The Oracle Sold Oracle: Michael Burry, the 13F, and the Quiet Math Behind an AI Exit

There is a particular silence in a quarterly filing. It is not the silence of a blank page. It is the silence of a name that used to be there and is now gone. I spent 2017 tracing the silence that broke the ICO boom, watching tokenomic schedules unravel before the press conference had even ended. Today, I am tracing a quieter one. On 14 November 2025, Scion Asset Management, the fund run by Michael Burry, filed its quarterly 13F with the United States Securities and Exchange Commission. The filing covered the quarter ended 30 September. Two names were missing: Microsoft and Oracle. Not trimmed. Not hedged. Gone. The market did not flinch. Between 30 September and the day the filing went public, Microsoft rose by roughly 2.5 percent. Oracle rose by about 8 percent. In a world that treats every famous investor as a weathervane, the absence of a reaction was almost louder than the absence of the positions. But silence is data. It just has to be read in the right order. Michael Burry is not a market commentator. He is a neurologist who left medicine to run money, then built one of the most famous trades in financial history. In the run-up to the 2008 crash, Scion bought credit default swaps on subprime mortgages while the world was still celebrating house prices. The trade made spectacular returns and made Burry the protagonist of Michael Lewis’s “The Big Short.” He has spent the years since as a professional contrarian, a man who buys what other people refuse to hold and dismisses what other people refuse to question. So when the November 2025 13F arrived, the crypto-native press did what it always does when a famous bear raises his hand: it reached for the word “bubble.” The original report, published by Crypto Briefing, framed Burry’s exits as a warning that AI infrastructure spending is not sustainable. The headline was crisp. The supporting evidence was thin. The report contained exactly two facts: Scion no longer owns Microsoft, and Scion no longer owns Oracle. Everything else was interpretation. This is where the story becomes useful. Microsoft and Oracle are not random positions. Microsoft is the largest commercial partner of OpenAI, the company that made generative AI a household word. Oracle has transformed itself from a legacy database vendor into a cloud infrastructure contractor for some of the most aggressive AI labs in the world. If Burry wanted to express doubt about the AI trade, he could not have chosen two more representative names. Why now? Because the AI capex cycle has reached the point where hope begins to meet accounting. The largest hyperscalers are committing hundreds of billions of dollars to data centres, chips, electricity and cooling. The market has rewarded that spending with rising multiples. The question is no longer whether AI is real. The question is whether the promises embedded in equity prices are real. Let me start with the forensic layer, because after 21 years of reading balance sheets and tokenomic schedules, I have learned that the first shot is almost never the one that kills. The only hard facts in the original article are these. Scion filed a 13F on 14 November 2025 for the quarter ending 30 September. The filing showed no shares of Microsoft. It showed no shares of Oracle. The previous quarter’s filing had shown positions. Therefore, between the beginning of July and the end of September, Scion sold every share. That is it. We do not know the price he received. We do not know whether he sold in July or September. We do not know whether he bought puts to hedge the sale or whether he reallocated the proceeds into cash, energy, healthcare or even Bitcoin. The 13F is a photograph, not a movie. It is also a delayed photograph, taken 45 days before the public was allowed to see it. This is the first cognitive discipline required to read any disclosure: understand what it is not telling you. What the 13F does tell you is that Burry made a commitment. A trim can mean many things. It can mean rebalancing, tax management, or harvesting gains. A full exit is a different animal. In the language of portfolio construction, a full exit means the thesis no longer works at the current price. It does not necessarily mean the company is bad. It means the trade is bad. The asymmetry that once justified the position is gone. Based on my audit experience, I have learned to read quarterly filings the way a coroner reads a body: quietly, without assuming cause of death. And the phrase I watch most carefully is not “buy” or “sell.” It is “the risk-reward has changed.” When a disciplined investor like Burry deletes a position, he is not issuing a weather forecast. He is saying that the payout structure of holding the stock no longer compensates him for the potential loss. That is a different message from “AI will collapse.” It is a message about price, not technology. One of the quiet lessons of the 2008 trade is that Burry is not a market forecaster. He is an asymmetry hunter. He