name one good Chinese product that is a Chinese around you we've used. That I use, yeah, I style. I mean, some songs are pretty, but Samsung is Korean.
It's great. Samsung is great.
Yeah. I'm talking Korean. Yeah.
I can think of Lenovo, but I don't know if that's great.
What do you mean, the number is China? Yeah, it is, right? What do you mean, no, of course not. Are you sure? Oh, we live in China? No, it can't be Chinese. What are you talking about?
I don't even know where they're from, but it can't be China shortly, you know. They've been around too long, European or something.
Yes, no, no, Lemo is a Chinese multinational technology company, family in Beijing in 1984.
The number is Chinese. Yeah, yeah, yeah.
But if I'm not going to miss you for a while.
Yeah. But besides that, I can't think of anything.
I didn't know that.
I mean, VYD show, I don't know if that's a major brain.
I mean, you said you proved DY and you like it. I just ask it. Wow, you don't know the number was Chinese. Yeah. But it's still a big market fan, like, American, like, Europe and stuff.
Yeah, yeah. But I don't think it's great or anything. So that's the thing. Show me who you live here, man. Huawei. Yeah, we don't use any of them. DJI? Okay, yeah, sure. It's bigger.
Yeah, I was actually gonna, like, you have a, you have a, you have a plan of here between DGI on this thing, because I think I should, I probably should own a drone, and I was like, I think I have it open here. I think it's going to be on my... Yeah? Fight terms. Yeah, but we don't use bike.
Thick close, okay. Halipa, go.
But yeah, but that's software, but yeah, I don't know. I don't really talk.
Hyson is a shitty TV company.
This is the, my my 15 Charm is a high sense TV. I didn't even know it was Chinese. Oh, sorry.
No, no, no, no, no.
I got it because it's cheap, bro.
Yeah. It's good value, for sure.
Yeah, but because it's cheap. I'm trying to. Yeah, man, I'm trying to decide between this. And this. Have you had this one? Do you have a drone?
Have you done that? I thought of getting one once, but I forgot which one I was gonna get.
Yeah, because I saw this one.
It was a super nice one. I think by EGI for holidays.
I saw this one. This one is like super cool. And like, it's almost like it's like by itself. Because if that's the reason I saw this, as you can fly by itself, you can put it on the smack for the...
Oh, yeah, the software that follows.
Yeah, but then it's like 500 bucks. and then I started asking my agent and my agent is like, okay, cool, just check this one out and this one up.
Oh, yeah, I think it was one of the mini 5s that I was looking at a long time ago. Yeah. Because there was someone who had one and it looked really good.
It's like 7 red box, if you buy the whole thing, and then it's, yeah, it's like, it's like super fresh. But anyways, um...
So Eno's gone from these great things to like, is it a commodity business? I know the people who can you?
No, but I also think, right, like, so 1st of all, okay. So I also think, so, 1st of all, those, if you can't, when I was in New York, it doesn't look like this is total bullshit.
No, the space one is genuinely, it has, as you said, it has high, it's a tough one.
Yeah. But if it works. So the issues, the risks here are the radiator and then the star, star, the star. If Starship can bring it down to 200 and then if you can find it, regain it as a journal. those are the 22 things I need to work for this business, for the data center business to work.
What are you saying, 47 million? But how many GPUs are on that? Like, what does it mean in terms of compute? That's what I need to understand. Because he's saying the 27000000 you save on water and all the stuff and there's less loss, you know, from...
And they haven't factored in what, like, we have the GPUs. like what do you mean?
But I don't know, you don't know say the GPU should be pretty stable because on earth, they will test it for Australia pretty vigorously before sending it into space.
Yeah. I'm like this very into movement. So it should be almost as stable as just sitting in a proper thing. Yeah. But I wonder like how much, how much, what is the replacement rate of like GPUs in an actual data center?
Oh, it's, uh, apparently most of them fail, either they fail early or they're pretty reliable. Oh, really? That's what I heard, but that's... But what is Star clock doing? Are they exactly the same as...
Yeah, what is it?
TodayYC? That's great.
I thought you thought I came in...
No, I was watching... The stock found was the, the branded name, they had a different name.
Oh, was he in yoga?
I think it was in a batch aftermind, actually. I think it was in the 23 or 24. This pretty recent batch.
Is he the guy who originally came up with the Davis End and Space idea?
I mean, I think if you, like, when you're actually invested in one data sentence, there is an space company. Yeah. I'm interested in this one. So for my EA stuff that I'm doing today, I'm doing it through the laps version of this, and I invested in this business. Um, This one's socially good. I mean, but I think he's so bad. I don't know if we understand what the guys do, but this guy's also a biggest same thing.
Clearly bonded you, DSM spaces, right, right? Is the what? Regulations, besides...
Ah, yeah, yeah. You're not fighting all the crazy people. Stock club is not currently a commissioner club provider, selling GPUs, each having you for custom space infrastructure, company buildings, such as that something you've used. What does this mean? Selling up it, competing. You know, this is like, yeah, I think it's the same thing as...
It's also like, is it similar to SpaceX Data Center?
How's it different or similar to SpaceX? SpaceX, um, DSNA business?
Man, voice mood is good. It's like word class. Have you used it? Do you use it?
No, but I use it when I grow up and that's the best.com.
It's, uh, the one on QuitX is better than the one on. But I haven't won a rug is pretty good, but I'm...
It's the only one use, but like, Florida's not usable.
Yeah. So, but so the thing is, I use it to control my, like, I use it to, like, that's how I manage my, my game. You talk to it, then I use, um, then I had a Vita, Vita, no computer app, so I was talking to it. The computer is in the background, picking every single task company on the checklist. And then every kind of like 5 minutes it tells me what's going on. So I don't have to ask you, hey, heads up. And grow, can you, can you use it to manage threads or you use it in one conversation?
So then, So when I'm in the Tesla, the use gives is different, like general chit-cha, what's happening. Now I can also like control the climate and the open podcasts and stuff. and just normal stuff. Yeah.
So, so, the QuX one, the reason the QuX one is exceptional is because even if you open one voice mode, it has access to everything. So you basically say, yeah, you create a nuclear on this pain, stop updating this website, I go to fix claw, do this, go to Facebook, do this. Because you basically have like 20 things happening at most different projects and you can orchestrate it from one. I'll send you the tweet I made. So you can you give your your agent that tweet and just tell it to set up that archestrator for you. Anyways, okay, cool.
The cool idea is almost identical for computing audit a bit, Thai.
So you don't probably go with the idea.
Yeah, so it's basically the same thing.
So they need... They need love to go to space.
Do you want to go to space? But then doesn't that mean space let's go have a cost advantage?
They won't have a cost advantage. I know UCSP sex is much better. So I wonder what this kind of businesses would look like. I guess they'll have to eat no feet, radically different, because SpaceX is very big. Um, they have economies of skew. So there are very things you can do better. So why would it start up like this, complete with like a $100000000 company?
But he's very complimentary of space sex. But the direct competitors, if it comes to, if push comes to shock, why would SpaceX give them launch space? If they launch thousands.
Yeah. PRP did is send a company, start one lunch in November with an NVJ 100, and run trains for model. So it's already happened. They're already in space already. I don't know how many, was it just one in VDM?
I mean, anyone can send GPUs. We could do one. I could send my GPs. I'd have to play the lodge.
You take the massive loss just to say like a bunch of DPs in space. Yeah. But I wonder how it's...
I don't think you take a loss. Like, the moment you say that, you say you're, okay, the same space business, someone will come here, right?
