When asked, “How can the U.S. win?”
Elon Musk paused for three seconds and replied: “China’s power generation is three times that of the United States.”
In Huang Renxun’s “five-layer cake” model, energy sits at the foundation.
Chips, models, applications—all rest upon it.
China: 10,583 TWh (terawatt-hours, equivalent to 10 billion kilowatt-hours). U.S.: 4,520 TWh.
China’s total exceeds the combined output of the United States, the European Union, and India.
One gigawatt-hour (GWh) of electricity—equal to one million kilowatt-hours—corresponds to approximately $6 billion in GDP.
Power infrastructure is among the most critical forms of foundational development.
This is not merely about AI; it touches on the fundamental design of a nation’s institutional framework.
The vast majority of China’s power facilities are state-owned, where profitability is not the sole determinant.
In contrast, the majority of U.S. power companies are privately owned, unable to sustain losses over several years—or even decades—without shareholder backlash. Market dynamics also dictate pricing, constrained by the absence of sufficiently high margins.
Musk was not speaking solely about electricity. He was referring to the foundational platform for AI competition.
Who controls energy, who holds the right to discuss computing power and competition.
Who stands a chance of emerging as the ultimate winner in the AI race—this is the essence of the matter.
Original: toutiao.com/article/1877806570081288/
Disclaimer: The views expressed in this article are those of the author alone.