18:24
give me $500 when I invested into SpaceX. I think I would. Yeah. No, I wouldn't. You would not go through. No, but I hope my index funds aren't automatically doing it. Hi, welcome to another episode of Cold Fusion. Space X is an incredible company with impressive achievements. It brought the price of space launches down by 85%. Objectively, that is astonishing. The company also made reusable rockets reality. Many critics said that was impossible. If you love your engineering and physics, yep, great company. But now there's a massive problem.
Recently, something has gone terribly wrong with the company. It's no longer what most people think it is. Let me ask you a question. What type of company do you think SpaceX is? You might say a rocket cargo company that operates satellite communications. For those who are more optimistic, you could say that SpaceX is an exploration company who will take humanity to the moon and beyond. Well, according to their own S1 filing, the documents a company needs to file before it goes public on the stock market. Both of those answers are wrong. Generally, S1 filings list the industry category that a company is under. For example, the industry code 3760 is for spacecraft. 3182 is for defense and
aeronautical. Other companies like Virgin Galactic, Boeing, etc. All file under these categories. But for SpaceX, they filed under the industry code of 7370. That's computer programming and data processing. Oh, it looks like SpaceX is officially filing to be an AI data center company. That already raises an eyebrow, but it's worse than that. To understand how, you first need to understand what a TAM is. TAM stands for total addressable market. It's the maximum revenue opportunity available for a product or service if you achieved 100% market share. It basically represents to your investors what the market for your product is. Looking at the prospectus, SpaceX's TAM is 28.5 trillion, approximately the size of the GDP of the United States. Okay,
that's pretty high. But that's not the bizarre part. The strangest part is only 15% of this market is space and communications, and it's written right here in front of your face in plain text on page 11. So 85% of the company is essentially just a rapper for yet another LLM and data center business like Anthropic or Open AI. I know some of you guys have commented that you're getting sick of hearing about AI, but this isn't funny anymore. It's getting really serious. When SpaceX went public, it was the largest launch of any company in the history of financial markets, larger than Amazon, Google, Alibaba, and even Saudi Aramco. In the first few hours of trading, SpaceX blasted past a valuation of $2.3 trillion, instantly becoming the
seventh most valuable company and worth approximately the same as the country of Canada. For some people, this seemed like a sign of strength, but for others, it was terrifying. The signs of an overheated market. But that's not all. The story of the SpaceX IPO is telling us something if we listen closely. It shines a light on the state of the current financial market. A corporation is hiding an AI company losing a billion dollars a month inside the brand of a perfectly healthy rocket company just so investors fund it. Let me say that again. In 2026, a corporation can disguise an AI company losing almost a billion dollars a month inside of a perfectly healthy rocket company, sell
it to investors, and get away with it. You know, your expectation is not going to be a good investment. And that's the part that I'm trying to understand. I found that never been a great bet to bet against Elon. I'll grant you all that. If you again look at the S1, the TAM, the total addressable market is defined at 28.5 trillion. 85% of that TAM is an AI. So even if everything you say about SpaceX and Starlink is true, the entire company is being bet on AI in terms of its future, not on space and not on Starlink.
Those are his words, not mine. It's a mess. And in this episode, it's going to be a wild ride. And a disclaimer, none of this is financial advice. If you want to buy the stock, buy it. If you don't, don't. But let's get into it. You are watching Toll Fusion TV. There's a lot of excitement about what's going to happen. Everything they can possibly do to guarantee a pop is being done. This is going to become the largest company on the planet. For over two decades, SpaceX earned its dominance the hard and old-fashioned way. Three catastrophic early failures, a near bankruptcy that must described as existential, a methodically driven slow
rise, and finally a near monopoly on orbital launches. In a company full of moonshots, if we isolate Starlink, it's the only thing that actually makes money. The Starlink internet service has a constellation of over 9,000 satellites and generated 11.4 billion in revenue and 4.4 billion in operating profit in 2025. But the rest of the company is another story. And trust me, we'll get to that. And from here, the controversy begins. When people talk about the broader market, they're referring to the S&P 500. It includes all the big names like Apple, Nvidia, Microsoft, and Amazon. SpaceX had hoped to fasttrack its way into the S&P 500 immediately upon listing. This was essentially changing
the rules. They were essentially forcing the general public to buy the stock. Fasttracking its way into the index would have forced every 401k and even pension funds to buy the stock. This isn't just American citizens we're talking about, but investment funds of everyday people all around the world. roughly 14 billion of automatic passive fund inflows going straight into SpaceX. It would have been a massive wealth transfer, but thankfully the index committee said no. Why? SpaceX was just too risky. To join the S&P, a company first must be profitable in its most recent four quarters. SpaceX didn't meet the mark. Despite posting 18 billion in revenue in 2025, which was a
33% year-over-year increase, it lost 5 billion for the same year. So the S&P 500 refused to play ball and now that $14 billion in passive funds is off the table. But the NASDAQ, they said yes. So there still will be some passive funding. And because SpaceX will be fasttracked onto the NASDAQ index, major funds will be forced to buy some shares. A lot of Australian investors will be exposed to SpaceX whether they want to be or not. SpaceX spent years dismissing the public markets. Elon Musk had watched what quarterly earnings obsession did to Tesla. For a company that measures
success in orbital mechanics in decades, impressing shareholders in a 90-day cycle seemed like a poor fit. In fiscal year 2023, Spac X's capex was manageable at 42% of revenue. And in 2024, SP X was even profitable, clearing about 791 million in net income for an expense-heavy rocket company. That is remarkable. So, what changed? two letters AI. In February of 2026, Space X merged with XAI. Now, the company's performance was in shambles. After the merge, the required retrospective accounting attributed to that net loss of 5 billion we talked about in 2025. XAI is a company that's bleeding so much money that it needed the public to fund it. Starink profits alone weren't nearly enough.