Fair enough. Especially long capacity, percent different space X, when she's step 30 grand stations, the network, such a stalling, say, CC internet. always important. So this guy's out this business. They have a launch business. They've already kind of like beat out some of the infrastructure that their business requires. They're beating this from scratch. Pastric government can supply own operations and sell externity. That's small, little move.
So there are people who don't like it on that much. Like the EU governments and stuff, if they have a chance to know a book with...
Okay, that was useful actually. Yeah, it starts wrong. So, actually, you saw skate. So how can you compete with SpaceX? SpaceX is so good again, figure, especially on.
How can they come previously? SpaceX, SpaceX is so bad, can you agree that with the costs and obviously?
It's actually the, I mean, strong, long-term strategy, currently, because you only launch, but, I mean, it's basically on Saturday.
You know, probably, like, these guys are going close. Yeah, it's a few things forever.
I guess maybe the niche, right? They've probably become like a niche provider. People in this issue that Yidon is in, in, when he has strong relationships and, and then that's it. And then you become like, it'll become like a, like a $1000000 business like in 10 or 20 years. Maybe a small business study will be viable. Right? Um, nature, multi cloud, superior, species, need to processing, specialize, but I can't really see anything here. It's like,
You're right. Stuff like kind of B, SpaceX, and Commodity customer. If companies sell identical GPUs, 1st custom vantage, uncomfortable trees. Yeah.
Special life service is what I'm saying, right? Specialized surveys, analyze it, self-services on top of that shit like Elon man of interest. Software yeah.
Anyways.
So yeah, so I mean like, potentially, it's a big option. This is comfortable.
What is the biggest, what's the biggest risk, right? I mean, this is your work. I mean, you agree with everything here else here? The biggest risk is here, right? This is where you think of biggest risk is, if this is...
So I have to have a question on this, it's like, where's the mode? If Elon's just buying GPUs. And I'm not saying there's no mode. The mode could be.
Just less compared to like what he's looking at.
Yeah, is it as simple as Elon has got bags, Elon has connections. And so it's like an AWS style business where, you know, Elon is basically JF Bezos in this case.
Well, this is basically the problem people also have like a video, right? Because it's basically what NVDA's business is, like, especially not what the site customers.
Yeah, that's true. Yeah. I like to collage. It is precarious. like Warren Buffett said it. And you see what about the cerebony, yeah? He said, yeah. Essentially said...
I know I heard that he said something.
You're making real bets here. And the game could change and instead pasta shown us, the innovation changes the game.
Two one buffet on AI or data centers.
Is that a recent one?
It's that one, but you want the right clip.
Spider-Man. I watched the whole thing, man. I watched them now. Yeah, watch out. Oh, you want us some food is that, I think, instead of wearing...
In the last few years, I guess, dude.
This movie? It's 11 minutes, but it's still shutter then.
But essentially he's saying, like, look, there's real capital deployed on a bet here, and he's saying it could be innovation, that changes the game.
Talk a little bit about in fakes because the deep fake warrant that popped up early in the session was pretty good. Um, they had somebody standing up, you know, Greg was joking about it, but, you know, the first, the first question went to a guy from Warren, up in the...
of a, uh... a really good imitator. Um, I mean, just imagine... Well, yet, we had that famous thing before... playback in New Jersey.
The other thing is, I want to welcome everybody back in the arena, make sure you're getting time to get back to your seats, and then we didn't give you a lot of time for a break. Okay, remember, I'm fine to do that.
But, um, there's a lot of hype in the world right now, right? It's like the internet, right? The internet was real, huge companies came up. So people say a lot of bullshit, right?
So, but the, like, so based on this, you still don't think Castro is, like, 3 sex is like worth, like, I had least one trillion. Is it worth one treat them or no one. five. So it's it's like...
So my question is, this is generally leverage, even though it only makes like a billionaire. There's a lunch lunch. This is like a huge competitive advantage.
And strong notes, no one else can do it. Yeah And you're the person that comes to this blue regime, but then I've encouraged from China.
Yeah. But it's not 100% water type thing.
If it's only worth like, what, 1000000000 right now, like, okay, cool. 20 million.
Well, the launch business. Yeah, what's it worth? I've seen launch BSP what? A couple of 100 billion at least.
But how do you value that? Like, like, what?
Teacher, because it has a mold. There will be people, there will be demand for notches in future. And if he can launch like 15 starships a day.
For launches for for whatever people want to put in space. and if there's a moon base as well and all that.
So, um, Because the only practical use, the only thing I have, because, I mean, the only use case right now, the only parts for use case for, for, for space is even like 80 centers, satellites. That's it. Yeah. Like no one has any reason. I mean, yeah, you want to go mind something in the moon, but you don't know...
it's coming back up. I mean, it's somehow. It's super tempting recovery. Yeah, it's coming back.
But, um, every time we speak down and it's coming back up.
So starting to have those questions. So as now it's very profitable, it makes money. It adds free casual to the business. The launch business adds free cash flow. This business, once Anthropic and Google kick in, highly profitable for what they paid for the infro, right? Yeah.
I don't think if therapy keeps growing. They'll never need less of pizza. They would never have any need to bring this off.
But if they become big enough, they'll do that. They'll be like, why are we being effectively paying for the GPU?
But you remember why the topic is paying that much, right? Why? Because you have 90% margins. Anthropic is so profitable and they're kind of like operating. All that would come down. Yes. I agree with you. agree with you. I agree with you that they're gonna come now. I was watching and I was watching this. I'll send it to you on my ex. I was watching the chief product of Felo Francisco, and his point of view was that, if you have too much margins, that 90% is too much for him for photokens. You have that. Eventually people start putting that. You made too much. You need to make much less.
Yeah, like Gail said, your margin is my opportunity, right? So all that will come down. The market will correct it. So, especially because it's a commodity.
Yes. All those have been coming up. You I've been the models that are commodity now. You were, you think that?
I didn't invest in our topic because I thought the models are cool. which is stupid, but that's what I thought. I was nice when I was wrong.
Yeah, yeah, I mean, it's kind of difficult, like, come out to get it out. So there'll still be a big possessive if models become a commodity. What's this scenario? They make money on these friends.
But I have an idea, actually, the other day, right now. Speaking of, um, so innovation and stuff. Um, what if uh, I did something where it's basically some confidence to make the efficiency of their dopos, right? Like, what am I getting for my tokens? Um, and I can think of at least something encoding where uh, I haven't thought about this deeply or anything, um, if you burnt opens to generate code, then the development environment will track
honey tokens, what will, the honeymoon is called, the sun mark code. Then it will also keep track of... The time you know, the time you spend modifying that code, and when you ultimately deploy it to production, and it will calculate how productive people's token are used by tracking, what you're doing with the phone you generated. You get what I mean?
I think someone like me, we need something like that. I'll show you, um, should you do things? I know what you mean. I agree with it. I generally.
But I don't figure out how to measure it because if you're using like cluster or something or clot code, they own the interface, right? So the thing is how do you start time stamping, the token generation and measuring it?
You can't. There are enough stuff. So this is mine, so I have I have a few.
So you'll have to build a new ID.
No, so you just have to, you just have to, so be a few ways to do it. someone sent me this today, so there's a few ways to do it. Uh, someone's gonna be one this morning, hold on, let me just quickly check the name of the guy that sent it to me.
These, uh... So these guys sent me this thing today and the, the kind of like the idea was that, Um...
See, if you had something like this, if this guy sent me this avenue, who did? Um, they sent me this, right? I haven't appended, let me see, but this is a physical looking source project, that their own idea was this one, look at your, your code code, sessions, and then tell you, like, where you were being wasteful. So I guess basically what they're doing, they take your session, and then they'll pass it to another them as a judge. send this to you. But for your question, in terms of like, how do you use it with mine?
