As the U.S. and China vie for dominance in AI, a new geopolitical dynamic is emerging among a group of so-called “middle powers.” From Canada, France, and Germany to Japan, South Korea, Israel, and the UAE, countries are attempting to build so-called “sovereign AI” systems designed to offer alternatives to technology from the U.S. and China. These sovereign AI efforts prioritize local companies building their own powerful AI models, local AI infrastructure, data control, and alignment with national or regional priorities.
That dynamic was on full display today as Canadian AI company Cohere announced a strategic partnership with German AI startup Aleph Alpha, forming a transatlantic alliance aimed at strengthening their position in the global AI market. Schwarz Group, a major Aleph Alpha shareholder, committed $600 million in future financing and will lead Cohere’s upcoming Series E round.
Cohere, founded in 2019 by Toronto-based researchers with roots at Google Brain—including CEO Aidan Gomez, who co-authored a seminal paper that introduced the ‘transformer’ architecture that underpins today’s large language models—has struggled to keep pace with U.S. leaders such as OpenAI, Anthropic, and Google, which have poured billions into talent, chips, and data centers. At the same time, Chinese firms like Alibaba and DeepSeek have expanded their global reach, intensifying the competitive pressure.
Increasing geopolitical tensions and a growing sense, particularly since the re-election of U.S. President Donald Trump, that Washington’s interests no longer align with those of many long-time allies has created an opening—and a sense of urgency—for companies operating outside both ecosystems.
Paris-based Mistral AI has built its strategy around that premise, positioning itself explicitly as a non-American alternative and emphasizing European control over AI infrastructure. CEO Arthur Mensch has framed the company’s mission in terms of independence, arguing that “AI should be a tool for empowerment, not dominance.”
A key part of that approach is Mistral’s use of “open-weight” models, which allow customers to inspect, customize, and deploy systems on their own infrastructure—sometimes even fully offline. The pitch has resonated with enterprises and governments wary of relying on opaque, centralized systems from U.S. or Chinese providers. (Chinese models providers have also offered open-weight models but users outside of China have raised concerns about biases in these models that favor views of Chinese history, politics, culture, and even medicine that meet the approval of the Chinese government.)
Mistral has also drawn backing from ASML, Europe’s most valuable tech company, which led a $2 billion funding round in September and signed a deal to use Mistral’s AI in its products and research. The round valued the startup at $14 billion.
The broader middle-power ecosystem is also expanding. Tokyo-based Sakana AI, founded by former Google researchers, is working to establish Japan’s domestic AI infrastructure. South Korea is backing a massive state-led effort to build out its own AI ecosystem. Israel’s AI21 Labs is reportedly exploring strategic options, while the UAE’s G42 is investing heavily in AI infrastructure and compute. There are also other European players such as Germany’s DeepL, Black Forest Labs, and the Dutch-based Axelera AI.
In January, Anton Leicht, a visiting scholar with the Technology and International Affairs Program at the Carnegie Endowment for International Peace, wrote that the AI middle powers—that is, most advanced economies that are not the U.S. or China—“need to find a strategy for participating in rapid AI progress.” They are in a tough spot, he explained, saying their participation in AI-powered growth is contingent on finding a niche; their sharing in AI benefits depends on access to compute, much of which comes from American and Chinese controlled supply chains; and their safety from AI-powered threats hinges on access to foreign top-tier defensive capabilities.
By combining resources and aligning across borders, Cohere and Aleph Alpha are effectively attempting to tackle all three of these challenges. It remains to be seen whether they will succeed, but for AI’s middle powers, the goal isn’t to win the AI race—it’s to make sure they’re not left out of it.
