Be informed about free updates
Simply log in to Artificial intelligence Myft Digest – delivered directly to your arrived mail.
Granica sale of energy and infrastructure stakes this week after the progress of artificial intelligence of Chinese start-up deepseek shows that little is understood in AI power requirements, an international energy agency warned.
Investors have largely indicated shares in companies ranging from European gas turbine producers to American electricity generators and infrastructure providers, as Deepseek’s surprise has suggested that the development and use of AI may require less power than previously thought.
Hit companies include Siemens Energy, Ge Vernova, Mitsubishi Heavy Industries, Constellation Energy, Vistra Energy, Schneider Electric, ABB and Legrand.
Schneider, who overcome the value of Totalengies last September, while investors rushed to buy companies that would benefit from seemingly insatiable AI supply requirements, has now laged behind French oil in market capitalization.
“This sudden reaction emphasizes that the market currently does not yet have the right tools and information to evaluate the chances of electricity demand focused on AI,” said Thomas Spencer, one of the analysts who run Energy and AI work at IEA.
He added that greater volatility in the energy sector is “an inevitable risk” given current opacity over operational and energy performance of data centers.
The Deepseek model seems to be dressed using much less electricity than its western peers, demanding that they develop less than 10 percent of computer resources of the metal model Llam, according to Nature.
“Electricity consumption in data centers is proportional to their calculations,” said analysts from Rystada, energy consultation. “If what they claim is true, the total energy consumption could be significantly lower.”
Citi analysts said that if the figures prove to be correct, they will have a significant impact on companies that provide energy infrastructure. Companies that supply data centers have benefited from strong growth expectations. “A more computer -efficient AI could put those trends in question,” they wrote.
The sale was caused by an alarm throughout the energy sector, and one higher executive director Siemens Energy exclaimed that 21 percent of the company on Monday is just “not reasonable”.
The price of shares of the German company, a four-year-old Spin-off of the former Siemens Gas and Power Department, increased 336 percent last year, before a collision on Monday. It has a completed order book for its gas turbines from municipal services that build gas power plants to AI data centers.
After the market closed on Monday, it said it generated 9 billion euros in revenue in the first quarter, significantly ahead of expectations, and organically grew 18 percent of the year. The company has announced that the sale is exaggerated because data centers will only take a small part of its future growth.
CEO, Chief Financial Director and Chief of Relations with Investors in an Electric Infrastructure Company, held an invitation on Monday to try to understand the impact of the NEWS DEEPSEEK. They concluded that the situation would stabilize “little by little” if the financial results were positive.
Others insisted that, although Deepseek’s model could be more effective, the breakthrough will accelerate the adoption of AI, creating more demand for energy as a whole.
This effect is known as Jevons’s paradox after the English economist William Jevon, who noticed in 1865.
“Jevons Paradox is hit again!” Wrote Satya NadellaMicrosoft CEO of LinkedIn. “As Ai becomes more effective and affordable, we will see that its use is increasing, turning it into goods that we simply cannot get enough.”
Rystad noted that a wider energy transition can be affected by less intense AI, as some of the greatest electricity customers of zero carbon were large technological companies.
Amazon, for example, is the largest corporate customer of renewable energy in the world, after signing more than 500 electricity purchases in 27 countries each year, for a total of 77 electricity hours each year, which is equivalent to the entire demand of Belgium electricity.
Because less renewable energy sources are going to AI, it may become available to other sectors “helping to dispute faster use of fossil fuels,” said Rystad.