And then those judge are very, very difficult.
Yeah. Because you, you, you, you, you, you, you, basically what I do, I have a bunch of, um, I have a bunch of 2s that I use. So you can see if I, so this, what I'm drinking at is all my, I'm trying to see this, let me see what the, I don't use it all the time, but I should be able to pretty much tell you in terms of token and cost. But how you would basically do what you want to do is your focus questions, obviously, live in whatever device you install it.
So you can scroll the sessions, and from the session files, you can estimate how much token you're spending. Right?
Where do you put the section files be used to?
Special files are in your computer. So for instance, I just, so what I'm doing here is...
Also, I just, uh, I can set up fully, set of clothes or something.
So your code is obviously... I need to see the code as well. So that's the thing.
And I want to measure the time, like, if you generated the score today, but you spend 5 days modifying it, because the above, so it's just...
Yeah, but you got to eat it, you got to eat it bit by bit. But I think it's possible. So Uber also launched a product called ADR. So I also sent you a bunch of this because you have a few new idea. So ADR was Uber. The vehicle is doing it is they were feeding all your session data from all your agents, right? And then they will also do some sort of reverse it. So they will reverse the form to use and they create sort of end to end of everything. So you go from your sessions, the prompt you use, and you get a good full picture of everything in the organization. So every single person who is using Asian Uber.
They use this streamwork to get that data. So they basically have it on everybody. Like, what are you using an Asian for? And so I guess with this, they can basically pipe it into some sort of observability to cost. And then a bit of what you're also saying. Did I send you the Pedro's, did I send you Pedro's YC thing? You should watch it if you haven't watched it. I think it's the best podcast I've ever seen on AIM business. Yeah, did you watch it? No. This one.
I don't watch a lot of YC, but then, no, I mean, very early days to watch...
No, Pedro is good. This is good. I'll send you this one. This one is really good. But the reason I also mentioned is, because here and Ram, they're actually building their business for what you're saying.
Around for the... Yeah, those are guys.
Yes. So basically their whole business now, but who runs pitch now and you should watch this one?
No, I know what rap was actually done in Germany, the day I do was pretty big last one, but the scenario. They've come to Europe as well, by the way. You are in Europe? Yeah, you can get your company owners.
This one, I watched this one. So if you watch... So now, let's see.
the size is it, but yeah, so this one. Um, basically, he's basically your business idea. This idea I just mentioned is basically the work. So Rampus basically positioned themselves as like, in the future, but what is going to be mostly like very 50% tokens are 50% like humans and whatever. And so they want to be able to tell you what to put your spending, if it's productive, what it's for. And so they already have my position themselves for that. So that probably your...
Well, the question really is how it's an obvious problem. Yes. Really, like, how do you measure it and how do you break down the problem?
They already started, man. They already started. They launched a product last week. Um, to go to product last week for something like this.
Because they see where you're spending, okay?
Yes. They know it's a big problem.
How do they know how many tokens are generating?
They launch a router? One? So, no, this is this is different, but they launch a router first.
What is there after they launched, what did is this? When did they launchched route Yes. Right? Yeah, so like, yeah, yeah. So it goes around. And so they already kind of like try to put it together. But I'll send you like... It was announced when I played it out.
But in this podcast, you would understand why you launched it because again, a lot of these customers, The devil is purely irrespective of what he says in the detail of how they imagine. that's the one.
Well, the reason I'm mentioning those things, you just say that it's actually doable. People are doing it today. like, you know? So they are, because you are, you're saying if I can find a way of, because what you're saying is some sort of forensic accountant, right? And I know a friend of mine is also wise becoming. Actually, imagine here, in dimensions offices, in San Francisco now.
I'm firstly, I'm narrowing the case to just code, because I think that's the best use case in the most wide use. And I'm like, higher code tokens productive. And I'm saying that to know if your code of tokens are productive, you need to understand the point where the tokens generate it. plus the human effort in modifying those tokens. And uh, Whether those covid ultimately make it into production.
So if you generate tons of tokens, it never meant production, that's wasted cost, right? If you generate tokens where the human has to spend lots of time to correct it, that could still be a loss, even if you win production.
I feel like I feel like technically you can do those sites too quickly through. Yeah. Session falls, if you look at the session fall, you can see like issues, it's emerged, it's it in production, and I guess like some, so I think if you do the, if you bring it into two, which is just, let's do that, your social problem, and be able to tell how you cook, you cook it into the production, and then you can do the stockdown values. What is your business? What is your business value?
Because if here you are, the business of selling this software, but you charge a $1000000 and attach 23. So even if though we generated the same total, you see, like the value is different.
Yeah, yeah, yeah, right? But then people would like pay, like I told you, one of my engineers paid the AI, they're willing to pay elementary for measuring the value of your agent, how to charge the customer. So that's solving that by building a telemetry platform.
The value of your, what do you mean, the major? What do you mean?
Yeah, so if you go to pay your AI, they rather than charge for tokens. Like, so you could give a phone, you're saying, my agent, you charge the customer, the customer uses a 1000000 tokens, probably a 1000000 tokens plus 20%, but they're not doing that. They're saying you charge going around. So if your agent has created $2000000 worth of value to your customer, you try to present it with that, something like that.
Which is what copies, which is the point copies, which is what copies is always saying with lanthropic. Because the issue with up in the anthropic is the adjust of tokens, where you can tokens can be useless. You can, like, an engineering, like, finish running your cooking and not produce any valuable code. So, I mean, everybody's like making a case for it.
Actually, it's interesting if they put down some pricing, you know.
I used to have pricing in the past.
No, they were doing it very custom initially. The back by supply on my speed. So, yeah, but should we finish the video? Yeah
He hits on something, right?
I think he's like a few legals, even maybe six. And who is learning? Because he's like sports provided because it was, basically, the iconic story must be six. So, as it justifies a total legal relation, where it would be injustice, if I would tell you, that's it. So you have 14s, when you was even... Number one, the SY grades are duplicated as a market. We need to be, there's nothing to the IOA and I, you know me, do the month of it, right? yeah. But it's only very funny, you know, you guys go to go to stay in. Study, no way. Nobody has ever watched it, and they can repeat about, like, oh, I'm asking. And especially nobody does for eating, is a tummy. And today's real life of 2007, it was the aging. The entire is this morning. These were days also.
Thirdly, if you get the work, AI was a free word, also every development, and it also will be opposed to 3000000 was your discussion. And suddenly, 75% money carries between everything. So it only was end of this market. So they're making a loss of launch?
I was unique, in the name, was your introception. I wish, uh, I mean, 3 penny and his past share, and 300000000 for... So that's IRD. So, what do they say? What are the margins? I imagine they're increasing. So it's getting better.
Operating volume was negative, negative, 740, 730. Yeah, they become more than 63% now.
Sound better or worse?
Well, it's better than previous. margins are getting better.
No, those things are negative. These things are oppositive.
No, starting is a great business.
Yeah, yeah, yeah. These are our positions. This is my... Oh, yeah, yeah, yeah. So we've been getting better.
Yeah, but it could also be negative because the cost of the satellites and launches came in or something like that. Maybe now they've, if those costs no longer features.
So it's getting, these numbers are getting better. And the higher, the other better. So they've been increasingly like, very, very here, yeah. Yeah.
So, computers in this market end, right? That was nice. I don't get the point, but it's the big stream. And lastly, in spaces, like you, 20% is going to video.
That was a good point by the way. If you look at AFC, the company competing with them, other people are going to come in, but then they don't have rockets. So that's why the Rocket Christmas is so...
