Starlink looks insanely profitable — turns out that's a lot easier when your sister division launches your satellites for free. If you're just joining, SpaceX isn't an IPO rumor anymore. It priced its offering at 135 a share, and now comes the test every debut gets: can the reported AI compute and data-center revenue justify a roughly 1.77 trillion-dollar valuation, with losses and huge capital needs still sitting there? That's the backdrop for everything we read today. This is AI IPO Watch — today, an accounting mirror trick at SpaceX, Asia's IPO machine hitting a five-year high, and a $20 billion hole nobody's pointing at. Stick around. Here's Rich Smith at The Motley Fool:
In 2025, SpaceX's rocket business lost money, while its Starlink division generated huge profits. Space (rockets) doesn't charge Connectivity (Starlink) for launch services. This inflates near-term profits at Connectivity, while depressing profit at Space.
So the SpaceX repricing argument we've been working all week just changed addresses. We started with AI revenue and xAI hopes; now it's parked squarely on Starlink's segment economics. And here's the mechanism, finally in plain view. Per the Motley Fool, the Space division doesn't charge Connectivity a dime for launch services. Starlink's margin in that prospectus is built on an intercompany transfer that's priced at zero. Right, and I want to be clear — fraud isn't the claim here. The point is the architecture: when you own both the cost input and the profit center, you decide where the dollars land. I've sat in the rooms where you decide exactly that. The rocket business lost money in 2025. Starlink printed huge profits. Move the launch cost back to where it's actually incurred and that gap narrows fast — retail buyers reading the Starlink line have no idea they're looking at a gift between divisions. Which means if you're modeling Starlink as a standalone, you have to impute a market-rate launch cost yourself. That number isn't in any document — and for once this week, the missing line item is literally the story. And it reframes the whole pitch. SpaceX spent a decade as the rocket company that cut launch costs 75%. Turns out that cost advantage is now the subsidy propping up the only profitable segment. Here's The Economic Times:
Technology companies have raised a total of $3.1 billion from stock market listings in China this year to June 18, more than five times the volume in the year-earlier period, according to LSEG data.
Five times last year's volume, $3.1 billion raised onshore, and Reuters counts nearly fifty companies in the Shanghai and Shenzhen pipeline with $18.7 billion in plans. Robotics startups and chip firms are listing at scale while US AI candidates are still in conference rooms trying to manage their boards. And read the regulator language out loud — Beijing supporting listings in 'future industries' to provide exit opportunities for investors. That's LP impatience with a five-year plan stapled to it. Right, but the branding matters. If you're a Palo Alto founder watching Chinese peers clear the exchange while you're stuck, they've got national policy creating window pressure. You've got a skittish board. Same mechanism we'd call VC exit pressure here. Over there it's a self-reliance campaign. The capital still needs to get out the door. Here's The Business Times:
Hong Kong share sales surged to a five-year high in the first half of 2026 as investor enthusiasm around the artificial intelligence boom overpowered the drag of a sluggish equity market and regulatory headwinds.
Forty-four billion out of Hong Kong in six months, up 29 percent — and the Hang Seng's down almost 12 percent on the year. So the index is bleeding while the share sales hit a five-year high. That tells you exactly where the appetite is, and it's not the broad market. Right, but read the line carefully — IPOs, placements, and block trades, all in one bucket. CATL and Victory Giant are driving the headline, and those are big secondary raises, not fresh debuts. When Bloomberg lumps a block trade in with an IPO, the 'AI fever powers share sales' framing gets a lot blurrier. A placement is existing holders cashing out. That's a very different signal than a market clamoring to underwrite new AI capacity. And that's the part that lands for me — back in Palo Alto, US AI candidates are still trying to keep their boards aligned. If you're an AI founder watching Chinese peers list at scale into a falling index, the window pressure is real. No banker had to manufacture it. This one's from AInvest:
SpaceX went public with a market capitalization of roughly $2 trillion. Trailing-twelve-month revenue sits near $19.3 billion, which means the stock trades at 104 times sales. The price-to-book ratio is 48.
Starfall gets the splashdown video, the FAA blessing, the third revenue leg headline. Meanwhile there's a $20 billion hole in the cash-flow bridge and a capsule that, per AInvest, generates exactly zero dollars for the next twelve months. A reentry capsule with no pricing, no commercial service, and no timeline. Calling that a revenue leg is generous; right now it reads like a press release with a payload weight attached. And the timing is the giveaway — the approval lands days before the IPO. I've sat in the room building that calendar. A regulatory headline that convenient in the week of your roadshow doesn't just happen by accident. Down 17 percent in five sessions, sitting at 153, a hair above the 52-week low. The market took the Starfall headline and then opened a spreadsheet. Layer that on the subsidy story we just hit — Space eating Starlink's launch costs. Now the same company's flagging a $20 billion gap. The architecture and the cash need are finally in the same frame, and they're pulling in opposite directions. Which sharpens the lock-up question, Eric. Ninety-six percent of the cap table is frozen, and the company's staring at a multi-billion-dollar funding need before that window even opens. So what comes first — debt, more equity, or both? That's the signal in the expiration. Insiders are waiting to sell, but fresh paper could dilute them before they even get the chance. The lock-up clock and the cash clock are racing each other now. If AI IPO Watch helps you stay ahead of the market, consider subscribing wherever you listen. And if you have a moment, leave a quick review — it really helps other people find the show.
Next, we're watching SpaceX's two FAA-approved Starfall reentry tests — the next technical checkpoint before any commercial-service valuation claims.
We've put links to every story from today's briefing in the show notes, so if one thread caught your ear, you can dig into the original reporting there. That's AI IPO Watch for today. This is a Lantern Podcast.