Career market analysts like Michael McCarthy say investors should be extremely cautious. No matter how we cut the numbers, they simply don't add up. In 2025, XAI was losing $2 for every dollar earned, [clears throat] burning through approximately $28 million per day. Quarterly losses accelerated through the year. By Q1 of 2026, XAI's capital expenditure alone reached $7.7 billion. That's an annualized pace of 30.8 billion, more than double the entire prior year. In that same period, Spac X's capital expenditure had gone from that manageable 42% of revenue to 215% of revenue.
Objectively, Spac X was a great company by itself. There was no need to shove AI into it. Every dollar that XAI burns is a dollar that the rest of the company could have used. So to summarize, in one transaction, a focused aerospace business became a conglomerate of rockets, satellites, social media, and a chatbot in the form of Grock. Up next, we'll see some of the bizarre things happening with XAI, and then we'll take a close look at Space X's biggest bet, data centers in space. But first, this message is sponsored by Raycon. Finding a gift that's actually useful is the hard part. But Raycon's essential open earbuds are one of those things I can genuinely see getting used everyday.
I've been using them quite a bit recently on walks, while running errands, and even working from home. What makes them different is the open ear design. They sit just outside your ear canal, so you can listen to music, podcasts, or calls while still hearing what's going on around you. Whether I'm walking through the neighborhood or just getting outside after working on a script the whole day, I don't feel completely disconnected from everything around me. They're also incredibly lightweight, and the flexible multi-angular ear hook means that they stay secure without becoming uncomfortable. Whether I've worn them on long walks, around the house, or during workouts. After a while, you honestly
forget you're wearing them. A few features that make this a go-to pair of headphones is the open ear design, the comfortable fit, and the multi-point connection. I can switch between my phone and laptop without having to reconnect anything, and the battery life is also excellent, up to 36 hours with the charging case. I charge them maybe once a week. Raycon now has over 3 million customers, and the sound quality is right up there with the far more expensive brands. Plus, they come with a 30-day guarantee, so there's really no risk trying them out. I got one of these for Mother's Day, too, and she started using it as well. So, if you're looking for something practical
this Father's Day, this is a great option. Upgrade your dad's workout routine. Go to byracon.com/coldfusion open to get 15% off. Thanks to Raycon for supporting Coldfusion. Now, let's get back to these data centers in space. Since AI is 85% of the company's proposed market, we're forced to look at how XAI is performing in the LLM landscape. So users switch between Chatg GPT, Claude, Gemini, and Grock consistently. Critics argue that AI is essentially a commodity at this point. Nobody talks about Grock being the market leader except for users on the Xplatform. In fact, according to some estimates, Grock only has 0.4% 4% of enterprise use in the AI sector.
