In a recent interview, a senior NVIDIA executive offered perhaps the most candid assessment yet of the global AI race, admitting openly that China is going full throttle across the entire AI ecosystem and will soon surpass the USA in building and deploying advanced AI platforms. According to him, AI is not a single monolithic technology; it is a vertical stack of more than five layers, stretching from the bottom layer of raw energy to the very top layer of applications. And across this entire stack, China is now moving with a speed, scale, and national purpose unmatched by any other country.
He began with the foundational fact that AI is ultimately an energy-hungry technology. Training frontier models requires immense electricity— entire power stations dedicated to data centres— and here the gap is stunning. China today generates over 9,600 TWh of electricity annually, nearly double the roughly 4,800 TWh produced in the USA.
Massive hydropower, coal, solar, and wind installations continuously expand China’s energy base, giving it the ability to power tens of thousands new data centres and fabrication plants. Without this energy, nothing else in the AI stack functions. The USA, he warned, simply does not have enough electricity to reindustrialize, restore manufacturing, or scale AI at the level required to maintain global leadership. Even if it decided now to double its energy output, it would take five to ten years to build the necessary power infrastructure.
The second structural difference is speed. In the USA, constructing a semiconductor manufacturing facility can take three years or more because of permitting, regulation, zoning, and litigation. In China, the same facility can be erected in a matter of months, sometimes weeks, because the national system aligns government, regulation, industry, and capital around a single purpose: build fast, build big, and build continuously. This United States and Europe, by contrast, expand capacity by barely 10–20 percent annually. If this trend continues, China’s domestic production will increasingly cover its own AI needs, reducing dependence on imported chips and eventually surpassing rivals in total volume. The idea that China would always lag in semiconductors is collapsing under the weight of its own accelerated progress.
Then comes the higher layer of the AI stack: the models, platforms, and applications that convert computation into economic value. Here the cultural and institutional differences are profound. In China, almost every major AI model is open source, from Baidu’s ERNIE to Alibaba’s Qwen to models emerging from startups like Zhipu and 01.AI. Tens of thousands of researchers, students, surgeons, engineers, and small businesses use these models freely, improving productivity and innovation across society. China is not earning money from these open models; instead, it is earning capability, scale, and national momentum. Open-source AI becomes a nationwide accelerator that multiplies learning, experimentation, and economic output.
In the United States, by contrast, almost all frontier models are closed and commercial. They are powerful, but they are fenced behind subscriptions, APIs, restrictions, and corporate ownership. This produces revenue, not widespread capability. The NVIDIA executive made a critical point: China is not monetizing AI at the application level; it is weaponizing openness to democratize AI across its entire population. This is why Chinese AI applications are spreading faster than American ones — because the barriers to entry are near zero.
The verdict from within the USA’s own technology leadership is unmistakable. China is not catching up; China is accelerating past. It is building more, deploying more, training more, and opening more. It has the energy, the infrastructure, the manpower, the optimism, the manufacturing base, and the national will to dominate the AI century. If trends continue, then sooner rather than later, China will indeed be number one in the AI business— and the world will be shaped by the platforms it builds.
He also offered a striking cultural insight. When surveyed, nearly 80 percent of Chinese citizens view AI as a positive force for society. In the USA, the sentiment is almost the reverse— roughly 20 percent view AI positively while 80 percent express fear, distrust, or resentment. Innovation cannot thrive in a cultural climate of suspicion. China has fused optimism, national pride, and collective ambition into its AI mission. The USA has not.
One example he gave was Huawei, now emerging as one of the fastest-growing AI companies despite US sanctions. The company builds chips, models, cloud platforms, and 5G systems simultaneously and is advancing at a speed that even American executives privately acknowledge with respect. When the US president asked him for recommendations on how to reindustrialize, he replied that the first requirement is not money or technology but energy— without doubling energy output, no amount of reshoring will succeed. AI factories, semiconductor fabs, and advanced manufacturing all require stable, abundant, and cheap electricity. China has it. The USA does not.
He also warned that the US. system, while superior in pure innovation and scientific breakthroughs, is deeply constrained by regulation, litigation, and slow infrastructure development. China, meanwhile, has government coordination, manufacturing speed, infrastructure readiness, and cultural enthusiasm. As a result, the remaining AI gap between China and the USA is now as small as six months— and shrinking.
Behind these observations lies a broader truth: China is building the full AI stack. Energy, infrastructure, semiconductors, compute clusters, models, applications, and societal adoption. Every layer reinforces the others. The USA excels in some layers, particularly research and frontier architecture design, but lags in foundational layers like energy, manufacturing, and large-scale deployment. Without rebuilding the bottom of the stack, it cannot maintain leadership at the top of the stack.`
Yet he was not pessimistic about the USA. He emphasized that it was a nation of extraordinary people, innovators, scientists, and extraordinary entrepreneurs. If the country decides to renew itself — doubling energy capacity, rebuilding industrial infrastructure, streamlining regulations, enabling faster construction, and fostering a cultural shift that welcomes AI rather than fears it— then the USA can still compete with China, perhaps even share global leadership. But without such a national transformation, the momentum is clearly with Beijing.
The verdict from within the USA’s own technology leadership is unmistakable. China is not catching up; China is accelerating past. It is building more, deploying more, training more, and opening more. It has the energy, the infrastructure, the manpower, the optimism, the manufacturing base, and the national will to dominate the AI century. If trends continue, then sooner rather than later, China will indeed be number one in the AI business— and the world will be shaped by the platforms it builds.
