Mistral CEO Challenges Claims of US AI Dominance Over China at Davos

Arthur Mensch, co-founder and CEO of Mistral, told Bloomberg Television on January twenty second, twenty twenty six that China is very much at parity with Western nations in AI capabilities. The French tech executive added that China's open-source AI technology is probably stressing the CEOs in the US, marking a stark departure from the narrative promoted by American political leaders and some technology executives.

Conflicting Assessments of the AI Race

Mensch's assessment stands in direct contrast to comments Trump made from the same forum just one day earlier. During his Davos speech, the president declared that we're leading the world in AI by a lot and we're leading China by a lot. Trump attributed American progress to policies allowing major technology companies to build their own electric capacity and power plants for energy-intensive AI facilities.

The divergent views underscore a growing debate among technology leaders about the true state of the US-China AI competition. The discussion has taken on increased urgency as artificial intelligence has emerged as a central focus at the twenty twenty six World Economic Forum, rivalling traditional issues such as trade tariffs and geopolitical tensions.

Evidence of Chinese AI Advancement

Mensch pointed to concrete evidence supporting his position, noting that Chinese companies have performed well in AI development even without access to the most advanced US chips due to export restrictions. He specifically highlighted the success of firms like DeepSeek, whose models have rivalled America's top AI chatbots.

DeepSeek's breakthrough has been particularly significant. On January twentieth, twenty twenty five, the Chinese research company released a high-performance large language model called R one at a small fraction of OpenAI's costs, significantly impacting the global AI landscape. Chinese large language models' global market share surged from three percent to thirteen percent in just two months, largely driven by DeepSeek's success. Chinese models captured more than ten percent penetration in thirty countries and twenty percent of market share in eleven countries.

In January twenty twenty six, DeepSeek started the year with a research paper on a new AI training method to efficiently scale foundational models and reduce costs. The paper, co-authored by founder Liang Wenfeng, proposes a rethink of the fundamental architecture used to train foundational AI models.

Industry Division Over China Policy

The Mistral CEO's remarks also put him at odds with Anthropic chief executive Dario Amodei, who told Bloomberg earlier this week at Davos that selling advanced chips to China would be a big mistake akin to selling nuclear weapons to North Korea. Amodei argued that the US chip embargo is precisely what has held Chinese AI companies back and that loosening restrictions would allow Beijing to close the gap.

The thing that is holding them back, and they've said it themselves, the CEOs of these companies say it's the embargo on chips that's holding us back, Amodei stated during his interview.

However, other industry leaders have offered assessments more aligned with Mensch's view. Nvidia CEO Jensen Huang said last year that China was nanoseconds behind the US, warning that China is going to win the AI race if the US does not surge ahead. White House AI adviser David Sacks has estimated that Chinese AI models lag American ones by three to six months, a timeframe that suggests a highly competitive landscape rather than clear American dominance.

European Perspective on AI Competition

The contrasting perspectives from two of Europe and America's leading AI executives highlight the uncertainty surrounding the technological competition between the superpowers. Mensch suggested the real concern should be Europe's position, stating that the continent needs to maintain its own AI development capabilities rather than relying on Chinese open-source models for critical applications.

Former Google chief executive Eric Schmidt echoed this concern at Davos, warning that Europe risks becoming dependent on Chinese AI unless it invests heavily in its own open-source capabilities. In the US, the companies are largely moving to closed source, which means they'll be proprietary systems, Schmidt noted, emphasising the strategic implications for European technological sovereignty.

Mistral's Growth Trajectory

Mistral, which has emerged as Europe's leading AI challenger to US firms like OpenAI and Anthropic, announced ambitious growth targets during the forum. The company said it expects to reach two billion dollars in revenue by the end of twenty twenty six and plans approximately one billion dollars in capital expenditure this year. Mensch revealed the company has a one billion dollar revenue target for twenty twenty six and is due to spend the same amount on infrastructure investments.

Implications for Global AI Competition

The debate at Davos reflects broader questions about the true state of global AI competition. DeepSeek's breakthrough underscores that the AI race is continuous, the gap between the United States and China is narrower than previously assumed, and that innovation by industry start-ups is the backbone of this race.

Microsoft president Brad Smith has warned that US AI groups were being outpaced by Chinese companies outside Western powers, especially in emerging markets. The competition remains intense, with the AI race in twenty twenty six still defined by a multipolar order, though the United States and China continue to yield the greatest influence.

The conflicting assessments from political leaders, established technology executives, and emerging AI companies suggest that the competitive landscape is more complex and uncertain than simple narratives of American dominance would suggest. As Chinese companies continue to demonstrate sophisticated capabilities despite restrictions on advanced hardware, the question of who leads in AI may depend less on technological superiority and more on how leadership is defined and measured.