Enterprise is where all the money is. So that's not a good sign. All right. Well, maybe the management team is a star-studded lineup of smart people, so they'll figure it out, right? Well, perhaps not. Musk founded XAI with 11 co-founders in 2023. Today, all 11 have left. There's zero remaining. Okay, so that's not encouraging, but that's still fine. They can still make money by selling compute to other AI companies. Well, there's some problems there, too. The deals so far don't give an aura of confidence. Google entered a partnership with XAI to lease some GPU compute for $920 million a month. Sounds great, but when you look deeper into the deal, there's a strange
smell. Firstly, either party can terminate the contract with 90 days notice. Okay, that's a bit tenuous. But reading further, you learn that Google has a 6% stake in XAI. And there you go, that smell gets stronger. It smells like yet more circular financing with a sprinkle of revenue boosting for the SpaceX IPO season. Okay, so what about Anthropic? Didn't they sign a deal with SpaceX in May of 2026? Yes, but the way that deal came about is hilarious. It turns out that XAI screwed up the design for their first mega data center. For some reason, they built it with three different types of GPUs. This meant that the fastest GPUs were just sitting there idle, waiting for the slower ones to
finish. The result was that AI training only ran at 11% of full capacity according to sources. Seeing that it was useless for training, XAI decided to rent out the data center and move the training of Grock to a new one. So, that's not a great sign in management competence, but okay, that's fine. The company can still make money with a data center in space. SpaceX has filed with the FCC to build a space cloud of up to 1 million satellites. Orbital AI compute is targeted to begin by 2028, but launching a million satellites is a lot. The Starlink network has 9,000 satellites, so imagine over 100 times that number
orbiting the Earth. 2 years ago, I literally moderated a panel on Space Junk, and I was even on the news talking about it. Although SpaceX's satellites have autonomous avoidance systems, coordinating 9,000 objects and coordinating 1 million objects are two completely different challenges. I truly believe that if we don't want space debris becoming a hot topic in 15 to 20 years, sending 1 million satellites into orbit with GPU technology that will be outdated in 3 years just isn't a good idea. Or am I the crazy one here? Maybe there's something I'm missing, but it doesn't look like a financially competitive idea at all. In fact, it seems kind of wasteful. It seems like even in the best case scenario, with every new GPU
release from Nvidia, the orbital data center will become less and less competitive over time. In the worst case, if LLM algorithms get vastly more efficient or AI running locally on devices becomes the norm, that seems to me like a huge threat to the central thesis of a space data center. But I do get the basic idea here. On my last episode about data centers, I got so many comments from you guys saying how horrible it is living next to them. Some of your stories were heartbreaking. So yes, I can see people wanting an alternative, but as for actually putting data centers in space, I'm just not so sure personally. So it remains to be seen how well this works out economically. From an engineering point of view,
leaving aside the cost of launching that much mass, dissipating the heat radiation from the GPUs will be a challenge. In space, there's no air, no water, and no convection. So radiating heat away becomes very inefficient. But this is their area of expertise, so they probably got it figured out. That being said, there are some other things to consider though. They are as follows. Cosmic rays and protons can cause bit flipping in memory resulting in errors in calculations. And that's not great for some of the most densely packed transistor technology in existence. And also, energetic solar particles can
destroy unshielded GPUs. Van Allen belt electrons degrade solar panels over time. And most devastating of all, solar flares can cause massive radiation sparks, resulting in the potential loss of the GPUs. To fix this, each satellite must require heavy radiation shielding and possibly redundant systems. Without it, the data center performance will inevitably degrade over time. All of this at mass and cost. But what do I know? Obviously, I'm not sure what the future holds. So, we'll have to wait and see if this risky long shot plays out. So, if orbital data centers don't end up working in the prospectus, there's still talk of asteroid mining and Mars colonies. So, that's got to be worth a
lot of money, right? or even so, SpaceX only believes that it's a subset of 15% of their market. Remember, 85% is AI. So, when you think about it, the company themselves don't really believe that. It's a bit of a shame that AI is now 85% of SpaceX's addressable market. This is essentially what investors are buying according to their own filings. So, what will happen to the stock? Well, nobody really knows, but the general consensus is that it's going to rocket up as it launches. Then, there will be a slow grind downwards. SpaceX was a genuine world-class aerospace company with a profitable satellite internet business that had no competition. If these were normal days,
they would have IPOed on the basis of steady growth. There was zero competition, so they could have jacked up their prices as much as they wanted. That should have been enough for investors. But everyone caught the AI bug. And now, for some reason, AI has to be 85% of their business. XAI is essentially a Trojan horse because in the prospectus you have to scroll through 11 pages of rockets and talk of reaching the stars before you get to the financial details that reveal that the whole plan is to float an AI company. Investors seem none the wiser here. Last time I checked there were like 200 something. They're probably going to go over to $500 million.
Hundreds of thousands of dollars. 35. Really? What is he selling? He's selling real estate on Mars. The S&P 500 thankfully said no to this, but the NASDAQ said yes. So, look, who knows? I could be completely wrong. A rabbit could be pulled out of a hat and SpaceX could rise to hundred trillion valuation in 5 years. But as for right now, it seems very risky and there's also a bad financial smell. Anyway, I'm done talking about it. I'm just going to get the popcorn. Right. So, thanks for watching. What do you guys think about the SpaceX IPO? Do you think things are going to go great and it's going to make a lot of money or do you think it's just a quick panicked cash grab off
the back of the AI boom? Let me know your thoughts in the comments section below. Feel free to subscribe if you liked it. My name is Dogo. You've been watching Cold Fusion and I'll catch you again soon for another episode. Cheers, guys. Have a good one.
