中等强国正在 AI 竞赛中输给 Nvidia、Anthropic 与科技巨头
Middle powers are losing the AI race to Nvidia, Anthropic, and Big Tech | Fortune
中等强国在 AI 竞赛中落后于 Nvidia、Anthropic 等私营科技巨头。斯坦福 2026 年 AI Index 报告显示,截至 2025 年美国有 59 个“值得关注”的 AI 模型,中国 35 个,韩国 8 个,模型开发高度集中于美中两国。Anthropic 9 月威胁报告披露 Claude 被滥用于针对东南亚政府机构的网络行动、中国威胁行为者的水下武器系统工作,以及伊朗针对美国海军军舰的图谋。
正文
当前显示原文,中文全文译文尚未就绪。
Lindsay Newman is an associate fellow at Chatham House.
Middle powers are trending.
Since Canadian Prime Minister Mark Carney warned the world’s middle powers at the World Economic Forum in Davos in January to get “at the table” or else end up “on the menu,” a movement has gathered momentum—built on the optimism that the way through today’s geopolitical uncertainty runs through the middle.
This vision rests on a flawed premise. The middle powers argument assumes the nation-state is still the primary actor in great-power competition. This is not the case.
The competition ahead will be one where private-sector actors already are, and will keep, pursuing global power.
This is not a moment of national dominance. It is the dawn of multi-stakeholder geopolitics—and middle powers are arriving late to the table after the private sector got there first.
Nvidia now has a market capitalization that sits just below Germany’s annual GDP and just ahead of Japan’s. Apple’s ranks below Japan but above India. Amazon’s is wedged between Canada and Brazil.
After decades where public policy shaped and constrained private firms, corporations are now setting priorities, building bridges, and leading on global challenges.
Biotech firms developed and distributed the mRNA vaccines that helped end the COVID-19 pandemic. The world’s electronics, consumer products, and innovation sectors run on the output of Taiwan Semiconductor Manufacturing Company, the Taiwanese chipmaker.
U.S. President Donald Trump’s approach to diplomacy mirrors the rise of the private sector. On a state visit to the United Kingdom last year, Trump offered investment deals in place of trade concessions. And when he hosted Chinese President Xi Jinping for a state dinner at the White House in September, he invited a vast coterie of corporate chief executives and finance-industry leaders to attend.
The AI frontier
Nowhere is the private sector’s geopolitical influence more apparent than in AI. Increasingly, state actors are relying on non-state, private sector technology firms for capabilities that were once solely the domain of national agencies. Private firms like SpaceX and Blue Origin are leading 21st-century space exploration. Elon Musk’s Starlink satellite network has reportedly been deployed across conflict zones from the Sahel to Iran and provides civilian internet access and battlefield intelligence in Ukraine.
AI models are being folded into operational planning, simulations, and cyber operations in service of foreign policy. Earlier this year, Anthropic, the AI safety company, clashed with the U.S. administration over how its models could be used by the Department of War. The Trump administration designated Anthropic a supply chain risk to U.S. national security—a first for an American firm.
Already in 2026, privately developed AI is being deployed for influence campaigns, biological weapons development, cyber operations, and surveillance by malign actors. Anthropic’s September threat report identified cases of misuse or attempted misuse of its Claude model for malicious activity. These instances include a cyber operation targeting a Southeast Asian government agency to extract civilian records; a threat actor based in China that used Claude to progress work on an undersea weapons system; and suspected Russia-based threat actors using AI with the intention of building an autonomous kamikaze drone swarm.
And, finally, Anthropic said it disrupted a plot by Iran to use Claude to target U.S. Navy warships in the Middle East.
To counter such activity, private AI companies are stepping into roles once held by national intelligence and defense agencies. There is no longer any denying that private sector actors are geopolitical actors.
The emergence of private sector actors as a source of global power does not, on its own, mean that states have become merely bystanders. The bad news for middle powers, though, is that they are only tangentially playing in the AI field amid its expanding influence.
According to Stanford University’s 2026 Artificial Intelligence Index report, for instance, AI model development remains highly concentrated within the U.S. and China. Through 2025, the U.S. had 59 “notable” AI models (models meeting certain notability criteria including citation rate, compute costs, and usage levels) to China’s 35. The only other country with multiple notable AI models worldwide was South Korea, with eight.
Redrawing the rules
In Davos, Carney spoke of a fading rules-based order—of how the great powers had broken a bargain with their less powerful counterparts. His call is for these less powerful middle powers to withdraw their consent from a global order that no longer serves them and to chart a new path “with impact.”
It’s a tantalizing idea, offered at a moment of great change, when AI is defining how we think, work, fight, and manage risk. Old rules are not just fading but being redrawn entirely
But as the debate over AI safety shows, there’s no global consensus about how to govern the future’s most powerful technology. The leaders of private firms are stepping into that vacuum of state-led leadership and beginning to set the rules we live by.
The seats at the table are being taken not by middle powers but by private sector actors.
And so long as middle powers are not leading on AI development, attracting AI investment, or dominating parts of the AI supply chain—so long as they are not on the AI field—they will not exert geopolitical leverage in this emerging world.
Instead, it will be private sector actors, alongside national leaders in the U.S. and China, who shape the next global system.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
Brainstorm Tech International will convene CEOs and founders, technologists and investors, thinkers and doers to tackle the challenge of turning technological promise into lasting business and societal value. December 3 in Macau. Apply here.