From euphoria to doubt: Markets question who pays for AI

In a single trading day, Wall Street went from celebrating artificial intelligence to questioning how it is paid for. Jacques Lemoisson, head of the economic intelligence firm Gate Advisory, says the warning signs had been building for months.

Thursday’s sell-off was one of the most violent in years. The Nasdaq Composite fell 5.52%, the Nasdaq 100 6.36% and the S&P 500 4.36%. The Philadelphia Semiconductor Index (SOX) dropped 12.2%.

Chipmakers took the heaviest losses. Nvidia lost 9.3%, while AMD and Micron each shed more than 17%.

About $3 trillion in market value was wiped off the S&P 500 in one day. The Nasdaq is now in correction territory, close to 12% below its October peak.

Lemoisson argues that markets did not overreact. “They simply took several months to grasp what was becoming increasingly obvious,” he writes.

Five forces at once

There was no single trigger, Lemoisson says. The rout came from five pressures feeding off one another:

• AI returns in doubt. Investors are no longer sure the hundreds of billions poured into data centers and chips will pay off.
• Tighter financing. Borrowing to fund the AI build-out is getting harder and more expensive.
• Midterm shock. The U.S. midterm election results added political uncertainty.
• Chip outlook reversal. Expectations for semiconductor demand, until recently seen as nearly unlimited, have turned.
• Crowded trades unwinding. Institutional investors piled into the same handful of AI stocks, and are now selling them at the same time.

Revolution and bubble can coexist

Lemoisson has been making the same point since the summer. AI can be a major industrial revolution and, at the same time, a financial disaster for those who overpaid to fund it.

The distinction was easy to dismiss while the Nasdaq kept climbing. After a 5% drop in one session, it is much harder to ignore.

As U.S. investors start to question the cost of America’s AI dominance, Lemoisson raises a strategic question. If the American infrastructure race slows, could China gain ground? Its open-source models and less capital-hungry approach to AI could prove an advantage if money for mega-data centers dries up.