Ask an American and a Chinese citizen the same question about artificial intelligence and you get almost opposite answers. A poll published last November found that 87 percent of people in China say they trust AI, against just 32 percent in the United States. That gap, more than any benchmark score, may be the most telling number in the technology's story right now.
The American mood has curdled into something political. As opposition to AI data centers spreads across the country, some Silicon Valley figures and Republican politicians have begun accusing China of stoking the anger, claiming on thin evidence that Beijing is paying local activists. Foreign Policy's reading is blunter: the unpopularity is genuine. Data centers are loud, thirsty for power and water, and increasingly a symbol of a technology many Americans never asked for. The fight over them has become a proxy for a broader unease that cuts across the usual party lines.
In China the same machines arrive to a warmer welcome. The government treats AI as a national project, the public is broadly enthusiastic, and the argument is cooler on both sides. In Beijing's tech district, a bar called AGI Bar lets patrons use DeepSeek free while they drink; the walls carry the logos of the country's big labs like sports badges. The contrast is not a law of nature. Plenty of technologies are unpopular in China too. But on AI, public feeling and state policy are pointing the same way.
That enthusiasm is showing up as price pressure. Chinese models, many of them open-weight, have forced America's leading labs to cut deals. OpenAI has slashed fees by 80 percent on its lightweight model Luna; Anthropic has released a system approaching its best performance at half the price. DeepSeek's V4 Flash currently tops the usage leaderboard on the developer platform OpenRouter, and, confident in its position, the company says it plans to raise prices. When the cheap challenger starts charging more, you know the market has shifted.
None of this means China has caught up at the frontier. By the raw numbers, the United States still leads. China's own 2026 Global AI Innovation Index puts the US on 78.44 points to China's 60.49, with Britain a distant third; Stanford's AI Index counted 59 notable American models last year against 35 from China and just two from Europe. The lead in building the best models is real. The question is whether it matters as much as the lead in getting people to use them.
That is the tension worth watching. Diffusion, not just capability, is how a technology reshapes an economy, and China is spreading AI faster, cheaper, and to a more willing public, while also pitching its open models to a Europe anxious about being trapped between two giants. Trust can be engineered, and China's long history of acquiring technology by less-than-open means gives foreign buyers reason to hesitate. But a country that greets a new tool with 87 percent trust will find more uses for it than one that greets it with 32. The frontier, for now, is American. The adoption curve is looking increasingly Chinese.