At a French Alpine town known for its bottled water rather than high-stakes diplomacy, the leaders of the seven largest and wealthiest democracies will discuss how to solve pressing issues like Ukraine and the Middle East through Wednesday.
But the Evian agenda topics also reflect two interwoven anxieties: the Group of Seven’s dependence on China’s supply chains and reliance on the United States’ AI. The Trump administration’s decision to place export controls on Anthropic’s frontier models Fable 5 and Mythos 5 will be a “key topic” on the summit’s AI conversation, according to Andrea Renda, research director at the Centre for European Policy Studies who focuses on AI policy.
“The other six are quite annoyed and upset by the fact that the U.S. has actually tried to implement this differential treatment in terms of access to Claude Fable 5 for non-U.S. users,” Renda told Fortune, saying it’s “inaugurating an era” of weaponizing U.S AI against traditional allies.
While OpenAI’s Sam Altman, Anthropic’s Dario Amodei, and Meta’s Alex Wang are among the 11 AI CEOs attending the summit, it’s unlikely the leaders of France, Britain, Canada, Germany, Italy, Japan, and the European Union can actually deliver the summit’s goals to align on AI because of the U.S. and China.
“G7 pledges to adopt a more inclusive approach for AI, with steps towards developing economies, are unlikely to garner much (if any) interest from the U.S.,” Agathe Demerais, senior policy fellow at the European Council on Foreign Relations, told Fortune in an emailed statement. “This reflects the fact that the two leading AI powers are the U.S. and China, with not much space for other G7 economies to play a driving role here.”
China is not attending the G7 summit, though Bloomberg reported French president Emmanuel Macron considered inviting Xi Jinping in November, but participated in an “unprecedented” call with Macron ahead of the G7, signaling China’s status in the G7 discussion, according to Alisha Chhangani, an associate director at the Atlantic Council’s GeoEconomics Center.
But this was also Macron’s effort to leverage the U.S.-China rivalry as Europe finds itself squeezed between the threat of export controls on AI from one side and critical minerals on the other.
“It’s really Macron making China the elephant in the room in an effort to show the U.S. that there are other countries that they can work with,” Matt Pearl, the director of the Strategic Technologies Program at the Center for Strategic and International Studies, told Fortune. “It’s his effort to play China off of the U.S., and to make the Trump administration aware of that in order to create some leverage.”
The U.S. and China control 90% of global computing power and attract most AI investment, and nearly eight in 10 AI companies started last year in the G7 were based in the U.S., according to Stanford’s 2026 AI Index report.
But another dependency issue for the G7 is on Chinese manufacturing, which the agenda’s theme of “reducing global imbalances” is actually hinting at “in a diplomatic or a nice way”, according to Chhangani.
“The focus of the imbalances has been the China question, the industrial overcapacity question, and trade deficits,” Chhangani told Fortune. “China is the elephant in the room. The U.S. has been talking about this issue for so many years and Europe is a little bit new to it.”
China’s control of the critical minerals that underpin green energy technologies is creating a dependence that can “fuel protectionist policies” and “heighten national security concerns” among U.S. allies, according to the memo prepared for the G7 by a group of leading economists. But the question of which of the two dependencies—on the U.S. for the software making up AI infrastructure and on China for the physical layer underpinning AI and clean energy—is more important to solve is a hard one to answer for Europe, Pearl said.
“If they suddenly lacked critical minerals from China or suddenly had their supply chains for the environmental transition interrupted, that would be pretty catastrophic, and I think similarly on AI models,” Pearl said. “I think they feel like they need to solve both of them.”
