저자: CathieWood · 외국인
요약 보기
테슬라 FSD가 네덜란드에서 승인되며 EU 전역 확산 기대, 이를 구독형 고마진 매출과 로보택시 확장의 계기로 봄. 스페이스X IPO 및 xAI·테슬라 합병 가능성을 언급하며 궤도 AI 데이터센터와 스타링크로 TAM이 수천억~수조 달러로 확대된다고 주장. 발사 비용은 현재 약 1000달러/kg에서 스타십 완전재사용 시 100달러/kg 이하로 낮아진다고 전망. 테슬라 주주는 장기 인내 시 보상받을 것이라는 시각.
다듬은 글
Big milestone for Tesla today. The official Tesla AI account posting this meme with the caption Tesla's in the Netherlands right now. For those listening to audio only, Neo from The Matrix with the caption I know driving. Netherlands has approved Tesla's FSD, which makes it the first European country to do so.
This is big. Jeff Lutz in the comments adding this, Netherlands regulatory approval will be adopted by 27 other EU European Union members likely in the next few months. Basically, other countries in Europe need not have their own regulatory or new test requirements is the key takeaway. This is a big deal, and it certainly is.
The floodgates are about to open. Keep in mind, despite the fact that Tesla has a global fleet now approaching 10 million vehicles, FSD is not legal, not approved in most of the world. Every country that approves FSD immediately enables Tesla to offer this product to their customers, some of which will immediately subscribe.
Tesla doesn't have to spend any additional money, and suddenly they now have customers paying high-margin recurring revenue. Not only is FSD going to produce an enormous amount of high-margin revenue region after region as it's approved, but the approval of FSD also somewhat paves the way for Tesla robotaxis in those same areas.
Now, over to Ark Invest, who, just as a reminder, have a current fair valuation for Tesla for 2029 predominantly on the back of successfully scaling robotaxis of $2,600 per share. All right. So, the this basics IPO is is upcoming. Um Ark has done a lot of work on this. In fact, last year, Sam Daniel and Brett published a valuation in collaboration with Mock 33, but so much has changed since then.
So, so Brett and Kathy, what is our updated thesis on SpaceX? Brett, maybe I will start and then hand it to over to you since I know you've been almost living in outer space for the last few months uh trying to figure out what's going on here. We've been talking for a long time, Ark has, about the convergence among the major technologies evolving today.
And it was, I'm going to say about 5 months ago that Elon said, "Hey, you know, I think that my companies are converging more than I even expected them to." And that was a real clue to us that something was happening. And so, we weren't shocked to see the combination of SpaceX and xAI. And yeah, a lot of people are speculating that Tesla's next.
Uh I can confirm I'm one of those people speculating Tesla's next. In fact, I'm not even speculating. I'm telling you it's absolutely going to happen. The question is not if, just when. The synergies are so obvious. The overlap is obvious. Just give it time. We think that may be true, but we do think uh that Elon would like to see the scaling of robotaxis reward very patient shareholders over the next year or so.
But And on this point, which I've heard Ark bring up a number of times now, absolutely agree. I think it's safe to say that currently, as I record this, with Tesla's market cap just over $1 trillion, and SpaceX rumored to IPO somewhere in the vicinity of 1.75, maybe approaching $2 trillion, were the two companies to become one at roughly two trillion for SpaceX and one trillion for Tesla, Tesla's shareholders would get meaningfully diluted.
And at a barely $1 trillion valuation, I think it's reasonable to say that most investors do not believe Tesla's going to have a meaningful size of the global autonomy market, let alone a dominant share, let alone take the lion's share of the profits. If we have a look at this five-year Tesla stock price chart, November 2021, stock was almost 15% higher than it is as I record this.
Of course, as I've showed on this channel before, that's not the first time it's happened. Here's an example from February 2014 to August 2019, stock down a comparable amount, roughly 15%. A half-decade period where the stock went sideways. And then, well, we all know what happened after that, right?
So, what Cathie Wood of Ark Invest seems to be suggesting, and that which I agree on is, let's just say hypothetically, after a period of essentially going nowhere for half a decade, investors realize Tesla's going to dominate autonomy, the stock potentially moves higher, the market cap potentially is much larger.
At that point, Tesla shareholders would be much more evenly and fairly rewarded for their patience in a potential merger between SpaceX and Tesla. Assuming SpaceX doesn't IPO at two trillion and immediately go to three, four, five, six, and then never get any lower than that. But for now, we do have this major convergence.
We're talking about robotics and and AI converging with lots of other technologies today. So, I know that in the early days, Starlink alone, our first models were saying Starlink would be able to generate or the total available market would be It was at the time in the 30s, the mid-30s. And now, over the last few years, as we've really dug in and done the work, it has scaled to 150 or 160 billion in um TAM for the Starlink opportunity alone.