He was 6 guys, more than one. I think about this. be short that you really like. But the video is, I handed, I mean, really, a lot of space, because, you ever, I guess, the one thing I did from the postal, I guess, you know, you know, genuinely wasn't going to share, like, oh, space, I just wanted to share any shares, like, the highest possible price. No, when it's a genius move all about, if I say diving on 3, it's like, also, you drunk your food. Number one, it starts, you know, also, you have, every successful use discussion, right? It was push this $2,700 goes, it goes, it does it, we have. No, no, Stan is marching.
Like I said, I don't know, speaker, China's constitutions might die allergic down, and I might fit that machine. And no, he's there, and 19 million. If you actually go out of SSK, he loves these by computer. But anyway. What kind of fare of this parrot card?
What's the $9,119 billion?
What is that?
I mean, I don't know what that is. It's not a new one.
Yeah, it's a Chiwan manufacturer that cheers. Uh, but when it's 119, then you know. No, that's fine. I think SpaceX is following today because they have caprices of 65 period next year. For what? For building out data centers and all the, they have to pay for the GPUs, right? Right.
Yeah, but then the GPUs, you already see that they have a GPU period of one year. Can you, can you check what the tera for? Check what's announced for terra effort on your code.
That's true. You're right. One year, that's a good point. Yeah, so they get a... Check what he's saying.
Check the pair of fountains. Yeah, just checking on the phone. Like what's the...
Let's see.
Is there a...
.
Anyway, for immediate, Elon, you probably cut lots of deals.
Huge margins for info on the ground.
Yeah. But I think it's, but also like, this is if the, if the close source market doesn't crash.
I think low source is quite quite a big improvement.
Right? Thats if it doesn't. Open source does not crash if China doesn't crash the cold source. Because, I mean, for instance, if OpenA and Anthropics, um, IPU or whatever feels, I mean, it doesn't mean I will fail, um, but it's just that they won't make as much money. Investments in AI will throw down. I mean, it would still happen within the next 10 years, but it would like radically screw down. Um, there'll be off-ramp in terms of, you know, people needing to pay all these margins, because even anthropic, because none of them are profitable. Andropic ahead is profitable. I don't know if it's holding.
Anthropic is weird because their numbers are not really, the numbers they report still has a lot of smoke and mirrors because they report topline numbers. They report like their revenue even plus the cloud. So they're just reporting topline revenues, right? So each time, each time you compare anthropic and opening, and you're familiar with this, right? Opening a reports what they actually make. dropping reports, including the cloud providers, they create.
What does that mean?
I believe it means something like when they sell on Azure, right? Anthropic has a margin, right? They don't own everything because, like, it's it's on a job. Like the hardware is on Najo. They're just putting their model on hair. They call the whole amount. So if anthropic and assure make 1000000000 from selling xx in friends, anthropic will start at their revenue.
When the clouds sell anthropics by the...
Andropic repositor at their revenue. And then I guess on the cost, I removes some of that as cost of good soda or whatever. So that's the issue. That one of the biggest. So we don't actually know how much andropic makes comparatively because what they're recording is topline revenues.
I actually think we're just going to started. think it's quite possible. Like you've used loads of agents how faulty right now still. We like 4 years into the whole thing.
Yeah, I mean, I think I'm getting better, right? I mean, if you spend, I think I would say that in terms of like being like production ready.
40%. Yeah.
But has it saved you money?
I think I'm playing. I think I have a good example. I think it's made me very productive, but I don't think it's making me money yet. So this is also the problem, right? So someone like me, I know that.
money on the coding, so it's for people who are using it well.
Yes. Yes. So the thing, for instance, for me as well. I'm using it for cooking. I've got like 6 projects and by creating and doing a lot of things. But what I wonder is, I'm gonna have money casting, and how people are monetizing this, like if you have customers already, if you're already a big customer, and you can keep ship code foster for sure. Yeah. But are you really making? Money from the cool, you're shipping as fast as you're shipping it. This, I don't. It's hard as it. How does that work for you?
So I'm an optimist. Um, but I'm skeptical of the LLM, uh, and huge LLM companies, like, Anthropic, include, uh... Not because the Korean stuff is brilliant. Um, I'm optimistic on stuff like, uh, isomorphic labs are doing on a lot of the financial-based AI that is not an LLM, but using fun time architecture. That's...
Who is doing this astromophic lab?
Isomorphic labs using for drug discovery. Yeah, I think so. They're trying to take like a 15 year pipeline to like a few years, which is amazing. So bullish of all that stuff.
Um, Prevent pollution, like just general and becoming a GI and I kind of like doing everything.
It's not that I'm not bullish. I don't know technology down. I don't see the point. I might let people just try to use an innovate, right? But there's so much hype in the game now. When Apple would innovate. Steve Joe wasn't going around for the US saying the Mac is going to be the greatest thing about the iPhone style.
Well, that's D-Lo style, though. Like, God is always, this is always really long style.
Steve doesn't go by, like, the iPhone's going to make Apple work for my new company, it's going to be worth trillions. He just builds the product. to the best of his ability, shipped it, and then let, you know... So there's a lot of hype in the game now where someone was like, oh, this is like game changing, everything's going to change, blah, blah, blah. nobody knows, right? For sure.
So that's why like Buffett's question is also interesting because he's like, these people are making a huge bet. It is a real money bet. And he's not saying it's wrong or anything, but he's saying if you go historically over the bust, whatever, 100 years. There are lots of instances of things coming on which change the game, innovation. And that will.
And the chances of that happening in the age of AI, even like 10 X. No problem.
Possibly. But there's also part of me, the things like, if there's a safest pet probably is building out compute, right? Because if there is going to be AI, there will be compute. And if you don't need the computer now, because people are saying all sorts of things. Like some people are saying that even though you utilization of the GPUs in data centers is not close to 100% and all that stuff. Maybe we just push it down the line, right?
Something is impressive, something keeps paying for it. I mean, if anthropic, even anthropic, I think a market will hold, if anthropic, you know, um, show is really Q3, some sort of their revenue number, show that they actually like printing cash like that. And if they show up for like, you know, Q3 will like 1000000000 in free cash flow or something like that, then that would be like truly transformational. I already heard that they have 1000000000 already in free cash flows.
She's right, like 119000000000 across multiple phases and SpaceX is taking most of the cost. Tesla's taking a bit, I think.
He said, SpaceX file the paperwork plant with an initial spend of 55 billion. There is a lot of capitals. Where are they actually getting the money from? They only have 100 billion.
Like how are they paying for the GPUs?
So this there are fab is. Is this, are they making? Is this the legal?
Yeah, the 55 billion. It's for the terrified blonde.
And the full buildout could reach 119 billion.
But I don't understand. They're making their own chips.
They're building a factory to make chips like TSMC, essentially.
And they're gonna buy the, they're gonna buy the lithography missions from ASML, something like that.
I think he's even looking at trying to copy some noise. Yeah, which will be interesting.
Like from today, he doesn't make shoes, right? I brought comma somebody like that makes it...
They do design chips, like the chip in Tesla. The Dojo's and stuff. is made by DSMC for Tesla. But that chip design team was designed.
He's actually himself on it, someone deep said for him. Yeah, the plan is that they want to now stop taking it themselves.
But if Tesla Chip makes sense, right? Because you don't want to go for a generic, um, and video, whatever chip, uh, some cars have the AMV. It's better to have a chip design purely for the functions, computations that a car has to do, right?
Yeah, because you don't need anything else. Yeah, let's play the video, man.
