AI experts come back from China amazement: the American grid is so weak, the race can already be finished

For AI American researchers, it’s almost unimaginable. In the United States, AI’s demand is collided with a fragile electricity network, the type of neck of extreme strangulation that Goldman Sachs warns could seriously suffocate industry growth.

In China, continued, it is considered a “solved problem”.

MA, a renowned expert in Chinese technology and founder of the media company Tech Buzz Chinatook his team on the road to look at the country’s progress in the first hand. She said Fortune That even if it is not an energy export, it attended enough meetings and spoke to enough initiates to leave with a conclusion which should send chills in the spine of Silicon Valley: in China, building enough power for the data centers is no longer ready to debate.

“It is a striking contrast with the United States, where AI growth is increasingly linked to the debates on the energy consumption of the data center and the limitations of the network,” she wrote on X.

The stakes are difficult to overestimate. The Data Center building is the basis of the progress of the AI, and the expenses of new centers now moves consumption expenditure in terms of impact on American GDP, which is worrying, because consumption expenditure is generally two thirds of the pie. McKinsey plans that between 2025 and 2030, companies around the world will have to invest 6.7 billions of dollars in a new capacity in the data center to follow the tension of the AI.

In a recent research note, Stifel Nicolaus warned against an imminent correction at the S&P 500, because it provides that this CAPEX Boom of the Data Center is a punctual construction of the infrastructure, while consumption expenditure is clearly on the decline.

However, the clear limiting factor in the development of infrastructure in the American data center, according to a survey by the Deloitte industry, is stress on the electricity network. Cities’ electrical networks are so low that some companies simply build their own electricity power plants rather than counting on existing networks. The public is increasingly frustrated by the increase in energy bills – in Ohio, the electricity bill for a typical household increased at least $ 15 this summer from data centers – while energy companies are preparing for a sea change in growing demand.

Goldman Sachs frames the crisis simply: “The request for insatiable power of the AI goes beyond the development cycles of the grid of the grid, creating a neck of critical strangulation.”

Meanwhile, David Fishman, a Chinese electricity expert who spent years following their energy development, said Fortune that in China, electricity is not even a question. On average, China adds more electricity demands than all annual consumption of Germany each year. The whole rural provinces are covered in solar on the roof, a province corresponding to the entire electricity supply of India.

“American decision -makers should hope that China remains competing and not an attacker,” said Fishman. “Because for the moment, they cannot effectively compete on the energy infrastructure front.”

China has an excess electricity offer

The quiet domination of China’s electricity, explained Fishman, is the result of decades of deliberate overdication and investment in each layer of the electricity sector, from generation to transmission through new generation nuclear.

The country’s reserve margin has never dropped below 80% to 100% nationally, which means that it has always maintained at least double the capacity it needs, said Fishman. They have so much space available that, instead of seeing AI data centers as a threat to network stability, China treats them as a practical means of “soaking up the excess offer,” he added.

This level of cushion is unthinkable in the United States, where regional networks generally operate with a reserve margin of 15% and sometimes less, in particular in extreme weather, said Fishman. In places like California or Texas, managers often emit warnings on red flag conditions when demand should force the system. This leaves little room to absorb the fast load increases AI infrastructure, Fishman NTOED.

The preparation gap is austere: while the United States is already experiencing political and economic fights as to whether the network can follow, China operates from an abundance position.

Even if the request for AI in China increases so quickly that renewable projects cannot keep pace, said Fishman, the country can exploit inactive coal power plants to fill the gap while building more sustainable sources. “It is not better,” he admitted, “but it is doable.”

On the other hand, the United States is expected to rush to bring new generation capacities, often faced with permit delays for several years, with local opposition and fragmented market rules, he said.

Differences in structural governance

Sweetness The material advantage is a difference in governance. In China, energy planning is coordinated by a long -term technocratic policy that defines market rules before investments are made, said Fishman. This model guarantees that the construction of infrastructure occurs in anticipation of demand, not in reaction.

“They are set up to hit the Grand Slam,” noted Fishman. “The United States, at best, can get on the basis.”

In the United States, large-scale infrastructure projects greatly depend on private investments, but most investors expect a return within three to five years: far too short for electrical projects that can take a decade to create and pay.

“Capital is really biased towards short-term yields,” he said, noting that Silicon Valley has channeled billions in “the nth iteration of software as a service” while energy projects are fighting for funding.

In China, on the other hand, the State leads money to strategic sectors before demand, accepting that all projects will not succeed, but ensuring that the capacity is in place when it is necessary. Without public funding to deactivate long-term bets, he argued that the American political and economic system is simply not set up to build the grid of the future.

Cultural attitudes strengthen this approach. In China, renewable energies are formulated as a cornerstone of the economy because they have economic and strategically meaning, and not because they have a moral weight. The use of coal is not thrown as a sign of wickedness, because it would be among certain circles in the United States – it is simply considered obsolete. This pragmatic framing argued Fishman, allows political decision -makers to focus on efficiency and results rather than political battles.

For Fishman, the point to remember is frank. Without a spectacular change in the way the United States builds and finances its energy infrastructure, the example of China will only widen.

“The capacity gap will only become more obvious-and develop in the coming years,” he said.