Crypto
19 Dec 2025
Read 13 min
Michael Burry Substack AI bubble warnings How to spot risks *
Michael Burry Substack AI bubble warnings reveal clear indicators and steps to protect your portfolio.
How to read Michael Burry Substack AI bubble warnings
Why he is speaking now
Burry closed his hedge fund to new outside money. He returned to posting on X in late October after a long break. He now writes detailed notes on Substack, and he joined Michael Lewis for a podcast that he suggests may be his last interview. The tone is direct. He explains his calls, lists the data he tracks, and names the trades he likes and dislikes.Core claim: AI looks like a classic mania
Burry compares the current AI excitement to the dot-com bubble and the mid-2000s housing boom. In his view, stock prices and capital spending have raced ahead of true demand. He expects the AI trade to unwind within two years. He advises anyone sitting on big wins to take some gains and reduce risk.What Burry sees inside the AI trade
Slowing demand versus record spending
Burry notes signs that cloud-computing growth is slowing. At the same time, big tech firms are spending more than ever on data centers and AI chips, especially from Nvidia. That gap worries him. If growth cools while spending soars, returns will fall. Capital can get trapped in hardware that turns old fast or never earns back its cost.Accounting and incentives that stretch the story
He calls out two accounting practices that can make earnings look stronger today: – Companies extend the useful life of servers and chips. This slows depreciation and boosts short-term profit, even if the gear may wear out or become obsolete faster than the books imply. – Heavy stock-based pay reduces true owner returns. It shows up as an expense, but dilution can still hide how much value goes to employees instead of shareholders. He also flags “give-and-take” deals among tech firms. For example, one company may buy compute credits from another while the other buys software or services back. Everyone reports revenue, and the hype stays alive, but real net demand may be lower than the headlines suggest.- Cloud growth slowing while capex hits records
- Depreciation stretched to flatter earnings
- Stock-based pay dilutes owners
- Circular “you buy mine, I buy yours” contracts
- Marketing buzz outpacing proven use cases
His positions and time horizon
Bets against the favorites
Burry says he is short market leaders like Nvidia and Palantir. He calls OpenAI “the Netscape of our time” and claims it is “hemorrhaging cash.” He thinks the AI bubble will burst within two years. That timing matters. It means he expects pressure to show up well before decade-long AI visions become real. He urges investors who rode the rally to lock in gains. If a stock looks priced for perfection, he prefers to take chips off the table. As he puts it, a number of bad years could be ahead for the broad market if profits fail to keep up.Where he still finds value
He has held gold since 2005 as a long-term hedge. Among the mega-cap tech names, he calls Alphabet the value investor’s favorite. That implies he sees steadier cash flows and lower hype relative to certain AI peers. He also shared that he knows Nvidia’s CFO, Colette Kress, and owned Nvidia shares years ago, which shows he is not anti-tech; he is anti-hype.Beyond AI: the macro backdrop
The Fed and the cost of money
Burry argues the Federal Reserve has done “a lot of damage” since its start. He says the Fed does not do “anything very helpful” and suggests the Treasury could manage interest rates and the money supply instead. Whether you agree or not, his point is clear: easy money creates bubbles; tight money pops them. In a tighter world, prices must line up with profits, not dreams.Banks and fault lines in the system
He warns the US banking system shows “fragility,” and says banks are “getting weaker way too fast.” That matters for risk assets. Weak banks lend less, credit gets tight, and growth slows. In that setting, stretched valuations are most exposed. He also defends past calls on regional banks, pandemic-driven inflation, and meme stocks. He admits he missed GameStop’s sudden 2021 surge; he had sold before the spike and says he had “no idea what was coming.” He plans to publish a fresh take on the stock now.- Own portfolio includes Lululemon, Molina Healthcare, Shift4 Payments, Fannie Mae, and Freddie Mac
- Long gold since 2005
- Sees Alphabet as the best value among mega-cap tech
- Predicts tough years for stocks if profits lag
- Bitcoin near $100,000? He calls that “ridiculous” and says it is “not worth anything,” now below $90,000
Practical ways to spot risk now
Start with the business, not the buzz
Ask simple questions:- Are customers paying more or less over time?
- Is revenue growth coming from real adoption, or big one-off deals?
- Does usage growth match the jump in capital spending?
- Are pilots moving to paid contracts with solid margins?
Read the numbers with a skeptic’s eye
You do not need to be an accountant to test the claims:- Depreciation: Did the company extend the life of servers and chips? If so, current profit may be inflated.
- Stock-based pay: How big is it versus revenue? Big dilution can cap future returns.
- Free cash flow: Is the company producing cash after big capex, or only earnings on paper?
- Customer notes: Does the filing mention “commitments,” “credits,” or “barter-like” deals?
Watch for market telltales
Manias have patterns. Burry points to echoes of the late 1990s:- Leaders trade at extreme price-to-sales ratios
- Index gains hinge on a small set of names
- Companies cross-buy from each other to keep growth optics strong
- Journal narratives run ahead of audited results
Build a margin of safety
Burry’s stance suggests a few risk controls:- Trim positions that outran fundamentals; take some profits
- Avoid leverage and complex options unless you can afford a total loss
- Diversify across sectors and factors, not just many tickers in one theme
- Hold some cash or short-term Treasurys to stay flexible
- Consider hedges like gold if they fit your plan
What to make of the message
Signal over noise
The Michael Burry Substack AI bubble warnings are not a blanket call to flee tech or AI. He is saying the math must work. If spending, dilution, and accounting tricks carry results today, the payback must appear tomorrow. If not, prices will adjust. His track record shows he prefers hard data over groupthink, even if he is early.Applying the checklist
Use his lens in your own process:- Map revenue growth to capex and share count
- Stress-test valuation against normalized free cash flow
- Discount “circular” deals and marketing-heavy boosts
- Lower expectations where hardware cycles are short and costly
FAQ
* The information provided on this website is based solely on my personal experience, research and technical knowledge. This content should not be construed as investment advice or a recommendation. Any investment decision must be made on the basis of your own independent judgement.
Contents