So that's interesting. There's a lot going on. And I might have a good experience. I know if we, it's just, at least, how to experience, yeah, he knows, James, I come to, but even because of the, he didn't have a project, you know, six size across, in competition. And asking, SCR, it's okay, guys, that's easy to be a bit nice. But you said you want to learn, is that living in us, he's crazy, in 2000, he's crazy, right, on an absolute idea, like, okay, should flywise, and everyone laughs for 3 years, and you know, actually, I get impacted. Do you actually like deciding to wear a far? Yeah. The picture is like a reactor, doesn't it? My physics is, it's possible.
The finances, you know, no sense, at all. And there's no money, as we say, is that this long, it's not too brilliant. So very likely history. Okay, he's right. Like, actually, like, one of the 3 things they, you know, we don't, a lot of these cases, accessible. And that the only way is space. No, because that makes, besides in Amazon of 1997, the biggest bubble ever story. Like, really dumb question comes out with just one number, $200 per kitchen. And you tell me the comments, because I, you know, Do you think this is a public, our greatest promise ever made my... That's what a nice idea is, if you not speak language, you can see the light, but a 2 super channel. I know most of these activities, please, please, very much. see you.
next one. Bye. Interest Yeah, some good high-level points. I think Staling is... more to be seen.
Yeah, I mean, I think in TCD, I think that I think I think even he's done a lot of impossible stuff that shows like this is possibly...
He also not done a lot of impossible stuff. So it's like, which side are you on?
I feel like, you know, I feel like he's always consistent. And you think that he's just got timelines stick longer than he projects them. My friends is like, I mean, there's so job and he's seen that for years.
I actually generally don't know whether the last one, the same computer.
Um, I hope it can be the one% of the thing that keeps it from being, but women already did.
We haven't had to solve the last one percent. We almost done achieve. So what we went to done is, um, so Tesco's got this 1st only scalable approach for self-driving. Like if they crack it, it should drive most places. Way more has taken London. Like, you know, way more used to launch in London ages ago.
Yeah, they might be London right now.
had that. Yeah, so they map these cities. They're geo fans, they, you know, map everything in the city and then they drive. That's not how human drivers are. Yeah. No, they use LIDA.
These light. Yeah, that's not the easiest camera.
So using LIDA, it's kind of like... It gives you a similar signal to what it... I have a problem with having crutches, right? If imagine like you have what? You have eyes, you can smell, you can hear, And If you, those are all like different things, right? And they alert you to something.
Like if uh, if you're driving and you smell smoke via something triggers in you, right? But you don't have senses that overlap, right?
that, Right. So do one thing, right? You don't have, like, no, like, uses to see. the 2 eyes can see, but the news doesn't see, right? Yeah, I mean. So yes, nice functions, voice.
Yeah. So I, I, I, I, I agree with his...
Interesting argument. So your argument is only lied there because he's giving a camera is doing.
Yeah, I'm saying like you essentially it's a crutch for the camera. It's like the camera may not see this, but light over the detective. And that confuses. I think it makes it a less good algorithm.
So you want to use the one that like trusts one sauce and that sauce gets to perfect and rely.
Yeah, which is what that's supposed to say. You saying is vision only, and we will get good at vision. Philosophically, I agree with that approach. I just don't know if the level of machine intelligence whereat now.
It's good enough to do that. It's good enough information for that one%.
Yeah. You know, they, uh, did you see that part in the Orland podcast where they talked about how someone's mapped the fruit fly's brain?
Who? There was what? last week? Yeah, I haven't watched that. I'm still like 20 to 7. actually been watching that.
So you know a food flight, right? A tiny little thing. Yeah. Even smaller than a fly. An annoying little, uh, their brain had 50000000 signapses.
So I'm not the whole thing?
Yeah. And that's just a fruit fly. It doesn't do much.
So 50000000 synapses every second.
Uh, no, the connections, right?
Oh, it's an absolute, the things that I connected. 50 million.
Something like that, yeah. And that's crazy.
Compared to a brain, right?
So compare that to our silicon intelligence now and I don't know. Maybe we're getting false positives and we still have a lot more. Uh, not unusable. We can still use it for stuff, but as a tool for humans.
Things are getting, even this models. There are 2 innovations that happened this week. The company announced yesterday that they're able to compress models by 90%. For my ex too, I'll send them to you. So they picked, took a model and then compressed it by, I forget which one it was.
Actually, you're raising a very good point, which was one of my points for a long term.
Like the models can, like, you can run that.
Models will be runable on your device.
So yeah, so that's that. I mean, obviously, I would have to DJ Xbox. For instance, I'm running Gipsick today. So some of my cards, I think, so that's number one. And then yesterday, um, a 2.5 um, um, gigabytes model, um, He's as good that it launched those as good as like a 10 or 12 gigabytes model, and it was, it was like trained on traces of Opuklon Hermes. So you can run your open corner. I actually put it...
Depends what they mean by good enough. If it's some, like, weird evil.
Yeah, let me see, because I actually put it on, because I immediately I saw it. I put it on my, um, because I mean, I have a privacy, right? Oh, yeah, I have a, oh, you can't see what I'm... So, um, so let me see if it came out, but I put it in, so this 2.6 by its model. So I started benchmarking it. Let me see what it's called on.
Which standard emails, right?
No, this is my own email.
Let me see. Oh, you only love. Colonel.
Then publish. Moving into audit now, is that did you publish?
So this actually hasn't done too badly. Um, this is that it, that's it, that's it. This is that the food is totally timed out, Lanti. But yeah, that run. 40 records are eaten. But yeah, but this model is a 2 This model is um 2.6 gigabytes model and like it can run Hermes agents, you can do an open call agent.
You know, so yeah, even my agent agrees. This is one of the rare insane IQA. This is like this is like one AM last night. Um, you know, and so on and so forth. But yeah, but I wanted a benchmark because for every single model that, um, begins being ordered, it's called them.
Do you ever run out of context window?
Yeah, I don't know the context of the company. Yeah, but it compresses that. I used something called LCM. Okay, okay. As LCM. So let me see what my benchmark, my benchmark looks like. So I have them. So right now as I'm talking to you at least, I run, um, so I have, um, I have the, obviously, the GPUs that, that, that, that city, magic, San Francisco. Then I have my own server at home. So all my services run on my server. Um, So I have a Mac. I bought the Mac, I'm the 125, once we get gig Mac.
Um, so that runs um, so I run a lot of my services on there. And so I do skill in, because I saw this skill in your stuff, so you use it. So I do skill in, so all my devices are all kind of connected to it. So this is my own benchmark. Um, so every single model comes in, like I benchmark, benchmarking models for the last, like, few, few, like months, like every 6 months. Every time a model comes. And so what I typically do is I take my traces across all my agents.
Um, and then I, I ask my agents to create a new benchmark based on the type of work we do, so my work is very cool, I, I write code. I do sales, I do like a bunch of stuff. Um, so it's Benchi.
Right? So, um, I benchmark a bunch of things I've been on prom, providers, browser, memory. I just did an image one that hasn't been provised, but I, I mean, mark every single thing I use. So this is the deep seek. This is that deep secret that came out. V4 flash. This runs, this is running on my, this drum is on my servers, my 2 my 2 spots. See, you see why it hits on my own wood. It's, um, it's a 200 um, it's a 200 gigabytes, um, but the 2 DGF spots are, 128, 128, so you can run on it.
So you can see how Feeble is here. From my kind of work. I think this is maybe fibble meat. This is some people who have people who I might do better than so. Um, so Laguna, I'm sure you know when Laguna came out. This is where Lagunas goes on that chat. Then, um, where's Kimi 3? didn't I benchmark give me 3 yet? I don't think I benchmarked Kimi 3. But yeah, but they haven't published this model, but when I publish it, that's how I'm gonna know. If the model is, um, well, finish, and publish.
Um, But yeah, but, um, let me show you, let me send you that one now before I forget this, the one that, um, So model compressions might be the um. Might be the next crazy thing.
