Stay in view of the free updates
Simply subscribe to artificial intelligence Myft Digest – it is delivered directly to your inbox.
The International Energy Agency has warned that the sales process of energy and infrastructure shares this week after applying to artificial intelligence by the Chinese company Deepseek shows how much the power requirements of artificial intelligence are understood.
Investors have greatly obtained shares in companies ranging from European gas turbine makers to American energy generators and infrastructure providers, as a sudden Deepseek penetration suggested that the development and use of artificial intelligence may require less power than previously believed.
It included companies that struck Siemens Energy, Ge Vernova, Mitsubishi Heavy Industries, Constellenge Energy, Visstra Energy, Schneider Electric, ABB and Legrand.
Schneider, who exceeded the Totalenergies evaluation last September, where investors rushed to buy companies that will benefit from the apparently irreplaceable energy requirements, have now retreated behind the French oil in market value.
“The sudden reaction highlights that the market currently does not contain sufficient tools and information to assess expectations to demand the electricity driven by artificial intelligence,” said Thomas Spencer, one of the IEA energy analysts.
He added that more fluctuations in the energy sector were “an inevitable danger”, given the current output on the operational performance and energy centers.
It seems that the Deepseek model has been trained using much lower electricity than its Western peers, which requires less than 10 percent of the arithmetic resources of the Llama model in Meta, according to nature.
“The electricity consumption of data centers is appropriate to its arithmetic strength,” said analysts at Rystad, an energy -consulting company. “If what they claim is correct, the total power consumption may be much lower.”
CITI analysts said that if the numbers prove that they are accurate, it will have a significant impact on companies that provide energy infrastructure. Companies that provide data centers have benefited from strong growth expectations. They wrote: “Artificial intelligence can raise more effective in terms of these trends,” they wrote.
The sales process caused an alarm in the energy sector, as a senior executive of Siemens Energy cried out that the company's decrease by 21 percent at its share price on Monday simply “unreasonable”.
The share of the German company, a four -year -old cross in the previous gas and energy department in Siemens, increased by 336 percent last year, before the crash of Monday. It has a gas turbine turbine notebook from facilities to build gas power plants to serve artificial intelligence centers.
After the market was closed on Monday, he said that he had achieved 9 billion euros of revenue in the first quarter, before expectations, and was organically growing by 18 percent on an annual basis. The company said that the sale has been late, because data centers will only explain a small segment of its future growth.
The CEO, financial manager and head of investor relations at one Electricity Infrastructure Company on Monday made an invitation to try to understand the impact of Depsic News. They concluded that the situation will settle “little by little”, if the financial results are positive.
Others insisted that although the Deepseek model may be more efficient, the achievement will accelerate the adoption of artificial intelligence, which creates more energy demand in general.
This influence is known as Jevons Paradox after the English economist William JPA, who in 1865 noticed that technological breakthroughs that increased coal efficiency increased coal consumption.
“Jevons Paradox Strikes again!” Satia Nadila BooksMicrosoft CEO of LinkedIn. “Since artificial intelligence becomes more efficient and accessible, we will see its use rises, and turn it into a commodity we cannot get enough.”
Rystad pointed out that the broader energy transmission can be affected by the least intensity AI, because some of the largest buyers of carbon electricity were large technical companies.
For example, Amazon is the largest buyer of renewable energy companies in the world, with more than 500 energy purchasing agreements in 27 countries, a total of 77 hours of electricity each year, equivalent to the demand for entire electricity in Belgium.
Risad said that with the transmission of less renewable energy to artificial intelligence, it may become available to other sectors “helping to remove fossil fuels faster.”
