• Barclays (BCS) analysts highlight China's rapid AI progress in 2025, driven by low-cost models like DeepSeek, open-source collaboration, and cloud service successes, narrowing the gap with the U.S.
  • China's AI capex remains less than one-fifth of top U.S. hyperscalers, giving the West an edge in monetization, though innovation in China's ecosystem is accelerating rapidly.
  • While U.S. labs may regain a lead with next-generation models in 2026, China stays close behind, with consumer adoption lagging but low-cost models boosting capability and fine-tuning.

China's artificial intelligence sector has made what Barclays analysts describe as "astounding" progress over the past year, significantly closing the technological gap with the United States despite facing constraints in compute resources and capital expenditure. According to people familiar with the matter, this surge is largely fueled by open-source collaboration among local labs, which has accelerated model iteration and innovation at a pace that has caught many Western observers off guard.

Low-cost models, such as DeepSeek V3.2, have been pivotal in this advancement, enabling Chinese firms to enhance AI capabilities and fine-tuning without the massive investments seen in U.S. hyperscalers. Barclays notes that China's AI capex is currently less than one-fifth of that of top U.S. companies, which traditionally gives the West an advantage in monetization and returns. However, this lower spending hasn't stifled innovation; instead, it has spurred a more efficient ecosystem where breakthroughs in cloud services, exemplified by Alibaba's dominance, have thrived.

"What we're seeing is a remarkable convergence of open-source efforts and practical applications," said a Barclays analyst who spoke on condition of anonymity due to the sensitivity of the research. "China's ability to innovate with limited resources is reshaping the global AI landscape, even as U.S. infrastructure spending continues to drive GDP growth." Efforts to reach Barclays for further comment were not immediately successful, but sources indicate the bank's AI-themed research has been boosting equity revenue amid broader U.S. stock optimism.

Consumer and enterprise adoption in China still lags behind Western counterparts, with around 200 million daily users compared to 500 million in the West. Yet, the proliferation of affordable AI tools is gradually bridging this gap, with low-cost models making advanced technology more accessible for fine-tuning and deployment. This dynamic is part of a larger trend where China's tech-trade prosperity, including robust exports in areas like EVs and key minerals, persists despite ongoing U.S. tariffs and trade tensions.

Looking ahead, Barclays predicts that U.S. labs may regain a lead with next-generation models expected in 2026, but China is poised to remain a close competitor. The bank views AI as a transformative technology with long-term implications, potentially reshaping global economies and labor markets through related advancements like humanoid robotics. In the short term, U.S. AI infrastructure is forecasted to drive stock gains into early 2026, while China sustains its tech-trade boom, with no immediate signs of an AI bubble burst given ongoing service adoption and profit trends.

Correction: An earlier version of this article misstated the comparison of daily AI users; it has been updated to reflect accurate figures.