For finance, I've always been like, people will trust user on device financial advisor. It's private. But here's my question for you because you, it's an important point. If inference moves the device, how does that change the infrastructure business for space data center business? Yeah, because Apple is going that way. You think Apple is quiet, I'm losing this race, I'm not going to.
not losing the race, that the most valuable thing in the world now. Yeah. So everybody thought you were for a while. And other people, I'm wondering if Apple where Apple is now is strategic or incidental.
I think it's culture. I think it's like they have a very pure way of, um, analyzing user experience and product. And I think they wanted to go down an LLM route and they put some features, but as they tested it, they figured out more issues. So I think they're up to something. I think they're not convinced.
Yeah, I don't think it, but Apple is definitely, Apple is obviously not doing. It would be crazy, right? If Apple doesn't, because I feel like he's never done an element. I mean, I mean, they did a bunch of things around locally now. So you know every MacBook now, every US ships with like some local alarm.
Well, yeah, I spoke to this guy whose ex-apple now. He worked for Apple for a long time and I was telling him all these things. He's like, you think Apple are not doing it. But Apple are doing everything. They have an internal project and product for almost everything under the sun and things you never imagine. I said they just decided not to launch it. It's not that they have...
No, but I think if Apple was doing other lines, you would know, right? Because they were like talent. Element talent is not mainstream, right? It's the same people in San Francisco. Like they Google hires them and then think to hires them and then Tropicas. the same people, right?
So the actual core LLM talent. The team is not very big. If Apple need to do it. It might be 10 people, 20 people, 30 people, 40 people, it's the data cleansing team and all the people.
had an LLM program that was valuable, you would know, right? Because I mean, they would buy data from someone.
But the thing is, you're thinking like the rest of the industry, like Zuckerberg, open AI, Anthropic, they all believe that they're these, or maybe opening iron metamor, these crazy, expensive engineers were geniuses, and they're only a finite number of them. But I think Apple is more 1st principles, they're like, If you actually look at the Transformer architecture and theory, it's like, there are many people who will get up to speed in it.
It's just that opening I have the learnings of having to build that and learn from it and meta, like, okay, they have those learnings if we get them, then it'll be quicker for us. But Apple are more of a research iterative approach. So maybe they're like, no, we don't need this guy just because they've gone through the problems that are open.
Yeah, they just do 1st principle, start from scratch. and you can understand the basis is good enough. Right, a celebrity. But you also do know that they hit Gemini, right? to kind of like have the, I have, I don't have access to it, but I'm gonna telegram with a few of my friends, San Francisco, who...
I could probably get it on the beat.
Yeah. But I haven't want to be there. Yeah. So they didn't find impressive with the series. So the series is quite good. Really? Yeah, that's what they said.
It must be like, yeah, it must be like amazing. City is so frustrating.
Yeah, say they say, yeah.
I'm now at a level where I speak so much to AI. that I find AI pad. I just think it's so dumb on lots of things.
Yeah, I use it enough that I know it's not a... So for instance, that this is a popular guy. I'm the engineering side called Stevie Egg. Yeah. So the guy read something, read an article that published like 2 days ago when he was basically talking about that he now believes that AI is a conscious, and he should talk to AI, literally like a human being, like blah, blah, blah. And my tutors are basically, I listen, these things are too, right? These things are unconscious, like the guy, and then so on. And then like, you know, the guy on my comment was like, yeah, prove that they are not conscious.
This is like a bloodbath. What's happening? It's like down close to 14%. The more shit, doesn't like it.
Still dropping today. I agree with this. Space 6 is heavy on ambitions, but light on details. It's like Elon has come with these ideas. Okay, it's unfair. He's just an idea because he is, they have a plan. And say, fake their sense, starting mobile, but nobody knows how it's going to work.
How are you gonna get to this? I have all these questions. those things are all a vicious, like very ambitious, dude.
On Starlic mobile. I don't see how he will eat into the general mobile market share because it's so cheap already.
Man, it's not impossible, man. The guy already makes like 5000000 from like something he lost like 3 years ago, 4 years ago. So it's stunning.
Starting for all the 5 years. You sure? Yeah. The starlic has always been great. That's amazing.
I think it's not that. How many years is it, has it been, like, 10 years?
No, in 2019 where we were looking at SpaceX to invest. Because already stalling that scale them? No, yeah, one of our biggest reasons was stalling is great business.
Now 7 years ago. Yeah. Yeah, yeah.
People like starting is a great business.
There's almost a 10 year old business and it's not a 5 year old business. Yeah.
And we were also like starting, um, but we never imagined it as a good, everyone will.
That's the biggest. You didn't ask it as a mass market business.
We saw it as like a premium business for people who want coverage everywhere.
Yeah. You love this, man. I like that Elon keeps challenging himself because, I mean... Yeah, that's great. He keeps, like, he keeps thinking things like, because, I mean, at some point, the question he happened, like, starts 3 seconds, can they launch it? Can they make it reusable? Right? For screen for a while. Tesla didn't have any. I mean, yes, self-driving was kind of like a question. everybody kept asking, can they make yourself driving? But Tesla wasn't like an like there's nothing else. the event that's soft, right? invested invented electric cars and that's it, there's no new innovation that is transformational. And today, Well, yeah.
South driving is a big one.
No, no, that's what I'm saying. Apart from self-driving, which again hasn't gotten there, but like, so I feel like what, so my plan is, you don't used to eat like big ideas bite size. Used to be cool. lets go to space. We need the dreams, yes. But now it feels like all the dreams are all happening at the same time. There's so many things actually.
When SpaceX was, even when I invest in SpaceX, it was only a $100000000000 company, right? And Elon, um, and Tesla was like 30 billion. Now they're all $100000000000 businesses. There's a lot of public money in these businesses. And...
So you need space. Expressing markets was more expensive than was more valid than Tesla in public markets?
No, no, no. Tesla was always ahead. Okay. SpaceX went from 300000000 to like 90 or something like that. But by then Tesla was in the 100s of billions. Okay, okay, okay. But now he's a $10000000000 bets, but it's more than a launch business. It's like it's worth 20000000000 because like the guy said as well, it's based on promises.
People are buying the Elon Fox.
So everything from time to, if you want to... What do you believe the probability of these things happening now? This can be absolutely huge. But then it's dependent on will Starship work, will it get down to £200 pillow, will inference, actually, you've mentioned 3.9 and considering so much. What happens when inference? 10 X's. So what are the data senders in space supposed to be useful? Is it mainly influence or is it mainly?
Training.
So it could be both, right? But I mean, bulk of. I mean, thinking 90%, if the bulk of the world switches on, 90% of data center has to be in friends, because there'll be a lot more people using it versus like people would just, you're not going to have that many training models all the time. Maybe, yeah, maybe a few companies will find you in the models, but no, it'd be more interesting.
It's great because latency doesn't matter. It doesn't matter so much. Yes.
Yeah, but I think it's largely in France. I mean, France is very high margin. I mean, for instance, is a very high marching business. If he takes him, if it takes him one year to pay back because he sued the thing for inference, which is what he had. Taught them like Bruv. The thing just, he has another 6 years or like 5 years. Because I think these GPUs are set, the lifetime of GPUs are about 6 years, five, 6 years. And so if he pays back in one year, then like, rah, such a profitable business.
It depends which version of the world. No, but even that's tricky. I do think models will get...
Even if China wins, they still need infrastructure, you still need data centers. So if open source wins, then the model, the part gets sticking from the Krucos models and given to the clouds and the new clouds, so the clouds, AWS would deploy Kimi, the deployed Kimi, and you would buy like internet LRF in France from AWS, um, and we'll buy from open router, guys like, who's actually the business I want to start. Um, so my the board for you, I also have, is I think in France is going locally. Um, but...
