The Tech Download: China’s AI surge — real threat or hype?

This report is from this week’s The Tech Download newsletter. Like what you see? You can subscribe here.

Most of the world’s population could be running on a Chinese tech stack in five to 10 years, one analyst told CNBC’s “Squawk Box Europe” earlier this week.

The U.S.’ “perceived monopoly” on tech and AI has been broken by China, said Rory Green, TS Lombard’s chief China economist, adding that the country’s rapid advancement is threatening to shake up American dominance in the market.

Green’s comments come as China races against the U.S. to develop artificial general intelligence (AGI) — where AI matches human capabilities — and roll out the technology across society. Big moves are being made to scale homegrown makers of AI chips to rival Nvidia and local AI companies are making waves on stock exchanges.

But could China really win the AI race?

2025 was the year that many in the West really began to pay attention to Chinese frontier AI companies, with DeepSeek causing a market frenzy and local tech giants releasing a slew of their own models since.

Despite progress putting many model makers in China “close to” leading AI labs such as OpenAI, Anthropic and Google, they still lag behind, Paul Triolo, partner at advisory firm DGA-Albright Stonebridge Group, told CNBC.

Compute is the big problem. Export controls limiting access to advanced Nvidia GPUs create a “real ceiling on the compute side of scaling,” Nick Patience, AI lead at research firm The Futurum Group, told CNBC.

These are shortcomings Chinese AI companies are all too aware of.

DeepSeek acknowledged in a December research paper it faced “certain limitations when compared to frontier closed-source models” such as Gemini 3, including compute resources. A technical lead of Alibaba’s Qwen team said at a conference in Beijing in January that there was less than 20% chance that a Chinese firm would surpass U.S. tech giants on AI in the next three-to-five years, the South China Morning Post reported.

But the country’s AI companies do have some advantages over U.S. counterparts.

The clearest area where China is outcompeting the U.S. is efficiency-driven model development — achieving strong performance at lower compute cost, said Patience.

“Whether driven by necessity (chip constraints) or strategy, Chinese labs have made notable advances in inference efficiency and quantization techniques that the broader industry must take seriously,” he said.

There’s also power. China is undergoing an energy boom and has added more power capacity in the past four years than the U.S. has in total, Bloomberg reported last month.

“This will help with the diffusion of AI in China, because energy will be more available to run data centers and other AI related infrastructure,” said Triolo.

By releasing competitive open-source or open-weight models, Chinese labs are “eroding the commercial moat that U.S. closed model vendors have relied on,” said Patience.

“If an enterprise can deploy a capable open-weight Chinese model on its own infrastructure at low cost, the business case for paying premium prices to U.S. providers weakens considerably,” he told CNBC.

With AI competition shifting from “model performance to value realization” — as Julian Sun, VP at research firm Gartner, told CNBC — that could be a significant boon for Chinese AI companies.

Green’s Chinese tech stack prediction is “plausible as a long-run scenario” for parts of the Global South where cost is the dominant consideration and geopolitical alignment with the U.S. is weaker, according to Patience, though he cautioned it’s a “speculative 5-10 year call.”

The U.S. still has some big advantages.

American companies continue to lead in areas such as advanced semiconductors, frontier-model research and hyperscaler infrastructure. They also continue to court huge sums from investors and companies, and governments have deployed their tools across the globe.

Sun says he sees the global AI landscape becoming “multi-polar” across different layers of the tech stack rather than being dominated by a single ecosystem. Time will tell how this plays out geographically as AI systems become omnipresent in societies.

More than 50% of enterprises’ current software could be replaced by AI, the CEO of Mistral AI told CNBC on Wednesday.

Meta announced it had struck a massive chip deal with Nvidia that will see the tech giant deploy next-generation Vera Rubin systems.

Europe is scrambling to undo the dominance of U.S. digital services in its infrastructure amid global geopolitical tensions, officials told CNBC.

Rapid advancements in quantum computing have intensified investment in the sector and sparked discussions about how these powerful computers will integrate with industries like the already booming data center sector.

A federal grand jury indicted three Silicon Valley engineers on charges of stealing trade secrets from Google and other technology companies and transferring sensitive data to Iran, prosecutors said Thursday.

The quote: Microsoft President Brad Smith told CNBC that American tech companies should “worry a little bit” about the subsidies Chinese competitors get from their government in the AI race.

The big picture: As competition ramps up between U.S. and Chinese model builders, Beijing is supporting its AI companies with measures such as a multi-billion-dollar national investment fund and vouchers for cheaper energy for compute.