And we know that Tesla is 10 years ahead of any other company out there. And now, on top of the Starlink opportunity, we have the orbital data center opportunity. And again, 6 months ago, we were not talking about this. And I know uh uh how quickly we embraced it. We got it because we've been expecting these convergences.
But you've been living, eating, and breathing this model, I know, along with Daniel in the last uh in the last few months. So, maybe I'll throw it to you because this opportunity is so much bigger than Starlink alone. And just to clarify for those you aren't super regular viewers, the orbital data centers are, in essence, the Terran project, Tesla to manufacture chips to go in satellites launched by SpaceX to run AI training and inference.
Shout out to xAI. So, again, we find ourselves talking about the synergies between SpaceX, Tesla, and xAI. By the way, those satellites require gigantic solar panels. How can you not see these companies all becoming one in the future? Yeah, and you can see it in their filing. So, Starlink, they had filed for approximately 40,000 satellites for the AI compute, they filed to launch at least a million satellites.
So, the very crude way to think about it is, well, wow, we need This is 25 times more of an opportunity than Starlink was. And if you adjust for the mass of the satellites that we think you need for AI compute, it could be 60 times more than the opportunity that we thought they had with Starlink. And relative to our model last year, they we already thought they were going to launch 10x more satellites just to support Starlink.
So, from today, we think in terms of total up mass that SpaceX could launch to support these networks, it could be on the order of 600 times more than where they sit today. Now, why would they do that? It's because um we think they have a great return on capital when they launch these satellites. For Starlink, it could be, you know, on the order of 1 and 1/2 to 2 years between when they launch the satellite and when they've got the money back from the satellite, and everything else is cash flow to them.
And with AI compute, there's less certainty here cuz we don't know what their cost of manufacture is going to be, but they could have extraordinarily high payback periods if they're able to convert those compute flops into AI model outputs that generate revenue on par with what OpenAI and Anthropic are doing against their AI models.
Even if they don't, they could on-sell their compute into kind of uh terrestrial players like Google and Anthropic. And there, they would have maybe a four-year payback period on their satellites. So, you should think of SpaceX as having this unique-to-the-world capability of being able to inexpensively launch satellites.
And given capital, they can turn a crank of launching satellites, collecting cash flow, launching more satellites, and they expand. And over time, their cost of manufacture and cost of launch comes down. And Yeah, just to be clear, Brett is currently describing an infinite money printer, a feedback loop.
He mentions that SpaceX is the only company that can launch at those costs and at that cadence, which is true. Cathie earlier mentioning that we know SpaceX is 10 years ahead, and that was a literal statement. It took 10 years for Blue Origin, the closest, quote unquote, competitor to SpaceX to land their first orbital class booster.
They did it in 2025, SpaceX first did it in 2015. 10 years, literally 10 years. And Brett describing a feedback loop here where they launch satellites, those satellites produce a financial return that more than pays for themselves being launched. So, they launch more satellites with that increasing cash flow, on and on and on and on, infinite money printer.
So, the convergence of AI compute and the Starlink services has led to an expansion of their potential TAM. It literally could be in the trillions of dollars to support the AI compute opportunity. Along the way, they're um using Wright's Law and good manufacturing practices, they've lowered the cost to launch from $10,000 to now a little less than $1,000 as we model it, all the way to once they launch a fully reusable Starship, it could be below $100 per kilogram.
So, um you know, a real step change down in how much it costs to launch things up, which gives them a step change up in terms of the amount of opportunity they can access in space. And this is even before thinking about Mars. So, it's a very open-ended I mean, I think that people misunderstand that they don't have There's a North Star for them, which is making humanity multi-planetary.
Exactly. But between Mars and here is the moon and they believe they could build infrastructure on the moon to lower launch costs again to $10 a kilogram. For some really strange reason while they were describing building out infrastructure on the moon, Tesla's Optimus humanoid robots came to mind. And so that the only reason to do that is because the opportunity in AI compute is so large.
If they were still just doing Starlink, um it probably wouldn't be be worth the fixed cost to install a moon base to lower Earth servicing cost to $10 a kilogram, but the number of satellites they need to continuously launch into orbit for AI compute um probably justifies the engineering and upfront cost of building a moon base.
Absolutely. Again, a stepping stone on the way to Mars. You know I've got this really crazy idea, but maybe after all SpaceX is going to be printing so much money from Starlink and the orbital data centers. Maybe SpaceX could look for a company that's developing useful dexterous humanoid robots. Maybe they could collaborate with that company, acquire that company, merge with that company cuz after all, they're going to need a lot of robots for both the stepping stone to Mars, the moon build out, and Mars itself. Anyone know any companies that are working on humanoid robot.
댓글 0