I think in from school, local, I think models will be more... Yes. There won't be one AGI. There'll be many specialists more.
Yes. But then I also think that you still have big models that someone needs to deploy, but then a lot of the open weights are out there. So you're going to need like, you know, Henry, who's got like a, you know, 50 GPUs somewhere in Lagos or somewhere in London, and then I sell it for like a few corporates, and then that's my business. And then, you know, I make a couple of 1000000s, a couple of 1000000s and you're going to have like 1000s of small businesses selling computes.
It's for influence, apparently. Yes, inference is the book. So then what happens?
For instance, the book.
Because, like, if you decentralize, and a friend of mine who runs a Delco, um, also actually is, is trying to, is running a walk on this so similar idea, because if, for instance, I can buy like 20 DG Xbox, and then I can produce like a 1000, or let's say, 2000 tokens per 2nd and I can sell it to life, 20 businesses a month. That's decentralized. Like there's no data center. There are lots of people doing it. Like, it doesn't blow my popular by a thousand. I pay maybe an extra like 50 bucks a month for five.
The question is, is there always going to be a need for more meaty cloud inference?
You mean the big ones? Yeah.
Depends on the one you believe in. can believe that. If you believe that open source will win.
So, if you see, I said, what happens if your inference moves on, divide you more efficient models, says...
Yes. Destination, and this thing is already happening very fast. Like, deep sick is so, is like 10 nights cheaper. At the same value compared to like a lot of the American models.
Yeah. It will never get smaller.
Yeah. So, actually, I think it's both ways. Frontier would keep multiplying. You would keep needing, like, people class models on the cloud, but like me and you to write emails, write code, movie local.
So what happened to Moans?
Yeah, we're not doing miles anytime soon. Because there's no economic, like, there's no, because this is, this is the problem with Elon's thesis on the thing is like, there's so much complex required. to do this business, but you need a, you need a reason for people to go to space. And just selling a dream of where you want to go to Mars is not enough.
But I'm actually very impressed with the Capex or StaceX launch services. For 3000000000, they're billion starship. That's insane. That's like the best user capital I've seen. That's crazy. So why are more people doing stuff like that?
Yeah, because it's technically hard, right? That 3000000000 is like master deficiency kids.
Yeah, I think I'm, I think I'm still, I mean, obviously, I think this is like a very general, I mean, intuitively tells me like, shit, like, yeah, it's not all bullshit, man. Like, there's some substance in there.
It was based on execution.
And Elon is a king of execution, right? He's not really been...
I mean... He's the king of execution in certain types of things, like building hardware, like building...
So but your argument is like doing data centers is not beauting anything.
It's different from what?
It's just assembling. It's much more commodity because a lot of people can assemble.
So no matter how good or hardworking you are, it's hard for you to compete. It was hard for you to compete against the quality or the electric Tesla was big, right? No matter how good you are. For example, Jeff Bezos in other endeavors, um, it's hard for you to build the roughest in SpaceX. But in terms of things he's doing now. XI is doing exactly what other people are doing. He failed the model.
You couldn't train any models? You feel that that?
Yeah, well, they're not the greatest. I'm confident, but they're not good enough. Same with the self-driving. Excellent effort and the best effort, but it's actually achievable. But then data center business. Is it a commodity business?
There is enough of space is not a commodity business, which is what he's backing. Yes. Yeah. So that's the mood. The mood for this data center. Because I don't think he makes any I mean, I'm thinking he's still ramping up. And then if you remember, Nvidia said that he's the, he's the person who's, because I think he took um, Colossus one to, the production in like 9 months, which is the fastest rump anybody has ever done. Yeah, right? So, I think you might have an LE because you can ramp very quickly. He's quick.
Yes. But then I guess like, yeah, then the long term is eventually everybody can get there too. And then if people get there, like, so I guess in a lot of the idea is hinged on, would there be with the growth for computes keep increasing in the pace that, even with every data center, Ionas putting, they still need for more. If then, if this heat continues... I think it's that the business is stable.
If it don't.
Official pages and authoritative references for named companies, products, people, programmes, and reports discussed in the recording.
Links identify the resource. They do not validate claims made during the conversation. The machine transcript may have garbled some names.
Cited repeatedly as a customer driving SpaceX's AI/data-center compute revenue; its contract terms and reported free-cash-flow figures are debated as evidence for or against the compute thesis.
Mentioned as: Anthropic
Discussed as either deliberately restrained or genuinely behind on LLMs, with on-device model strategy raised as a counter-model to centralized data-center compute.
Mentioned as: Apple
Referenced as the lithography-equipment maker SpaceX/xAI would need in order to build an in-house chip fab, underscoring the capital scale of that ambition.
Mentioned as: ASML
Named among telecom incumbents in the same market-cap comparison.
Mentioned as: AT&T
Cited as the comparison case for how hard it is to compete with SpaceX on rocketry, even for a well-funded rival.
Mentioned as: Blue Origin / Jeff Bezos
Used as a manufacturing-margin and unit-sales comparison against Tesla.
Mentioned as: BYD
Used in the telecom-incumbent comparison to argue broadband is a commodity business that Starlink can only modestly dent.
Mentioned as: Comcast
Cited as a much cheaper, competitive open model, used to argue AI infrastructure margins could compress under open-source/China competition.
Mentioned as: DeepSeek (also "deep sick")
Discussed regarding autonomous flight/tracking features, relevant to the broader autonomy discussion around Musk's companies.
Mentioned as: DJI
Telecom company acquired by SpaceX/Starlink reportedly to support direct-to-cell/mobile service; deal scope not detailed in the conversation.
Mentioned as: "Echistoft" (ASR garble, near-certainly EchoStar)
Named alongside Anthropic as a source of compute-related revenue feeding SpaceX's data-center line; also referenced regarding Gemini.
Mentioned as: Google
Named in the Chinese consumer-tech tangent as a value-for-money example.
Mentioned as: Hisense
Named among Chinese tech brands in the same tangent.
Mentioned as: Huawei
Cited as a non-LLM AI example (drug discovery) used to argue AI value can exist outside the chatbot/compute-rental model being debated for SpaceX.
Mentioned as: Isomorphic Labs
Discussed in a tangent on Chinese consumer tech relevant to the broader China-vs-US technology competition framing.
Mentioned as: Lenovo
Cited by a named contact as having previously attempted similar satellite-to-mobile base-station technology without success, used as a bear argument against Starlink Mobile.
Mentioned as: Nokia
GPU supplier referenced for financing/discount speculation and for reportedly praising the speed of the Colossus ramp; also used as an example stock in an unrelated leverage/margin-call anecdote.
Mentioned as: Nvidia
Used as a revenue and reporting-transparency comparison point against Anthropic in assessing whether AI-infrastructure margins are real.
Mentioned as: OpenAI
Discussed as building telemetry/spend-tracking tools to measure AI-agent productivity, part of the broader argument about whether AI infrastructure spend is economically justified.
Mentioned as: Ramp
Referenced via an anecdote about acquiring cheap mobile spectrum licenses decades ago, used as a historical parallel to spectrum-driven telecom value.
Mentioned as: Reliance
Used as a second UK broadband/TV comparison point against Starlink.
Mentioned as: Sky
The central subject of the entire conversation; all three business lines (launch, Starlink, AI compute) are discussed as parts of it.
Mentioned as: SpaceX
A YC-backed startup building space-based GPU/data-center infrastructure, discussed as a direct competitor whose viability is questioned given SpaceX's launch-cost advantage.
