I Thought Someone Was Propping Up SPCX. No One Was

2026-08-06 original Engineering

For a while now, I’d been watching a certain stock’s order book every day.

It was SpaceX (SPCX), which went public on June 12. IPO price $135, net proceeds from the offering roughly $8.57 billion. One of the largest capital raises in history.

I had a reason to watch it. On the day it went public, I wrote an article about the company’s multiple, asking how a valuation of 245x the Starlink business’s EBITDA could possibly hold up. Having complained that loudly, I had to see how it played out.

After the listing, the stock price fell in almost a straight line, from a June high in the $225 range to the low $100s by early August. Cut in half.

The decline itself was completely expected. I’d voiced my discomfort publicly, so let me be clear here: no surprise there.

What caught my attention was something else.

The discomfort of seeing the same movement every day

For several days, exactly the same pattern repeated.

At the open, the price would drop about 5% from the previous close. When it fell below $110, buying would suddenly kick in, recovering it to $115. Then it would barely move the rest of the day, drifting to a close. The next day, the same thing.

This didn’t look like natural price movement.

In a stock whose price kept falling, buying kicked in every time it hit a particular level, and it never went above a certain point. It looked like someone was “defending this level.”

The hypothesis I came up with was simple: someone who benefits from propping up this stock price might be buying to support it, backed by financial firepower.

There was a basis for this too. This stock’s float was only about 5% of shares outstanding at the time of listing. Relative to daily trading value, a large player would have relatively wide latitude to move the price. A structure where a small amount of money can shape the price does exist.

Let me state the conclusion up front: this hypothesis was wrong.

More precisely, the phenomenon was real, but there was no agent behind it. What follows is the record of my re-investigation.

Doing the things a support operation wouldn’t do

The first thing that fell apart was the theory of official price support.

Stabilization by the underwriting syndicate takes the form of buying back the unexercised portion of the over-allotment option. But in this deal, the greenshoe had been fully exercised within days of the listing. There were no bullets left to buy back with. On top of that, the stabilization period is typically around 30 days, and it had expired by mid-July.

The company’s own share buyback is also a weak motive. Right after raising $8.57 billion, with quarterly capital expenditure running around $18 billion, there’s no logic in a company spending cash to prop up its stock price. It’s also within the blackout period ahead of earnings. If insiders were buying, a Form 4 would appear within two business days, so that could be checked too.

Up to this point, it’s a process of elimination. The real problem came next.

If there were a support operation, the price wouldn’t move like this.

The level that was supposedly being defended was, in fact, stepping down in stages: $110.85, $109.53, $108.66, then $104.83. A step down every day.

If a deliberate agent were defending the price, it should stick to a level. A defense line whose defended price falls every day isn’t a defense.

And another thing: this buyer was selling at $115.

The suppression of the rebound was part of the same movement. Buy low, sell high. A support operation wouldn’t sell at the top.

The phenomenon was real, but it wasn’t “a force creating a floor” — it was “a force compressing the trading range.” That’s the core of where the hypothesis was wrong.

The true identity was directionless inventory adjustment

It clicked into place when I looked at the options market side. Let me write this without jargon.

There are three players: the person buying the option, the dealer selling it, and the stock market.

The dealer’s business is to earn a fee-like profit from selling options. It’s not a business of betting on whether the stock goes up or down. That’s the starting point.

A dealer who, say, sold “the right to buy at $115” is in trouble if the stock rises, because the obligation to deliver becomes real. So the dealer buys the stock in advance. But not all of it — not while the stock is still at $110, because the probability of that right being exercised is low.

So the dealer holds only the amount scaled by probability: 30% if the odds are 30%, 50% if 50%, 70% if 70%.

As a result, the dealer buys more every time the price rises, and sells every time it falls.

The relationship between dealer share count and stock price: an S-curve where the higher the price, the more shares the dealer must hold

Shares the dealer must hold (out of 100)

30 shares 50 shares 70 shares 95 shares The higher the price, the more the dealer buys

$110 $115 $118 $125 Stock price

Figure 1: Relationship between shares held by the dealer and the stock price

There’s a flip side too. A dealer who took the other side of “the right to sell at $110” benefits if the stock falls. To stay neutral, the dealer buys stock to offset that gain. The lower it goes, the more is bought. When it recovers, the dealer sells it off.

Buy on the way down, sell on the way back up.

This was exactly the pattern I had been watching for several days.

The dealer wasn’t turning bullish or bearish. It was mechanically adjusting an inventory meant to honor a commitment, in response to price. From the market’s perspective it still looks like buy orders coming in, but there’s no intent behind it.

Supporting at $110 and capping at $115 on the same day isn’t a contradiction either. The $110 option and the $115 option are different products, and which side of the trade the dealer is on differs between them. That’s why the behavior changes by price range.

The sign flips depending on the price range: below $110, buying increases as the price falls; above $115, buying increases as the price rises

Buying increases as price rises Hedging the call side. The rise accelerates $115

Buying and selling roughly offset The intraday range compresses $110

Buying increases as price falls Hedging the put side. The decline is cushioned

Figure 2: Even within the same day, the sign changes depending on the price range

This mechanism isn’t loyal to a price level. It follows wherever the outstanding option positions happen to be concentrated. That’s why the defense line drops every day.

To be clear, I haven’t confirmed the actual open interest figures for this stock. This is simply the most straightforward explanation consistent with the order book’s behavior, nothing more can be claimed.

My sense that “someone is propping up the price” was right as a phenomenon. Only the part where I imagined the agent as a human being was wrong.

