China's AI Progress: Closing the Gap but Facing Innovation Challenges

Davos, Switzerland—The global AI race just got more interesting. Speaking at the World Economic Forum, DeepMind CEO Demis Hassabis delivered a striking assessment: Chinese AI companies have narrowed the technological gap with Western counterparts to just six months. This marks a dramatic shift from earlier estimates that placed China "months behind" in artificial intelligence development.

The Shrinking Divide

The DeepMind chief pointed to China's rapid advancements in large language models as a key factor in this closing gap. He singled out one standout performer—the DeepSeek R1—whose capabilities sent ripples through Silicon Valley and even affected stock prices of chip manufacturers like NVIDIA.

"What we're seeing is remarkable progress in engineering implementation and product application," Hassabis observed. "Chinese teams have demonstrated they can not only keep pace but often exceed expectations when working with existing architectures."

The Innovation Gap Persists

However, the praise came with an important caveat. While acknowledging China's strengths in scaling and optimizing known technologies, Hassabis emphasized that breakthrough innovations continue to emerge predominantly from Western research institutions.

"Reproducing and improving existing models is one thing," he noted. "Creating the next paradigm-shifting architecture—that's where we're not seeing equivalent contributions yet."

The comment highlights ongoing concerns about China's ability to move beyond incremental improvements to fundamental discoveries—whether in developing successors to the Transformer architecture or advancing theories of embodied intelligence.

Policy Shifts and Market Realities

The discussion took an unexpected turn when Hassabis addressed recent U.S. policy changes. The Trump administration's decision to ease restrictions on AI chip exports to China drew measured approval from the DeepMind CEO.

"This adjustment makes practical sense," he said. "It relieves supply chain pressures while allowing American companies to participate more fully in China's growing AI ecosystem."

The policy shift creates what Hassabis described as a virtuous cycle: technology exports generate valuable data feedback that drives further model refinement back home.

Looking Ahead: The Next Frontier

As for Google's own ambitions, Hassabis revealed significant investments in robotics research aimed at bridging digital and physical intelligence. "We're pushing hard on embodied AI—systems that can perceive, reason and act in real environments," he shared.

The remarks come at a pivotal moment for global AI development. With computing power and data becoming increasingly commoditized, Hassabis suggested the next competitive edge will belong to those who can ask new questions rather than simply run existing algorithms faster.

Key Points:

• Narrowing gap: Chinese AI now estimated just 6 months behind Western counterparts
• DeepSeek R1: Chinese model making waves in Silicon Valley
• Innovation challenge: Breakthrough discoveries still dominated by West
• Policy impact: Eased chip export rules creating new market dynamics
• Next frontier: Embodied intelligence emerges as critical research area