Tech stocks slump on AI spending concerns as investor fear prompts short covering
Dan Niles said extreme investor fear prompted him to cover short positions after a tech selloff. He also cited data center constraints and AI companies' rising costs as concerns.

A sharp selloff in technology stocks has left investors fearful enough to create the possibility of a short-term rebound, according to Dan Niles of Niles Investment Management. He also pointed to AI companies' costs and restrictions on data center construction as continuing concerns.
In a Yahoo Finance interview on November 21, Niles said market indicators showed sentiment had deteriorated so sharply that he had covered short positions in some stocks. He said it would only be possible to identify a market peak in hindsight.
Niles recalled that the Nasdaq fell 78% from March 2000 to its October 2002 low during the dot-com bust, with seven rallies of 25% along the way. Over the same period, he said, the S&P 500 fell 49% despite seven rallies of 14%, while the semiconductor index fell 83% despite seven rallies of 28%.
He said he did not view current conditions as a financial crisis and suggested the decline could resemble the semiconductor selloffs of 1997 and 1998, given that the AI industry remains at an early stage. The semiconductor index fell about 40% from peak to trough in each of those years but finished both years higher, he said.
Niles allowed for both a further selloff and a rebound in the near term. He said the volatility index was higher than when the United States attacked Iran and that demand for near-term protection was substantial. His indicators, he said, suggested it was time to prepare for a potential rally, prompting him to cover shorts even in stocks he was reluctant to buy back.
He said year-end trading could be affected by individual investors selling holdings at a loss to offset gains and by fund managers seeking to protect their bonuses. He also raised the possibility of a Monday crash, without presenting it as a certainty.
Among the changes to the AI investment landscape, Niles cited the passage of 40 of 41 ballot initiatives to halt data center construction. Constraints on building data centers could eventually reduce demand for chips if companies lack places to install them, he said.
He also flagged AI companies' costs. Combined sequential revenue growth at Anthropic and OpenAI slowed from 75% in the second quarter to about 48% in the third, while their combined operating losses increased from $7.9 billion to $8.4 billion, he said. The companies are cutting prices to gain market share while facing competition from open-weight models and other providers, he added.
Niles said most everyday tasks do not require the most advanced AI models, and that improving cheaper models could put more pressure on OpenAI's and Anthropic's high-end offerings. He also cited OpenAI's paying users, who he said accounted for just 10% of its user base, as a challenge for the company.