AI-exposed power stocks attempted a rebound on Tuesday after getting crushed the previous day alongside a tech rout following the release of a new AI model from the Chinese upstart DeepSeek.

Constellation Energy (CEG), the largest nuclear plant operator in the US, was trading flat midday after tumbling 21% during the prior session. 

Electricity and power generator Vistra Corp (VST) rose more than 4% after sinking by a record 28%. Power equipment maker and servicer GE Vernova (GEV) also gained nearly 4%, following a 21% drop. 

In late January, DeepSeek released a new AI model that was competitive to those of US rivals and cheaper to produce. The model surged in popularity, rising to the top of US iPhone downloads ahead of OpenAI's ChatGPT, and won praise from prominent figures. American venture capitalist Marc Andreessen called the model “one of the most amazing and impressive breakthroughs I’ve ever seen.”

The developments have rattled investors, who bid up power stocks on the belief that US hyperscalers will pour money into building out data centers required for AI use, thereby driving a surge in energy demand. Power stocks are coming of a ripping rally in 2024 and start of the year, with Constellation, Vistra and GE Vernova all touching record highs just last week. 

Now, questions are emerging whether long term energy demand will be as strong as expected. 

"Hopefully we can kind of resume some of the upward [stock] trend, but this really does kind of cast some uncertainty into that thesis which has become an investors’s base case," said Paul Zimbardo, managing director for US power utilities and clean energy research at Jeffries. 

The analyst noted data center deployment is "very likely intact" in the medium term. The question going forward is what happens in the later part of the decade and 2030's. 

He said, "Does that trend accelerate in the future? Does it kind of stabilize and even see some moderation of demand growth out later in the decade?"  

Wall Street analysts recognize estimates over data center energy usage are bound to change over time as the technology becomes more efficient. 

"AI is of course very nascent, young technology which means that trying to estimate how much energy AI will ultimately need is very speculative," Pavel Molchanov, investment strategy analysts at Raymond James, told Yahoo Finance on Tuesday.

"It's going to be a lot, and we should expect that electricity demand in the US and various other economies will accelerate as a result of AI. It's just a question of how much," he added. 

The increased energy needs are already visible. After nearly two decades of relatively steady demand, US electricity consumption jumped by 2% in 2024, according to government data. 

The Energy Information Administration forecasts electricity demand will continue growing at that rate in this year and 2026.  

ASSOCIATED PRESS

Data center deployments is also happening at the same time that utility companies build out their electric grids and infrastructure.

"All of this is a marathon rather than a sprint. It takes years to build data center projects, and it takes years to build power plants and transmission grids," said Pavel. 

Ines Ferre is a senior business reporter for Yahoo Finance. Follow her on X at @ines_ferre.

Click here for in-depth analysis of the latest stock market news and events moving stock prices

Read the latest financial and business news from Yahoo Finance