Article — "SpaceX targets a $1.75 trillion IPO, unleashing a global hunt for proxy plays": A retail investor in Anhui province named Hu Xiaobin did something that captures the entire spectacle of this IPO in miniature.
Locked out of a $1.75 trillion offering he couldn't legally touch, he reverse-engineered SpaceX's supply chain by hand. He bought Sunway Communication, which supplies Starlink ground terminals, and Western Superconducting Technologies, which makes specialty metals for rockets.
"I dug out adjacent companies in A-shares by myself and made quite a bit of money," he said, calling it "value speculation." Then, a Musk true believer, he sold both before the IPO even priced.
Hold that phrase — value speculation — because it is the whole story. It is an oxymoron doing honest work.
Hold that phrase — value speculation — because it is the whole story. It is an oxymoron doing honest work.
Last week, the "Public SpaceX" thread was about secondary-share trading and pre-IPO stakes, the murky private market where SpaceX exposure changed hands at guesswork prices. This week it has a hard number and a hard date: a $1.75 trillion valuation, pricing June 11, trading on Nasdaq June 12.
And the reaction to that number is the tell.
When a deal is genuinely priced on fundamentals, institutions crowd the book and retail scrambles for scraps. Here, the polarity is inverted. Retail is scrambling worldwide while the professionals hang back.
"It's a great story if you're a trader, and you'd want to bring it up," said Nicholas Smith, Japan strategist at CLSA in Tokyo. "But I doubt people would be making big bets on this."
That shrug is the most important sentence in the coverage. The people who model cash flows for a living are treating a $1.75 trillion company as conversational, not investable.
Start with the mechanics, because they force the strange behavior.
This is an all-primary offering. Every one of the at least $75 billion in shares is newly issued, and every dollar goes to SpaceX itself. Existing shareholders sell nothing and cash out nothing.
So the deal is not employees and early backers looking for liquidity. It is the company raising war-chest capital and inviting the public to fund it.
Layer on the access restrictions. SpaceX is weighing an unusually large retail allocation — as much as 30% — while its underwriters, per Bloomberg's reporting, barred investors in mainland China and Hong Kong from participating. SpaceX's own IPO marketing documents weren't even reachable from those markets.
The result is a demand imbalance with nowhere to go. Retail is invited in size, then a huge slice of global retail is fenced out — so that fenced-out money goes hunting for a side door.
The side door is the supply chain. In Asia, Taiwan is the gateway of choice, home to component makers Chin-Poon Industrial, Wistron NeWeb and Universal Microwave Technology, all of which say they supply Musk's rocket company.
Central Asset Management's Jeffrey Chan pointed to more proxies — Taiwan's Compeq, Tong Hsing Electronic and Kinpo, Japan's Meiko Electronics. These are printed-circuit-board and microwave-component shops now trading as levered bets on a rocket IPO they have no direct stake in.
The optics feed the fever. Lens Technology, an Apple and Tesla supplier, is up nearly 50% to record highs this year after flagging commercial space as a growth driver — helped along by optics of its chairman being seated between Tim Cook and Musk at a Beijing banquet.
That is the mechanism laid bare: a banquet seating chart becomes a valuation input. This is not analysis. It is proximity trading.
The ETF side is louder still. The Tema Space Innovators ETF, which holds a 6.49% private exposure to SpaceX's pre-IPO shares, is up 29% since March. The Tradr 2x Fly Long Daily ETF offers twice-levered exposure to Firefly Aerospace — an unprofitable company that reportedly performed the first successful commercial Moon landing last year.
When the cleanest way to express your conviction is a doubly-levered wrapper around an unprofitable firm, you are not investing in a business. You are buying volatility and calling it belief.
Now the vertiginous part — the gap between what is being valued and what is being sold.
What is being valued: Starlink. Per Reuters, it is SpaceX's only profitable segment, while the other businesses are burning cash. Starlink's revenue is reportedly the single real anchor under $1.75 trillion.
What is being sold is something else entirely — an orbital-compute story stitched to the hottest narrative on Earth. Musk wants to put AI data centers in space, powered by Nvidia's next-generation Vera Rubin chips, launching the first satellites in Q4 2027 and reaching "significant scale" by 2028.
The pitch is seductive because it routes around the AI industry's actual bottlenecks: power grids, land, water. Bypass Earth's constraints; run inference in orbit.
The relationship is real. On Nvidia's earnings call, CFO Colette Kress named SpaceX alongside Oracle and AWS as a Vera Rubin customer, and Nvidia disclosed holding 122.8 million SpaceX shares — supplier and shareholder at once.
But here is the number that should govern the valuation. Evercore, which rates SpaceX "Outperform," includes zero orbital-compute revenue for 2026 and 2027.
Zero. The firm doesn't expect the first gigawatt of orbital AI capacity online until fiscal 2029. Substantial revenue from the story anchoring the enthusiasm is at least three years out, gated by unsolved problems in heat dissipation, radiation, collision, and regulation that the Brookings Institution flags as open questions.
Evercore even admits it doesn't know what Musk means. "The key question is, what exactly does Musk's so-called 'significant scale' in 2028 really mean," its analysts wrote — revenue-generating capacity, or a technical validation phase before commercialization?
So the valuation leans on optionality that its own bullish underwriter won't book for years, on a business that today loses money everywhere except Starlink. The profits are terrestrial and modest relative to the price; the trillion-dollar premium is orbital and hypothetical.
Then there's governance. Dual-class shares concentrate voting power with Musk and insiders — you fund the war chest, he keeps control. For an all-primary raise, that means new money buys the risk and the narrative without a meaningful say in either.
Which brings us to the sharpest framing, from DataTrek's Jessica Rabe. She argues this year's IPO market has actually been disciplined — 68 IPOs so far versus 90 last year, nothing like the froth of 1999 or 2021, with quality rewarded and mature businesses favored.
The winners she cites — Fervo Energy, Cerebras, Forgent Power Solutions — cluster around AI infrastructure and energy, real secular tailwinds. Her thesis: investors will fund growth, but only when it's backed by durable business models and competitive moats.
SpaceX is the stress test of that thesis. If a $1.75 trillion valuation resting this heavily on unbuilt orbital data centers clears cleanly, the discipline was never real.
And that is why Hu Xiaobin's exit matters more than his entry.
He is a self-described Musk fan who did the work, found the proxies, made his money — and got out before the thing he loved actually priced. He treated the IPO not as a company to own but as a wave to surf, and he read the wave well enough to leave before it broke.
Value speculation. The speculation was real. The value was the part he didn't wait around to test.