Clay Tech

"clay-works make things real"

translated from clazytech.com

A 245x Valuation Only Makes Sense If You Understand America's IPO Logic

original Startups

Starlink’s adjusted EBITDA runs at roughly $7.2 billion a year. The IPO’s assumed valuation is $1.77 trillion. Divide one by the other and you get roughly 245x.

For a normal telecom or infrastructure business, 20-30x would be reasonable. Here the valuation comes in an order of magnitude higher, on track to become the largest IPO in history. Even conservative estimates, like Morningstar’s, put fair value at $780 billion — less than half the IPO target.

In America, this goes through anyway.

What caught my attention was this: place the question “why does this pricing work in America” next to Japan and China, and a difference emerges in the underlying design philosophy of what “going public” even means.

I’m not a professional, just someone who’s had investing as a hobby since my student days, but I found the angle interesting enough to write about.

Going public is “a device that converts expectation into cash, now”

Let me abstract this a bit.

An IPO is, in essence, a device that converts future expectations not yet realized into cash at this present moment. In engineering terms, it’s a transducer that turns the input of expectation into the output of capital.

The problem is that the gain of this device, and who controls that gain, differs completely from country to country.

Same “expectation → cash” conversion device, but who reviews the story and when cashing out is permitted differ entirely. Here’s each one in turn.

America: the market reviews the story (and often gets drunk on it)

The American device runs at extremely high gain.

245x is tolerable because the market is buying “the story” rather than “the fundamentals.” In SpaceX’s case, the February 2026 merger with xAI produced the ultimate story: space × AI. The price tag isn’t on Starlink’s current cash flow — it’s on the dream beyond it, AI and Mars.

Something I felt firsthand running a company in the Bay Area was that over there, the story itself is the product being financed. In Japan, “so how much profit are you making?” comes first. In America, “so what future are you building?” comes first. The order is reversed.

There’s also something easy to overlook: the structure of founder control. Even after this IPO, Musk retains over 82% of voting rights. Through the dual-class share structure, he raises enormous capital by selling shares without ever giving up control.

In other words, the market “can buy the story, but can’t buy the company.” That’s the American way.

The timing of cashing out was also entirely up to the founder. SpaceX stayed private for 24 years, holding out until it emerged, fully primed, as the largest IPO in history. A high-gain device, held in reserve until the moment it would hit hardest.

A high-gain device naturally overshoots. That’s a bubble. And yet the fact that it keeps spinning regardless is both the astonishing strength and the terrifying part of the American market.

Japan: there’s simply not much of a story to review

The Japanese device runs at low gain.

So low that regulators’ energy isn’t directed at “cooling down dreams” but at “scolding companies for being too cheap.” This is the exact opposite of America, and I find it the most interesting part.

The symbol of this is the Tokyo Stock Exchange’s request for “management conscious of cost of capital and stock price” — the so-called sub-1x PBR problem. It began in March 2023, and 2026 marks exactly three years since.

A PBR below 1x means, bluntly, that the share price is cheaper than the assets the company holds. The market is effectively saying “you’d be better off dissolving the company.”

Lining up the numbers, it’s brutal.

Three years after the request, the figure has fallen only 6 points from 50% in July 2022. Disclosure rates have climbed to 92%, but the substance of improvement is only halfway there. That’s why, entering 2026, the Tokyo Stock Exchange has started publishing a side-by-side “capital cost disclosure list” comparing each company’s efforts — pressure that amounts almost to calling companies out by name.

America worries “isn’t the dream priced too high.” Japan worries “why can’t we even get our own assets reflected in the share price.” The direction of the problem is reversed from the start.

The philosophy around cashing-out timing is equally contrasting. In Japan, the common pattern is listing small and early on the Growth Market, then stalling out. Rather than holding a high-gain device in reserve, Japan fast-forwards a low-gain device, and the signal ends up buried in noise.

China: the state screens the story

In China’s device, the state holds the gain control knob.

The stage is the STAR Market (科創板). Xi Jinping announced its establishment in 2018, launching it as a pilot for a registration-based system — rare in China. Today 592 companies are listed, with “hard tech” as its core.

What matters here is that despite being called a “registration system,” whether a listing is approved is effectively determined by whether it aligns with the state’s industrial strategy.

There’s a concrete example. From around the end of 2023, the CSRC (China Securities Regulatory Commission) effectively froze IPOs for loss-making companies (the Fifth Listing Standard). The stated reason was to protect market balance, but in effect the tap was shut off. As a result, funds raised in 2024 collapsed sharply from the year before.

Then in June 2025, at the Lujiazui Forum, CSRC Chairman Wu Qing announced the reopening of that Fifth Standard for “quality” loss-making hard-tech companies in fields like AI and robotics. IPOs in 2025 nearly doubled overnight, and GPU companies aiming to rival Nvidia — Moore Threads and MetaX — listed one after another.

