Cooling towers at the Three Mile Island Nuclear Power Plant in Middletown, Pennsylvania, on October 30, 2024.
Danielle Devries | CNBC
Earlier on Monday, the most tech-exposed power companies tumbled as the debut of China’s DeepSeek open-source AI lab led investors to question how much energy would actually be consumed by artificial intelligence applications.
Constellation energy and Vistur More than 16% in early deals. GE Vernova Slip about 18% voice energy The loss exceeds 15%.
Constellation, Vistra and Ge Vernova have led the S&P 500 this year as investors speculate that AI data centers will increase demand for massive amounts of power. All three markets suffered as investors worried about China’s AI competition.
DeepSeek released an AI model over Christmas that scaled AI CEO Alexandr Wang described as “earth-shattering” in an interview with CNBC last week. Scale AI provides training data for AI applications.
DeepSeek last week released an inference model called DeepSeek-R1, which competes with OpenAI’s O1 model. DeepSeek has since risen to the top of mobile app stores. Wang said DeepSeek has essentially caught up with Openai.
Wang told CNBC’s Andrew Sorkin in a Jan. 23 interview at the World Economic Forum in Davos, Switzerland: “Their model actually performed best, or roughly on par with the best American models.”
Microsoft CEO Satya Nadella DeepSeek describes it as “ultra-efficient.” Bank of America analysts said in a note on Monday that DeepSeek “challenges U.S. leadership’s conception of AI and casts doubt on high expectations for cloud capital spending, chip growth and power requirements.”
Technology companies anticipate needing so much electricity to power data centers that they are increasingly turning to nuclear power as a source of reliable, carbon-free energy.
For example, zodiac signs have been linked to Microsoft Restarting the Three Mile Island nuclear power plant outside Harrisburg, Pennsylvania. Tarun is working on Amazon Data center, powered by nearby Susquehanna Nuclear Generating Station.
Vistra has yet to sign a data center deal, despite investors seeing promise in its nuclear and natural gas assets. GE Vernova has soared this year on bets its gasoline and power grid businesses will benefit from demand for AI.
Analysts at Bank of America said there is still a need for grid investment in the United States and Europe.
“Grids in Europe and the United States remain underinvested and are one of the key bottlenecks in meeting load growth demands,” analysts said.