Michael Saylor Didn't Make $15 Billion With ChatGPT. He Made a Story About It.

Michael Saylor claims he used ChatGPT to create roughly $15 billion in capital last year. The headline version of that story is seductive: one person, one AI subscription, fifteen billion dollars. It's the kind of sentence that makes you want to sign up for a tool you already have.

But the claim doesn't survive contact with the actual instrument. The security Saylor helped design with ChatGPT — STRK, an 8% perpetual preferred stock — raised a fraction of that amount in its initial January 2025 offering. Not $15 billion. The $15 billion figure comes from bundling the entire capital-raising machine Strategy (formerly MicroStrategy) ran through 2025: five preferred stock IPOs, at-the-market common equity programs, shelf offerings, and existing convertible debt. STRK was one part of that machine, and a relatively small one at that.

You can see why Saylor would collapse the whole operation into a ChatGPT story. It's a much better narrative than "we issued perpetual preferred stock with a monthly-resetting dividend to buy more Bitcoin, then kept printing more preferred stock through an SEC shelf registration." The AI angle turns a leveraged financing play into a parable about individual creativity. It makes institutional finance look like a weekend project.

The real question isn't whether Saylor was dishonest about the numbers — he frames them the way a founder frames any product demo, as the tip of a bigger arc. The real question is whether ChatGPT actually did anything that a room of lawyers and bankers couldn't have done without it.

STRK is a perpetual preferred stock with a fixed 8% dividend, a $100 liquidation preference, and a conversion feature that lets holders swap into Strategy common stock based on MSTR's stock price. The structure is novel in its Bitcoin linkage — the company uses proceeds to buy BTC, and the dividend is funded from operations, Bitcoin appreciation, and further capital raises. But the plumbing itself is standard mezzanine equity. Perpetual convertible preferred stock has existed for decades. The "never been done before" part is the collateral, not the instrument.

Here's where the story gets more interesting. The instrument Saylor built that's closer to genuinely new is STRC — the variable-rate perpetual preferred stock that launched in July 2025 and scaled to an aggregate stated amount of $3.4 billion in nine months, making it one of the largest preferred stocks by market capitalization. STRC had grown significantly and paid regular dividends.

That structure — a Bitcoin-collateralized money fund alternative — is closer to something financial engineers hadn't seen before. But even STRC was built by a team that includes Strategy's CFO, investment banks, and SEC filings spanning hundreds of pages. ChatGPT may have helped Saylor think through the concept. It certainly didn't build the execution apparatus.

Saylor's actual advice is simpler than the headline suggests: Don't try to outwork the robots. The sentence sounds profound until you notice who's saying it. Saylor had an accounting scandal in 1999. He rebuilt by committing his company to Bitcoin in 2020, turning a software business into the world's largest corporate Bitcoin holder — 818,334 BTC as of May 2026, at a cost basis of roughly $76,052 per coin.

What makes Saylor valuable isn't ChatGPT. It's the willingness to bet on a thesis nobody else was pricing, then build a capital structure around it. The AI tool is a drafting aid. The creative work — seeing that perpetual preferred stock could bridge the gap between income investors who wanted yield and Bitcoin believers who wanted exposure — that's a judgment call no model would have volunteered.

But there's a crack in the machine, and it's worth looking at. STRC closed at $89 recently, its lowest level since launch, pushing its effective yield to 12.9%. The preferred stock trades below its $100 par value, which means the market is pricing in a concern: can Strategy actually sustain these dividends, or is it funding today's payouts by issuing tomorrow's securities?

The numbers work as long as Bitcoin keeps appreciating and capital markets keep accepting the product. If BTC stalls and preferred issuance slows, the dividend machine becomes visible for what it is: a refinancing loop where new securities fund old dividends.

Saylor told a reporter that the way to build wealth in the AI era is not to compete with machines but to use them to your advantage. That's true. But the things he's created — STRK, STRC, STRF, STRD, STRE — are all variations on the same theme: perpetual preferred stock backed by Bitcoin. He's not using AI to do something that's never been done before. He's using AI to tell a better story about something that has been done before, repeatedly, and at scale.

The test for anyone inspired by Saylor's ChatGPT claim is simple. Take your $20 a month, ask the AI to design a novel financial instrument, and see whether you can register it with the SEC, convince Barclays to underwrite it, and find investors willing to buy perpetual preferred stock from a company you've never heard of. If you can, then the story is real. If you can't — and you can't — then the AI didn't create the $15 billion. The access to capital markets did. The AI just wrote the pitch deck.