looks for securities where the probability of being right does not need to be high to make money. That was the beauty of the subprime short: he did not need to predict the exact month of the collapse. He needed to occupy a position where, if the mortgage loans failed, the payoff was enormous, and if they did not, the ongoing losses were manageable until they did. The Microsoft and Oracle positions were a different bet. They were not built to survive a long wait. They were long-duration assets whose value depended on the AI buildout producing cash flow faster than the world could spend on it. When a company like Microsoft trades at a multiple that assumes a decade of accelerated earnings, the margin of safety is thin. A single quarter of disappointing cloud growth can reset the narrative. An investor with no structural hedge must therefore be very sure about the next twelve quarters, not the next five minutes. I do not know Burry’s exact cost basis. But the shape of the decision is visible through the lens of financial engineering. If you buy a mega-cap technology stock after a period of dramatic repricing, you are effectively short volatility and long execution. You are paying the market for the privilege of assuming that the management team will spend capital better than any other management team has ever spent it. That is a high bar. Burry’s history suggests he has no interest in clearing high bars. He wants low hurdles and oversized payoffs. The public market conversation about AI is dominated by revenue growth and earnings per share. The private reality is dominated by free cash flow, depreciation lives and the timing of capital expenditures. In my years building financial models for exchange flows and token valuations, I can tell you that the metric most likely to catch investors by surprise is not AI revenue. It is the conversion of AI revenue into free cash flow. Data centres are expensive. They require enormous upfront spending on land, construction, servers, networking and power. Accountants have a great deal of latitude in deciding how long those assets live and when the cost of them hits the income statement. The more aggressive the depreciation schedule, the higher today’s reported earnings and the bigger tomorrow’s headache when those assets need to be replaced. In an environment where capital costs are still meaningful, a company can tell an incredible earnings story while its cash flow story quietly deteriorates. Microsoft and Oracle are both investing as if AI demand is infinite. That may be true in twenty years. It is not true in twenty months. The market has been willing to finance the gap between story and cash because interest rates, while down from their 2023 peak, have not fallen enough to make free capital free again. Burry, a man who built a career reading the gap between what people believe and what payments actually roll in, may have looked at the cash flow statements and decided that the gap was too wide. Why did the market ignore the sale? Because the 13F is a delayed, low-signal document. It is also because famous investors are easier to mock than to follow. By the time the filing was public, Microsoft and Oracle had already risen above their quarter-end prices. The market had moved on. The absence of a price reaction was taken as proof that Burry was wrong. That reasoning has a flaw. Markets do not react to information; they react to the narratives that information supports. A single investor’s exit is easy to absorb into the dominant story called “AI is the new industrial revolution.” It becomes noise. It becomes a “gotcha” tweet. It does not become a reason to question the central assumption that the entire market is organized around until a second, third and fourth investor do the same thing. Catching the signal before the market blinks has never been about the first voice. It is about watching whether the first voice is followed by a second. The next round of 13F filings, which will land after the new year, will tell us more than any single interview with Burry. If other institutional investors quietly reduced their mega-cap tech exposure, the signal will compound. If the opposite happened, Burry’s sale will look like an idiosyncratic preference, and observers can move on. There is a behavioural reading here as well. Mapping the emotional value of digital assets taught me that markets can ignore an ugly fact for a surprisingly long time. The reason is not stupidity. It is social cohesion. Human beings in a market are not price-taking automatons; they are members of tribes. The AI tribe is enormous. It includes software engineers, venture capitalists, politicians and retail investors who have been told that anyone not owning AI stocks will be left behind. An exit by a famous investor threatens the tribe’s identity. The tribal response is to diminish the source. This is why the original Crypto Briefing report is useful even when its evidence is thin. It translates the threat into a familiar frame: the smart money is leaving. For its audience, this