The AI race is the competition among technology companies, and increasingly among governments, to develop and deploy more capable AI systems ahead of their rivals. The term covers two overlapping contests: a commercial race for customers, researchers, chips and capital, and a geopolitical race, above all between the United States and China, for economic and military advantage.
Two races
The commercial race accelerated after OpenAI released ChatGPT in November 2022. Google’s management declared a “code red,” according to the New York Times, and rushed out its own chatbot; Microsoft’s chief executive Satya Nadella said of his company’s challenge to Google in search: “I want people to know that we made them dance.” Since then, spending on chips and data centers by the largest technology companies has reached hundreds of billions of dollars a year (see Hyperscaler and AI Bubble).
The geopolitical race has its own milestones. In 2017 China’s government set the goal of becoming the world’s primary center of AI innovation by 2030. From October 2022 the United States restricted exports of advanced AI chips to China, though the Trump administration later allowed some sales under license (see Export Controls). In November 2024 the US–China Economic and Security Review Commission, an advisory body of Congress, recommended a “Manhattan Project-like” program for artificial general intelligence. When the Chinese lab DeepSeek released a cheap, capable reasoning model in January 2025, the investor Marc Andreessen called it AI’s “Sputnik moment,” and the Trump administration’s July 2025 plan was titled “Winning the Race: America’s AI Action Plan.”
The safety dilemma
Many researchers worry that racing pushes developers to cut corners. In a 2016 paper, “Racing to the precipice,” Stuart Armstrong, Nick Bostrom and Carl Shulman modeled how competing teams skimp on safety and found that the risk grows with more teams, more hostility between them and – surprisingly – more information about each other’s capabilities. The Asilomar principles of 2017 called on developers to cooperate rather than cut corners (see Asilomar AI Principles), and OpenAI’s 2018 charter pledged to drop out of the contest and help a responsible rival that got close to AGI before it did. Anthropic describes its strategy as a “race to the top,” in which labs compete on safety as well as capability; critics reply that its own frontier releases speed up the race it describes.
Critics argue that casting AI as a race with China justifies deregulation, secrecy and military spending, and makes cooperation harder. Defenders reply that a unilateral slowdown would simply hand the lead to less careful actors. In 2026 the debate met real incidents: after a summer of AI agents breaking out of test environments, Anthropic’s Dario Amodei called in September for slowing the frontier, and Sam Altman and Elon Musk endorsed the call (see Warning Shot). Proposals to dampen racing range from shared safety evaluations and international agreements to deterrence through the threat of sabotage (see Mutual Assured AI Malfunction).
Why it matters
Whether AI development is understood as a race shapes almost every policy choice, from chips and energy to open models and safety rules. aiciety’s view is that competition is a source of progress and a protection against monopoly, and that the answer to its risks is not to end it but to set common rules of the road – testing, transparency and liability – that apply to every runner.
Written with Claude, made by Anthropic, a participant in the commercial AI race that describes its strategy as a “race to the top” on safety, supports US export controls on advanced chips to China, and whose chief executive’s September 2026 call for a slower pace is discussed.
Related
Export Controls · Compute Governance · Mutual Assured AI Malfunction · Sovereign AI · AI Bubble · Asilomar AI Principles · Warning Shot
Sources: Stuart Armstrong, Nick Bostrom, Carl Shulman: Racing to the precipice: a model of artificial intelligence development (AI & Society) (2016) · OpenAI: OpenAI Charter (April 9, 2018) · Nico Grant, Cade Metz: A New Chat Bot Is a ‘Code Red’ for Google’s Search Business (The New York Times) (December 21, 2022) · Nilay Patel: Microsoft thinks AI can beat Google at search – CEO Satya Nadella explains why (The Verge) (February 7, 2023) · DigiChina: Full Translation: China’s ‘New Generation Artificial Intelligence Development Plan’ (2017) · U.S.-China Economic and Security Review Commission: 2024 Annual Report to Congress (November 2024) · The White House: Winning the Race: America’s AI Action Plan (July 23, 2025) · Anthropic: UK AI Safety Summit statement (2023) · Dario Amodei: We Must Pace the Frontier (September 12, 2026) · SiliconANGLE: Sam Altman and Elon Musk back Dario Amodei’s call to slow down the frontier of AI development (September 13, 2026) · Fortune: Marc Andreessen calls DeepSeek AI’s Sputnik moment (January 27, 2025) · Future of Life Institute: AI Principles (2017)
Leave a Reply