George Noble warns of an AI bubble crash: What it could mean for crypto
“An AI bubble crash could be 17 times worse than the dot-com bust and wipe out far more than the $5 trillion lost from the Nasdaq,” former Fidelity fund manager George Noble warned. That is not background noise for crypto investors. It is the risk. If an AI selloff rattles the wider market, investors may sprint for cash. Bitcoin’s safe haven claim would face a brutal test. Altcoins? They could fall much further.
“Noble’s forecast comes down to the enormous sums being spent on AI infrastructure and the chance that profits never cover the cost.” Polymarket traders put the chance of an AI bubble bursting in 2026 above 17%. The figure recently dropped to 14% after touching 30%. Related contracts now show odds of 16% to 24%. Prediction markets cannot see the future. Still, I would not dismiss that shift: tech stocks are falling, revenue forecasts look less convincing. The weakness has also reached markets around the world.
“Semiconductor stocks are already getting hit.” The Wall Street Journal reported that U.S. stock futures fell as concerns about AI moved through Asian markets. South Korean chipmakers SK Hynix and Samsung Electronics each lost almost 9%. Both plan to spend billions on semiconductor plants and additional AI capacity. Why does this matter? Because investors are no longer simply asking how large AI can become. They are asking whether revenue from AI services will ever pay for all of it.
If it does not, the damage may spread well beyond chip stocks. Crypto would probably follow. Most commentary treats the 2022 collapse as a separate story because inflation fears and rising interest rates were the trigger. That is only half right. Bitcoin dropped from about $69,000 in November 2021 to below $20,000 by mid-2022, and the underlying mechanics were painfully familiar: cash left speculative assets. Fast.
“IBM’s worst one-day stock decline since 1968 may be an early sign that investors are pulling money from established technology companies.” IBM shares fell almost 25% earlier this week, then closed another 2.7% lower at $211.20 on Wednesday. The stock lost 26% across several sessions. IBM said corporate spending on AI infrastructure was eating into software purchases, leaving revenue growth below expectations. The selloff erased tens of billions of dollars from IBM’s market value. Other software and IT stocks went down with it. I’ll be honest: that looks more troubling than a routine bad earnings reaction.
Crypto is exposed because institutional investors do not place every market in its own tidy box. When technology holdings disappoint, the same firms may cut altcoin exposure. Other speculative positions can follow. Nasdaq declines have often coincided with crypto selloffs, and Ether has sometimes behaved more like a tech stock than Bitcoin. My take: betting on that relationship to disappear during a panic is wishful thinking.
“A draft U.S. Treasury Department report found that AI companies are more closely tied to the American economy than internet companies were during the dot-com boom.” Using research from the University of Texas at Austin, the report examined how an AI downturn could propagate through interconnected balance sheets and capital commitments. If productivity and profits fall short, losses could reach private credit and chipmakers. Cloud companies, electric utilities and data center lenders are exposed too.
The Treasury did not predict an imminent crash. That distinction matters. It identified possible pressure points instead: electricity shortages and tight financing, plus supply disruptions and geopolitical conflict. Counter to the usual advice, diversification may offer less protection when those connections all lead investors toward cash. Bitcoin is often called digital gold, but lately it has traded much like a conventional risk asset. During a serious shock, liquidity may matter more than the safe haven story. A fall below $60,000 would put that story under real strain.
“Ray Dalio thinks a shortage of liquidity, not defective technology, could end the AI boom.” The Bridgewater Associates founder said investors often confuse rising valuations with spendable money. Consider a private company valued at $1 billion after raising only a small part of that amount. Its shareholders cannot convert those paper gains into cash until someone agrees to buy their stakes. Simple on paper. Messy in practice.
The trouble starts when too many holders want out simultaneously. Crypto knows this one. Many projects, particularly in decentralized finance, depend on steady inflows and high token prices. If cash dries up across markets, forced selling can reinforce itself. Would established protocols be spared? No. They would take a hit as well, and the slide can turn ugly quickly.
“Economists Bernstein and Cummings say the AI bubble is ‘still inflating,’ while technology investment has already passed dot-com-era levels.” In a recent Substack post, they estimated that technology investment had reached almost 5% of U.S. GDP. They also found that large technology companies were committing enough cash to AI projects to shrink their reserves. I keep coming back to that second point. A boom feels different once the biggest participants begin draining their own cushions.
Noble’s warning and Dalio’s liquidity argument point toward the same unresolved issue. The spending data makes it harder to ignore: Can AI profits catch up with the money companies have already committed? If enthusiasm fades first, some capital may move into safer assets. The rest may vanish on paper as valuations fall. Crypto suffers either way. Yes, that sounds more categorical than the earlier caveats—bear with me. Federal Reserve policy has repeatedly shown how quickly changes in available cash can move token prices, and a funding squeeze caused by AI losses would travel through many of the same channels.
What this means
“Warnings about an AI bubble suggest that investors may already be losing their appetite for risk.” A severe AI correction could spill into the broader stock market and push Bitcoin lower. Altcoins would probably fare worse because their price swings tend to be larger. The Treasury report adds weight to the concern: AI spending now reaches lenders and utilities. Chipmakers and cloud providers are tied in as well. This would not stay inside Silicon Valley.
“Crypto investors should keep an eye on technology stocks, institutional fund flows and the prices that may trigger heavier selling.” Nvidia (NVDA) and AMD are useful indicators because both depend heavily on AI infrastructure spending. Further sharp declines would suggest that investors are running out of patience. Crypto fund flows provide a separate signal. Slower purchases—or prolonged withdrawals—may indicate that institutions are cutting risk across their portfolios. I would watch both rather than wait for a neat declaration that the bubble has burst.
For Bitcoin, $60,000 is the price to watch. If it remains below that level, further losses could follow. Ethereum’s equivalent marker is $3,000. Is one threshold enough to call the market? Of course not. Inflation data and corporate earnings could move prices in either direction, especially if they change expectations for interest rates. The next FOMC meeting matters too, though I would pay more attention to what officials say about liquidity than to every carefully chosen adjective in the statement.