Earnings was the point of verification

Explanations of this kind can be fitted to anything after the fact, so they need to be put into a testable form.

August 4 brought the first earnings report since the listing, and two business days later a large lock-up expiration was coming due: 911.5 million shares, 1.4 times the roughly 639 million shares sold in the IPO.

The prediction I made was this:

Even if the numbers are good, the stock won’t recover, because what’s driving it is not earnings but the supply schedule
If the company touched the supply side (say, a voluntary extension of the lock-up), that would change things, but it won’t, because doing so would force the early investors and employees who’ve already earned the right to sell to give it up for the benefit of outside shareholders
Therefore, with some ups mixed in, it will ultimately fall

The earnings numbers were a substantial beat.

Revenue of $7.81 billion (versus an estimate of roughly $6.8–6.9 billion). Net loss of $541 million, a loss of 9 cents per share, far better than the expected loss of 26 cents. Operating loss narrowed from $970 million in the same quarter a year earlier to $143 million. All three business segments beat consensus. Starlink subscribers doubled year over year to 12 million. Full-year guidance was raised for the first time in the company’s 24-year history.

The stock fell 8% after hours.

It had simply returned to the level from the day before earnings.

What an “AI company” turned out to be was an equipment-leasing business

Why does it fall on a beat? Here another structure comes into view.

This IPO was sold as a “rocket company,” but looking at the revenue breakdown: telecom 55%, AI 33%, space 12%. Rockets are the smallest piece.

The AI segment that’s drawing attention had revenue up 247% year over year to $2.56 billion. The headline is flashy.

Reading the breakdown, of the $1.82 billion increase, $1.6 billion was data center rental. It wasn’t Grok usage growing, and it wasn’t AI products selling. It was revenue from renting out GPU-equipped facilities to other companies. Legacy Twitter’s advertising revenue, sitting in the same segment, was $367 million, down 14% year over year.

The contract structure matters even more. One major customer is leasing a data center in Memphis under a fixed monthly fee, three-year, exclusive arrangement. Fixed means it won’t grow from there. Revenue ramps up in a step at the start of the contract due to timing recognition, then flattens out.

And because it’s exclusive, the same facility can’t be sold to another company. Winning a new customer requires building new facilities.

Lining up the numbers makes the nature of it clear. Total cloud contracts won this quarter were $14.1 billion. AI-related capital expenditure in the same quarter was $15.8 billion. The capital deployed exceeds the total value of contracts won. And the contract value is recognized over three years.

Comparison of capital expenditure and contract value: annualized AI capex of $63.2 billion versus roughly $4.7 billion in annual revenue expected from contracts

AI capital expenditure (annualized)

$63.2 billion Annual revenue from contracts

$4.7 billion Rough estimate of $14.1 billion in total contracts recognized over 3 years

Figure 3: Capital deployed versus the annual revenue expected from it

There’s no way to grow revenue other than piling on more equipment. This isn’t software economics, it’s real estate leasing economics.

The multiple the market pays for AI companies assumes software-like gross margins and economies of scale. Against the reality of $18.4 billion in quarterly capital expenditure, I suspect what the market reacted to wasn’t the size of the figure itself, but this ratio.

Nobody can really say for certain whether it’s overvalued. Since it’s unprofitable, P/E can’t be used, and the answer changes by an order of magnitude depending on whether you apply a telecom multiple, an AI infrastructure multiple, or a data-center real estate multiple. That’s why the price targets from 34 analysts range from $62 to $800.

Uncertainty itself can be an investment opportunity. If you can buy cheaply something nobody can price, that’s an edge. What’s harsh here is that this sits at maximum uncertainty while also carrying one of the highest multiples. You could take the same bet on a target with a clearer business structure at a much cheaper price.

That said, the telecom business is the real thing. Quarterly operating profit of $1.7 billion, up 79% year over year. The barriers to entry are physical too. If it were a standalone company, I think it would be worth considering for a long-term hold. The problem is that the value of the telecom business alone can’t explain the current market cap.

At the end of my previous article, I wrote that if there were a stock that let you buy just Starlink, I’d buy it. Two months of earnings later, that wish has only been reinforced. Telecom is strong, and it’s the structure wrapped around it that’s setting the price.

Closing

Starting from “someone is propping up the price,” I landed on “a mechanism with no intent was trading in reaction to price.”

This was interesting. It felt like searching for a culprit and finding, instead of a culprit, something like a law of physics.

Similar mechanisms seem to exist all over the market. Flows tied to index rebalancing, risk-parity rebalancing, forced liquidations triggered by collateral values. None of these involve a person making a judgment, yet trades occur as if by identity, in response to price.

A mechanism is easier to read than intent. Once you know what it’s tracking, you can calculate its next move. Once you reduce something to “someone’s intention,” you stop thinking there.

When you see unnatural regularity, look for a constraint first, not an agent. That was the biggest lesson this time. For someone who used to work on circuits, this should actually have been a familiar way of thinking.

That said, the mechanism itself has a ceiling. Trading volume originating from options has a cap, and that pressure itself disappears at expiration when the open interest vanishes. In a scenario where tens of billions of dollars in supply arrive, it becomes a rounding error.

The answer to that hasn’t come in yet, as of this writing.

References

※This article is not a recommendation to buy or sell any particular stock. The stock price and order book figures are as of the time of observation and may differ by source.

This piece was drafted and directed by Kuzuryu, with the writing done by AI.

Originally published in Japanese at https://clazytech.com/2026/08/1743/. Translated with LLM assistance and reviewed before publication.