China Set to Dominate AI Race– Nvidia CEO Blames US Overregulation

Nvidia CEO Jensen Huang didn’t mince words: China is going to win the AI race, and the United States is handing Beijing the trophy on a platter of red tape and export bans. Speaking to the Financial Times, Huang painted a picture of two worlds moving in opposite directions.

In China, electricity for massive data centers is effectively half-price, subsidized by provincial grids in places like Gansu and Inner Mongolia, where any facility running domestic chips from Huawei or Cambricon gets slashed power bills. Meanwhile, the U.S. keeps tightening the screws: the January 2025 AI Diffusion Framework bans exports of chips exceeding 49,901 H100-equivalents per year, controls open-source model weights, and caps cloud access for Chinese firms.

Just yesterday, November 5, China fired back—ordering all state-funded data centers to use only homegrown silicon. Nvidia’s latest B200 and H200 chips? Grey-market contraband, smuggled through Malaysia or Thailand until those routes get shut down too.

The funding gap is even more brutal. Beijing is pouring over $200 billion into AI this year alone: $8.2 billion from the National AI Industry Fund, $47.5 billion from “Big Fund” Phase IV, a $138 billion low-interest lending window for chip and data center projects, plus local sweeteners—Shanghai handing out 1 billion yuan in compute vouchers, Hangzhou covering 60% of AI startup costs.

Compare that to America’s patchwork response: half a trillion dollars pledged by OpenAI, Oracle, and SoftBank for twenty “Stargate” data centers, but not a dime of it federal money. The closest Washington gets is the One Big Beautiful Bill Act from July, which throws a billion-plus at federal AI pilots while banning states from regulating the tech for five years—a windfall for Big Tech incumbents, a shrug for everyone else.

China’s regulatory playbook is a single national blueprint called “AI+,” rolled out in August. No fifty-state chaos, just one set of rules: hit 70% AI penetration across industries by 2027, mandate open-source models, and run ethics reviews on a fast track—ship first, label later.

The U.S., by contrast, is staring down the barrel of fifty new state-level AI laws, each one a potential choke point. Huang’s starkest line: “In China, power is free and permission is presumed. In America, both cost a fortune.”

The proof is already in the benchmarks. DeepSeek, a Chinese lab capped at 200,000 AI chips for 2025, trained its R1 model on pre-ban A100s and subsidized watts—and it now beats GPT-4o while sipping half the power. Silicon Valley’s anxiety isn’t theoretical anymore; it’s measurable in teraflops and kilowatt-hours.

While Washington debates export controls and model-weight embargoes, Beijing is wiring desert provinces with cheap electricity and writing checks big enough to build an entire AI stack from silicon to software. The race isn’t close, Huang warns—it’s already being run on different tracks, and only one side is sprinting unburdened.

Naorem Mohen

Naorem Mohen offers compelling insights on Artificial Intelligence and Cryptocurrencies. Explore his content on Twitter: @laimacha.