In other words, the state decides when cashing out is permitted through a policy switch. If the story aligns with national goals like semiconductor self-sufficiency or “new quality productive forces,” the tap opens. If it doesn’t, the tap closes.

An even more blatant example is the 2020 case of Ant Group’s $35 billion IPO, halted at the very last moment. The state can kill a story overnight.

Not the market, not the founder — the state is the story’s examiner. That’s the Chinese way.

Why does Japan turn conservative while America turns bullish? — min vs. argmax

So far this has been about who reviews the story. I want to go one level deeper and lay out, as a structural matter, why Japanese pricing turns conservative while American pricing turns bullish.

Japanese pricing isn’t the “average” of stakeholders’ wishes. In practice it’s a min operation.

Going public involves a long line of parties holding something close to veto power: the lead underwriter, existing shareholders, business partners, the main bank, the auditing firm. The price converges toward the minimum value that all of them can tolerate. The single most conservative link determines the overall value — just as when a signal passes through a series of filters connected in series, the narrowest filter determines the bandwidth.

Brand, face-saving, external relationships — these effectively function as constraints that raise each participant’s floor of tolerance. The more constraints you multiply together, the more the solution gets pushed into a corner.

America is the opposite. Price isn’t decided by consensus across the whole market, but by the “marginal optimist” who fills the order book during bookbuilding. There’s no need to convince everyone. As long as a sufficient number of believers buy at the marginal price, that clears it.

In other words, Japan is min (the most cautious party decides), while America is closer to argmax (the party that believes most strongly bids the price up). Same “review of the story,” but one is a committee’s unanimous consent, the other an auction’s highest bid — entirely different things.

The idea that “it’s a contest of whether you can get people to believe the dream” is half right, and I want to correct the other half. What’s happening is less fraud and more a credit auction with ex-post settlement.

America’s approach: get people to believe in the dream for now and sell high — then if it misses, lawsuits, short sellers, and the SEC come swinging afterward. Discipline is applied after the fact. Japan is the reverse: since all stakeholders finish their price assessment beforehand, by the time the company lists, all the conservatism has already been priced in. Discipline is applied beforehand. You could say it’s the difference between placing the control loop on the input side versus the output side.

There’s a very direct piece of evidence backing this view: the long-standing criticism that Japanese IPOs are structurally underpriced, with first-day prices popping too high.

The lead underwriter’s real incentive isn’t to “maximize the amount raised for the issuer” but to “not let the listing fail, and not damage relationships with existing clients.” So they err on the safe side. Face-saving and relationships literally push the price downward. When the Japan Securities Dealers Association revised its process for setting offering prices in 2022-23, that was precisely a response to this “too conservative” problem. The multiplicative structure was real enough to demand institutional reform.

That said, let me offer a counterargument too. This explanation is strong on the supply (issuer) side, but there’s a hole on the demand side as well.

Suppose you removed stakeholders and let issuers price boldly and bullishly — is there actually a thick enough pool of “marginal optimists” in Japan willing to buy that dream at a high price? Funds with mandates to bet on growth, and individual investors literate enough to bet on dreams, aren’t as thick a layer as in America. In other words, Japan’s conservatism isn’t a one-sided problem; it’s driven from both sides — the seller’s min and the buyer-side thinness. Even if you strip away the stakeholders, if there’s no aggressive buyer on the other side, the contest never even begins.

To summarize, market character splits along three axes.

Lined up along these three axes, China falls into a somewhat unusual quadrant: “the state pre-censors argmax.” Argmax is permitted only for stories that align with state strategy.

Which device do you want to live inside?

Let me lay out everything above once more, plainly.

Here’s where the paradox emerges.

Even America, which looks the freest, is a place where “the market can’t buy the company” in the sense of founder control. Even China, which looks the least free, allows the fastest cash-out imaginable, so long as it aligns with state intent.

Freedom and control aren’t as cleanly opposed as you’d think.

Japan might be neither free nor controlled — just structurally bad at telling stories. It has the assets. It has the technology. And yet the gain of the device that converts these into expectation remains structurally low.

The unease I felt at SpaceX’s 245x probably wasn’t simply “this is too high.” I think it was unease at the gap itself — the realization of just how differently the same device is designed from country to country.

For what it’s worth, I’m the suspicious type, so I won’t be riding the 245x dream. But if there were a stock that let you buy only Starlink, I’d buy it. Unfortunately, no such device exists anywhere in the world yet.

This piece was conceived and directed by Kuzuryu, drafted by AI.

Originally published in Japanese at https://clazytech.com/2026/06/1630/. Translated with LLM assistance and reviewed before publication.