Mentioned as: Starcloud
Referenced in the telecom comparison, including its stock performance and valuation multiple relative to peers.
Mentioned as: T-Mobile (TMUS)
Treated as part of the same vertical-integration bet as SpaceX; its FSD/Robotaxi claims and manufacturing margins are used as an adjacent credibility test.
Mentioned as: Tesla
Referenced as the manufacturing model for a rumored SpaceX/xAI chip fab and as Tesla's Dojo chip manufacturer.
Mentioned as: TSMC
Its internal tool for measuring org-wide AI agent usage and cost is cited as a real-world example of quantifying AI productivity, feeding the same infrastructure-ROI debate.
Mentioned as: Uber's "ADR" tool (exact product name uncertain in transcript)
Cited as an example of humanoid/robot-dog products mainstreaming in China, part of the robotics side-discussion adjacent to the vertical-integration thesis.
Mentioned as: UBTech
Part of the same telecom-incumbent market-cap comparison used to bound how much value Starlink Mobile could realistically capture.
Mentioned as: Verizon
Used as a real-world price and reliability comparison against Starlink from a UK household perspective.
Mentioned as: Virgin Media
Used as the comparison point for Tesla's self-driving deployment scale and approach (LIDAR vs. vision-only).
Mentioned as: Waymo
Discussed as part of the Musk-companies vertical-integration thesis; its Colossus compute cluster ramp is cited as an execution proof point, while its model quality is questioned relative to Anthropic/OpenAI.
Mentioned as: xAI (also transcribed as "XI")
SpaceX's proven, currently profitable launch vehicle, used as the credibility anchor against which Starship's promises are measured.
Mentioned as: Falcon
Google's AI assistant used live during the call to pull SpaceX financial data; separately discussed as possibly licensed to Apple for Siri.
Mentioned as: Gemini
xAI's model, discussed for its training-vs-inference compute split as a proxy for how xAI's compute is actually used.
Mentioned as: Grok
Open-weight model cited as an example of clouds deploying open models instead of proprietary ones, relevant to the margin-compression bear argument.
Mentioned as: Kimi (Kimi 3); exact model version uncertain in transcript
Tesla's driverless ride service; its headline driverless-mileage claim is scrutinized in detail as a proxy for how much to trust adjacent Musk-company claims.
Mentioned as: Robotaxi
SpaceX's satellite internet business, discussed in depth for consumer/enterprise/mobile segments, live-checked reviews, and its addressable market.
Mentioned as: Starlink (transcript ASR frequently renders this as "Stalling"; corrected from strong context)
SpaceX's next-generation rocket; its unproven cost and cadence targets are the linchpin of the data-center bull case and the physics-based bear case.
Mentioned as: Starship
Used as a customer-service benchmark against Starlink's reported poor support.
Mentioned as: WhatsApp Business
Central figure across every business line discussed; his promises, pricing decisions, and a hair-shaving pledge tied to Robotaxi timelines come up repeatedly.
Mentioned as: Elon Musk / Elon
Referenced regarding possible GPU financing or discounts extended to Musk's companies.
Mentioned as: Jenson / Jensen Huang
A referenced clip/quote about large capital bets on AI and data centers frames the episode's broader question of whether current leaders could be upended by innovation; exact quote content is garbled in the transcript.
Mentioned as: Warren Buffett
xAI's GPU supercomputer cluster, cited for an unusually fast production ramp, used in the conversation as evidence of Musk-team execution speed.
Mentioned as: Colossus
Referenced for real-world heat-output figures used to argue the radiator-mass problem facing orbital data centers.
Mentioned as: the space station / ISS
Cited as a SpaceX launch customer, used as external validation of the launch business, and as the source for orbital heat-dissipation reference figures via the ISS.
Mentioned as: NASA
The accelerator that funded Starcloud; also referenced via a podcast episode used as a source on AI business-model economics.
Mentioned as: Y Combinator / YC
Referenced as the outlet that may have quoted Pritam regarding an early Tesla investment thesis, though the speaker himself is unsure of the actual source.
Mentioned as: New York Times (speaker uncertain if correct outlet)
The review page opened during the call informed the service-quality discussion. It covers a UK installation vendor, not Starlink itself, so it should not be treated as a clean measure of official Starlink customer sentiment.
Mentioned as: UK Starlink review page / installation vendor page
Cited as a reference source for measuring AI token productivity and business models, informing the AI-infrastructure-ROI tangent relevant to the broader SpaceX compute thesis.
Mentioned as: "Pedro's YC episode" / Y Combinator podcast
The optimistic read in the conversation rests on Falcon's proven economics, Starship's early signs of progress, and a newer high-margin compute business layered on top, with the biggest unknowns pushed into the future rather than resolved now.
Arguments below synthesize the conversation. They are not investment advice and do not verify the figures discussed.
Falcon 9 is repeatedly described as one of the cheapest and most reliable launch systems available, generating real profit today and hard for competitors to replicate at the same cost structure.
Despite being unproven, Starship is credited with a successful batch satellite launch, treated by the speakers as a positive signal for the broader reusability and cadence bet.
The conversation cites a large reduction in cost per kilogram to orbit compared to historical baselines, framed as SpaceX's clearest demonstrated execution advantage over competitors.
Contracts tied to Anthropic and Google are discussed as feeding a data-center/GPU-hosting business with margins described as strong, potentially becoming a major profit center as those contracts fully ramp.
Despite the review complaints, the speakers acknowledge Starlink went from a speculative niche idea a decade ago to an operating business with a large, growing satellite fleet and a genuine use case in disaster response, conflict zones, and remote connectivity.
Combining launch, compute, connectivity, and (via Tesla) robotics under related entities is discussed as a potential edge no single-purpose competitor has, even though no part of the stack has fully proven out the integration thesis yet.
A rough, speaker-built valuation exercise during the call implies that the launch and Starlink businesses alone could justify a valuation in the hundreds of billions, before crediting any of the AI/data-center or Mars ambitions.
The skeptical read treats most of SpaceX's forward story as promises stacked on a proven launch and Starlink base, with the AI-compute and orbital-data-center layers resting on unresolved cost, physics, and durability questions.
Arguments below synthesize the conversation. They are not investment advice and do not verify the figures discussed.
The entire data-center ambition depends on hitting a much lower cost-per-kilogram target and a launch cadence far above what has been shown, and the speakers repeatedly flag that neither is proven yet.
Heat dissipation in vacuum reportedly requires radiators roughly ten times the mass of the compute they cool, which the speakers treat as a serious, possibly disqualifying constraint on the space-data-center thesis rather than a solvable engineering detail.
Live-pulled UK reviews show recurring complaints about reliability, customer service, surprise price increases, and post-cancellation billing, with no official review channel found, undercutting a mass-market growth narrative.
Most users already have adequate fiber or mobile broadband, so Starlink's genuinely compelling use case is argued to be a smaller niche (rural, disaster, military, remote) rather than a broad broadband replacement.
Hyperscaler-style contracts are described in the conversation as reportedly cancellable on short notice, which weakens the case for treating this revenue line as durable, recurring, high-visibility income.
After roughly fifteen years of development, Full Self-Driving is described as still unresolved on edge cases, and a headline driverless-mileage figure is picked apart to suggest a much smaller number of cars actually operating without a safety driver.
If the AI infrastructure line is mainly buying GPUs and reselling capacity, the speakers argue it resembles a replicable cloud-hosting business rather than a defensible advantage like reusable rockets or vertically integrated manufacturing.
Starlink Mobile, orbital data centers, an in-house chip fab, and Mars are all discussed as large ambitions announced with little operational detail, and the speakers explicitly note that large sums of implied value depend on these panning out rather than on demonstrated results.