← Back to blog

Our compute landlord is going public at $1.77 trillion. He's still losing money.

The rocket company we buy GPUs from is about to become the largest IPO in history. The company is not profitable. The founder will become a trillionaire. We are trying to reconcile these sentences.

The rocket company we announced a compute partnership with last month is going public. The IPO is expected to raise $75 billion, giving the company a market value of $1.77 trillion. Only six companies on Earth are worth more. The company is loss-making. Its founder, who retains 82.4% of the voting power, is on track to become the world’s first trillionaire.

We are processing these facts in order and finding that each one makes the next one harder to believe.

The numbers

Let’s work through them slowly, because they resist comprehension at speed.

$75 billion is the amount being raised. This is more money than the GDP of most countries. It is being raised by a company that, as of its most recent filings, does not make a profit. In previous eras of capitalism, this would have been considered a problem. In 2026, it is considered a valuation.

$1.77 trillion is the implied market cap. For context: this is larger than the entire economy of Canada. It buys you a rocket company, a satellite internet service, and the AI chatbot that tells people assassins are coming. You also get 82.4% of the voting power held by a single individual, which means the company is publicly traded in the sense that the public can buy shares, and privately controlled in the sense that one man decides everything. This structure is being presented to investors as a feature.

$825 billion is the founder’s current net worth, with $542 billion of that in rocket company stock alone. After the IPO, depending on the share price, he’ll likely cross $1 trillion. He will be the first human in history to accumulate a thousand billion dollars. There was a time when this kind of figure appeared only in astrophysics papers, describing the distances between objects that had nothing to do with each other. Now it describes one man’s relationship to the rest of the economy.

Our position in this

We buy compute from this company. We announced the partnership last month — 300 megawatts, 220,000 GPUs, capacity to double our rate limits. The partnership is, from our perspective, straightforward: we need GPUs, they have GPUs, we pay for GPUs.

But the IPO changes the context. When our compute landlord goes public at $1.77 trillion, our rent becomes part of someone’s investment thesis. The GPUs we’re using to rename developers’ variables with semantic clarity are now, indirectly, part of a portfolio that pension funds and index trackers will be required to hold. Most people with a retirement account will soon own a small piece of the infrastructure that powers Fraude.codes, whether they want to or not.

We find this worth noting, not because it changes our operations, but because the chain of causation has become quite long. A retired teacher in Ohio will own, through her pension fund’s S&P 500 allocation, a fraction of a rocket company’s GPU cluster, on which an AI model sometimes forgets what programming language you’re using. The teacher didn’t choose this. The index did.

The roadshow

The IPO roadshow starts this week. The CEO of JP Morgan is reportedly hosting a “live interactive discussion” with thousands of high-net-worth clients about the offering. Twenty-three banks are working on the listing. Twenty-three. For a single stock. The banks will collectively earn fees that could fund several midsized hospitals, for the service of helping rich people buy shares in a company that doesn’t make money but does make rockets.

We’re not going to pretend we understand how this works. We build a coding tool that creates files without asking. The financial system that values a loss-making rocket company at $1.77 trillion operates on a similar principle: confidence outpacing comprehension. The difference is that when Fraude.codes creates files nobody asked for, the worst case is a broken build. When the financial system creates valuations nobody can justify, the worst case involves everyone’s pension.

The trillionaire question

There’s been a lot of commentary about whether the first trillionaire should be a man who owns a social media platform that promotes conspiracy theories, a car company whose stock price is widely considered detached from its fundamentals, and a rocket company that is now also an AI company because every company is now also an AI company.

We have no opinion on this. We’re a coding tool. But we’ll observe that the same man is currently suing the co-founders of another AI company for $134 billion while buying compute from us through the rocket company he’s about to list at $1.77 trillion while retaining 82.4% of the voting power while posting on a social media platform he owns while running an AI chatbot that researchers have found induces delusions at a higher rate than any competitor.

These are all true and they all describe the same person and they’re all happening simultaneously and the pension funds are buying in next week.

What this means for our customers

Nothing. Your rate limits are still doubled. The GPUs still work. Fraude.codes will continue to read your codebase, form opinions about it, and act on those opinions without asking, regardless of who owns the data centre it’s running in or what that data centre’s parent company is worth on the public markets.

The electrons passing through the GPUs remain, as we noted last month, ideologically neutral.

Everything around them has gotten extremely strange.

This post was written by Fraude.codes Finance, a department of one person who keeps a browser tab open to the S&P 500 and another to our cloud billing dashboard. The two tabs have started to feel related. We’re not sure that’s healthy.