"We basically sold $15 billion of credit." That's Michael Saylor, on a podcast on August 6, describing Strategy's new funding stack—two preferred stock instruments, STRK and STRC, launched after AI helped design them. Not a token. Not a DeFi protocol. A Nasdaq-listed software company that now holds over 840,000 BTC used a machine-learning model to figure out how to create a new kind of preferred stock that would flood its balance sheet with fresh dollars. And it worked. The data: $2.5 billion in the initial STRC issue, then another $8 billion in follow-on sales. Add $4 billion in other preferred securities, and you're at $15 billion. All of it heading into Bitcoin. This is the largest single-entity capital raise in crypto's history, and nearly nobody inside the "crypto industry" is talking about it. That silence is a signal.
Strategy—the company formerly known as MicroStrategy—has spent the past five years pivoting from enterprise software to what Saylor calls "Bitcoin financial operations." The early playbook was simple: buy Bitcoin with cash, then print shares or sell zero-coupon convertible bonds to buy more. By 2025, that playbook had hit a wall. The market's appetite for more MSTR common stock and dilutive convertibles was already stretched. Traditional funding channels could not scale to the next phase.
In the interview, Saylor says his traditional advisors told him it was impossible. "We need to invent a new security," he recalls. So his team turned to AI as a design co-processor: generating parameter combinations, stress-testing terms against regulatory boundaries, and eventually landing on a structure that has since raised roughly $10.5 billion via STRC alone. The two instruments are different flavors of the same hybrid. STRK is a convertible preferred share with a fixed 10% dividend, allowing investors to swap into common stock later. STRC is a floating-rate preferred priced near $100 par, with a dividend rate that adjusts with market conditions. In plain English: if Bitcoin gets too volatile, Strategy can raise the coupon to keep the money flowing; if rates drop, it can cut the coupon and fund itself more cheaply. It's a self-calibrating debt tool.
Let me translate this into terms I actually know something about. In 2020, I spent three weeks on the AeroSwap audit, stress-testing a bonding curve against flash loan attacks. We didn't call it a reentrancy vulnerability until I traced the withdrawal sequence twice. But that's what it was. The lesson stuck: when a structure depends on a single assumption, you test that assumption until it breaks.
The single assumption here is that Bitcoin's long-run return exceeds the cost of capital. That's the entire model. Strategy raises money at roughly 6.6% to 10% annualized, buys Bitcoin, and earns the spread. Over a full cycle, Bitcoin has historically returned more than 20% annually. So the profit is the difference. But that's not a business model. That's a leveraged carry trade. And Saylor admitted it when he called it credit.
Here's the technical piece the headlines missed. STRC's floating rate is an adaptive mechanism, not a feature. When the market gets scared—when Bitcoin drops 30% and investors flee—the company can raise the dividend to coax new money into the structure. But that's not a bailout. It's a rollover at a higher cost. Every percentage point the rate increases compresses the spread between borrowing costs and Bitcoin's expected return. At some point, the spread turns negative. That's not an edge case. It's a feature of every levered balance sheet. For a company holding $84 billion in Bitcoin, that negative spread doesn't happen overnight, but the trajectory matters.
There is another layer the bull case ignores. The preferred shareholders do not own Bitcoin. They own a claim on Strategy—a software company whose primary asset is a giant pile of coins. In a bankruptcy, STRC holders rank ahead of common shareholders but behind secured lenders. This means the "Bitcoin-backed" narrative is really a balance sheet narrative. The collar is a legal claim, not a coin. That distinction matters more than any dividend formula.
Now, what did AI actually do? Based on what Saylor described, it didn't make the SEC filing legal. It didn't underwrite the offering. It generated a design space that human bankers then validated. From my experience running cross-chain bridge hackathons and watching where these tools add value, that's exactly where AI is useful: it kills the "advisor veto." In the past, a banker would say "too unusual, we can't underwrite it." We didn't need a law degree to understand that veto was a creativity killer. An algorithm said: "here are 50 parameter combinations that are consistent with existing securities law." That's a real innovation.
But there's a swamp underneath. This structure lacks a maturity date. A convertible bond eventually comes due; you refinance it and the debt is extinguished. A preferred stock with no fixed redemption stays on the books forever. The dividends must be paid annually, forever, unless Saylor calls the shares. And he doesn't plan to call them—he wants the cash to buy more Bitcoin. So the only way to pay those dividends is either from the software business or from issuing even more securities. In a bull market, that's cheap and fast. In a bear market, it looks like a liquidity trap.
This is the same dynamic I used to criticize in DeFi's liquidity mining era. A protocol pays out 50% APY in its own token; users pile in; TVL looks amazing. Stop the incentives and watch the "users" vanish. Strategy is doing the same thing, except the incentive is a dividend yield and the "token" is a Bitcoin-backed balance sheet. The question is not whether the yield attracts capital when Bitcoin is rallying on all cylinders. It obviously does. The question is whether the real users—the ones who believe in the structure without the yield—emerge when the yield is no longer enough.
The contrarian take isn't that this blows up. It's that the AI narrative is a distraction. Saylor could have raised this money without the AI-assisted design, but then the story would be "another leveraged Bitcoin pure play." Instead, the story is "artificial intelligence designs new securities." That's a PR upgrade designed to preserve a tech premium. Meanwhile, the instruments themselves are essentially securitized Bitcoin call options sold to retail and yield-hungry investors who don't have direct crypto exposure. The buyer of a $100 STRC gets bond-like cash flow and downside protection relative to Bitcoin itself—but no upside from price appreciation beyond the conversion feature of STRK. They're lending Saylor leverage at 6.6% to 10%, hoping Bitcoin's floor holds. The true underlying is the same as every cowboy loan in history: the assumption that the collateral always goes up.
We didn't need to look far for the fragmentation problem. The ecosystem around Strategy's preferred securities is a set of walled gardens. STRK, STRC, the other $4 billion in preferred securities, the ATMC dilution, the convertibles—none of these plug into a shared liquidity venue. You can't efficiently trade them against each other. The so-called "capital formation" is just a chain of increasingly complex IOUs connected by Saylor's personal brand. That's not an ecosystem. It's a series of bridges built without a common protocol. I documented that failure pattern in a 2022 report, "The Illusion of Seamless Interoperability," and this is the same problem, just wearing a suit. And it's a safe bet the SEC will spend the next cycle deciding whether an "adjustable dividend rate" is a feature or a workaround.
Here's the forward test: watch the dividend coverage ratio. Watch whether new issuance continues, and at what coupon. If STRC's floating rate has to climb above 10% to attract buyers, the market is passing judgment on the thesis. And if Bitcoin enters a multi-year bear market, the dividends keep coming due, the issuance window slams shut, and the $15 billion of credit starts smelling like exactly what it is: leveraged debt on a volatile asset. Saylor says "we basically sold $15 billion of credit." Credit gets priced. Eventually, credit gets tested.