is a compelling story. But the story also reveals a bias. Crypto Briefing is a publication that covers digital assets and market volatility. Its audience has a structural interest in the idea that mainstream tech is a bubble and that decentralized alternatives are the next stop. That does not make the AI bubble thesis wrong. It does mean the messenger is not neutral. How we taught the streets to read the blockchain is, at its core, a story about teaching people to ask better questions rather than to memorize price levels. The same lesson applies here. The Bitcoin and crypto markets know what it feels like to be dismissed from the institutional table. They also know what it feels like when the hard questions finally arrive. Burry’s exit is one of those questions, arriving in a form that traditional markets have not yet learned to translate. Here is the unreported angle: Burry’s exit may not be a prediction at all. It may be an admission of a different failure. For several years, Burry has talked about the dangers of passive investing, the distortion of index funds, and the unsustainable concentration of a handful of tech companies in major benchmarks. If the market’s entire structure is distorted, then stock-picking in mega-cap technology is almost pointless. The fundamental investor’s job is to find mispriced securities. In a market where a handful of companies are so large that they move the index no matter what the fundamentals say, the mispricing is no longer individual. It is systemic. A single manager cannot short a systemic delusion without asking to be steamrolled by momentum. So he retreats. Seen this way, the sale of Microsoft and Oracle is not a warning shot at AI. It is a quiet admission that the game he plays best, buying distressed, overlooked assets, has migrated away from these names. Burry is not saying “AI is a fraud.” He is saying “I no longer have an edge in this part of the market.” That is a far more honest and far more powerful statement, and it is invisible to anyone who reads the 13F as a stock tip. The market non-reaction is part of the same story. When an investor sells a stock and the stock does not move, the common conclusion is that the investor was wrong. The alternative is that the investor was too small to matter. Burry’s fund is not a macro force. It is a small, concentrated vehicle. The real money has already moved on. The silence after the sale is not the silence of a bad trade. It is the silence of an entire market that no longer listens to individual humans. This is the invisible contract binding our digital tribes: we agree to keep repeating the story until the market stops paying us to repeat it. Burry broke that contract. He did not announce it on television. He just filed a document and let the silence do the talking. If I were running risk management for a family office today, I would not sell everything because Burry sold Microsoft. I would do something more useful. I would create a dashboard of trailing 13F filings, capital expenditure guidance and free cash flow conversion for the five largest AI beneficiaries. Then I would wait for the first material break. The next Microsoft and Oracle earnings calls are the first checkpoint. The number to watch is not “AI revenue.” It is capital expenditure guidance and free cash flow. If a company raises its capex plan while shrinking its free cash flow guidance, the market will begin to understand the tradeoff. If a company signals that AI revenue is growing faster than capex, the Burry exit becomes a footnote. If a company does the former while another major investor also exits, the footnote becomes a chapter. From tokenized silence to decentralized truth, the lesson remains the same: listen to what is absent. A missing name on a 13F is not a headline. It is a clue. It becomes useful only when it is placed next to other clues and read as part of a larger pattern. For retail investors, the temptation will be to do one of two things. The first is to copy Burry and sell all technology stocks. The second is to dismiss him entirely because the market did not crash the day after the filing. Both are lazy. The better response is to look at your own holdings and ask a genuinely uncomfortable question: what is the margin of safety in the price I am paying? If you cannot answer that question without using the phrase “AI will change everything,” then you are not investing. You are participating in a ritual. The next 45 days will be more valuable than the last 13F. The next earnings calls will tell us whether cash is following narrative. The next 13F filings will tell us whether Burry was a lone voice or the first crack in a consensus. In the meantime, do not let the absence of panic convince you that danger has passed. The market’s calm is not a verdict. It is a mood. Leading the herd through the volatility fog means knowing the difference between a pause and a reversal. Michael Burry just made a choice. The rest of the market has chosen, for now, to look away. The silence is the signal. It always has been.