Энергия становится золотым стандартом для искусственного интеллекта: США в авангарде Translation: Energy Becomes the Gold Standard for Artificial Intelligence: The U.S. at the Forefront

The United States is home to 5,426 data centers—more than all other major countries combined. Furthermore, the nation is projected to construct data centers worth $40 billion—a staggering 400% increase since 2022, as noted by the analytical publication, The Kobeissi Letter.

For the first time in history, the value of under-construction data centers in the U.S. will surpass that of office buildings. Three years ago, the ratio was 1 to 7.

“This is a modern-day gold rush,” emphasized The Kobeissi Letter.

Since the launch of ChatGPT in November 2022, the value of data centers being developed in the U.S. has risen from $12 billion to $40 billion. Predictions indicate that energy consumption will continue to rise. By 2030, data centers are expected to account for approximately 8.1% of the nation’s total electricity usage, up from 3.9% in 2023.

Morgan Stanley has forecast a power shortfall of 36 gigawatts in the U.S. over the next three years.

However, the number of data centers does not tell the whole story, as pointed out by The Kobeissi Letter. China has around 449 data centers, yet its energy consumption is comparable—12 gigawatts versus 14 gigawatts in the U.S. This is due to the larger size of Chinese data centers compared to American ones.

Per capita energy consumption in China is now on par with that of the EU, having increased by 350% since 2000.

“This is why we keep reiterating the same point: energy will soon become the world’s most valuable resource. Nations are engaging in an arms race in artificial intelligence, and the world will need alternative energy sources,” The Kobeissi Letter remarked.

Since the introduction of ChatGPT, electricity prices in the U.S. have surged by 23%. Since 2020, it has increased by 40%. This greatly exceeds the overall inflation rate in the country.

As a potential solution to electricity shortages, the analytical outlet cites nuclear energy. Unlike solar and wind installations, nuclear plants operate continuously, which is essential for the uninterrupted functioning of AI. Moreover, it is one of the most cost-effective energy sources available.

However, constructing nuclear power plants is a lengthy process. Currently, there are no major reactors being built in the U.S., while China is constructing 29.

“Without a significantly greater volume of electricity, the AI revolution simply cannot happen. By 2035, data centers for AI will consume around 1,600 terawatt-hours—approximately 4.4% of global energy supply,” the analysts concluded.

OpenAI stated that the U.S. must significantly ramp up investments in new energy capacity if it wishes to outpace China in the AI race.

“Electricity is not merely a utility. It is a strategic asset that is crucial for building AI infrastructure. It will ensure our leadership in the most important technology since the advent of electricity,” the company indicated in a letter sent to the White House Office of Science and Technology Policy.

They urged authorities to commit to annually constructing new energy capacities.

“Electrons are the new oil,” the startup emphasized.

In July, the CEO of SAP SE, Christian Klein, declared that Europe does not require a surge of new data centers to compete in the AI sector.