The quiet rise of Chinese AI inside Western companies

For all the attention paid to export controls, tariffs, and technological decoupling, something unexpected is happening inside the very companies Washington was desperately trying to protect: some of America’s most successful technology firms are quietly turning to Chinese artificial intelligence (Opens in new window).

Not because they have suddenly become sympathetic to Beijing. Not because they are making a geopolitical statement. They are doing it for the same reason Australians will happily swap one flat white for another if it tastes the same and costs less.

The transition has largely escaped public attention. There have been no headlines announcing a “pivot to China”, no carefully crafted corporate campaigns. Yet beneath the surface, Chinese AI models are finding their way into products used by millions of people across the West, and Airbnb offers a revealing example. Alibaba’s Qwen is among 13 AI models the platform’s customer service agent is built on, and Airbnb CEO Brian Chesky (Opens in new window) described Qwen as “very good ... also fast and cheap.” That may prove to be one of the most important sentences in the global AI race – though it has attracted Congressional scrutiny and Chesky has subseuently clarified (Opens in new window) that Chinese companies do not have access to any data.

The challenge for American AI companies is that many customers are beginning to view artificial intelligence less as a national security asset and more as a commodity. If two models produce roughly the same result, just like the flat white, few executives are eager to pay several times more simply because one happens to be developed in California rather than Hangzhou.

The trend extends beyond customer service. Cursor, now one of the most popular AI coding tools in the United States, has reportedly incorporated versions of Kimi (Opens in new window), developed by Beijing-based startup Moonshot AI, but most users would never notice – they open the application, write code, and move on with their day.

And that invisibility may be China’s greatest advantage.

For years, discussions about technological competition assumed that success would belong to whoever built the most advanced model. Increasingly, however, another question is emerging: what if the real winner is the one that can deliver good-enough intelligence at a dramatically lower price?

In many ways, this captures the difference between the American and Chinese approaches to AI, where the Silicon Valley continues to dominate the frontier and dictate innovation, spending extraordinary sums pursuing ever more powerful models, constructing massive data centres, and pushing the limits of computing infrastructure. The ambition is clear: build the future before anyone else can.

On the other hand, Beijing’s approach has often looked different. Rather than focusing exclusively on the next breakthrough, Chinese firms have become adept at squeezing more value from existing technologies. Efficiency, not necessarily novelty, has become the most lucrative competitive weapon.

DeepSeek’s (Opens in new window) rise captured this dynamic: cut off from some of the world’s most advanced semiconductors, the company was forced to optimise. Necessity became innovation. Whether every claim surrounding its development costs withstands scrutiny is almost secondary. What mattered was the signal it sent: constraints can sometimes produce advantages of their own.

There is a certain irony here. American restrictions were designed to slow China’s AI progress. Instead, they may have accelerated the search for leaner and more commercially sustainable approaches.

Investors, meanwhile, have noticed.

Hong Kong’s stock exchange (Opens in new window) has reclaimed its position as one of the world’s leading IPO markets – by some measures the busiest of all in early 2026 – propelled largely by enthusiasm surrounding Chinese AI firms. Companies such as Zhipu AI and MiniMax have attracted extraordinary attention from investors eager to participate in what many view as the next great technological transformation – the atmosphere occasionally feels familiar.

Not because the technology lacks value. The internet transformed the world, despite the collapse of the dot-com bubble. Railways transformed economies despite speculative excess. Revolutionary technologies and financial bubbles are not mutually exclusive; history suggests they often arrive together.

That is what makes the current moment so difficult to assess.

In the United States, hundreds of billions of dollars are being committed to infrastructure on the assumption that future demand will justify today’s spending. In China, investors are attaching breathtaking valuations to companies that, in some cases, remain largely unknown outside specialist circles.

Both stories may ultimately prove correct. Artificial intelligence could become as foundational as electricity or the internet. But markets have a habit of pricing in success long before success actually arrives.

The real question, then, is no longer whether Chinese AI can compete – that debate is fading quickly – but whether the extraordinary sums now flowing into AI – whether in Silicon Valley, Hong Kong, or Beijing – reflect the emergence of a new industrial revolution or are simply the latest chapter in an old story: technological optimism outrunning economic reality.