4 hours ago
4
OPINION — Washington’s strategy for artificial general intelligence (AGI), or the ability to replace human cognitive labor, assumes the United States is locked in a decisive race with Beijing—one requiring maximum acceleration and denial of Beijing’s access to semiconductor chips and technology. This approach, as captured in the White House’s AI Action Plan from last year, echoes the race in the 1940s to build the atomic bomb and during the Cold War to dominate space. It risks refighting the Cold War, which is ill-equipped for a technology-based struggle. This posture misdiagnoses the nature of the AI competition and risks degrading, rather than strengthening, America’s long-term strategic position. It also has a sizable blind spot: dealing with an inevitable dislocation in the global workforce.
Presidents Trump and Xi have an opportunity to reset the terms of this competition over AI when they meet next month.
The current U.S. AI strategy amounts to a wartime footing defined by denial and containment of competitor capabilities, hundreds of billions in capital expenditure in AI capabilities, and expansive export controls of diminishing effectiveness. But AI is not a binary capability—either you have it or you don’t. It is a continuous, evolutionary technology with no single threshold that confers decisive, let alone permanent advantage. Our national workforce policies have remained remarkably stable so far, though AI is but one of many emerging technologies that may upend the global economy for which the U.S. is well positioned.
The international AGI ecosystem is rapidly evolving with many competitors entering, replicating others’ advances, and exiting to pursue niche applications. It was once assumed the U.S. held a year-plus advantage over China in frontier AI models. That gap has dwindled to 2-3 months, despite stringent export controls. Even if these controls have slowed China’s training on new frontier models, they have not dampened China’s advantages in AI deployment and diffusion. China’s AI influence on the global stage has only grown, aided by increasingly capable models, dramatically cheaper end-user pricing, and leverage of the global open-source developer community.
The economic advantage from AI does not stem from being first to develop frontier models, but from being first to diffuse capabilities across industries and scale across the economy. China rarely competes on frontier quality (it prefers being “good enough”), but on quantity, price, time to market, and speed to dominate supply chains. In this race, China is likely outpacing us. ByteDance’s Doubao chatbot exceeded 100 million daily active users. Alibaba’s Qwen models have surpassed 700 million downloads globally, spawning 180,000+ derivative models. Chinese open-source models are fast becoming the de facto platform for sovereign AI efforts across the Global South and startup companies globally (even in the U.S.).
China leads in 66 of 74 critical technologies tracked by the Australian Strategic Policy Institute, accounts for 54% of global industrial robot installations (International Federation of Robotics, 2024), produces about half of the world’s AI researchers, and builds more new electricity capacity annually than the rest of the world combined. These are the foundations of AI deployment at scale; denying chips won’t offset these structural advantages.
Washington often perceives the Chinese AI effort as a state-directed monolith. The reality is a fiercely competitive and innovative commercial ecosystem with creative business models. ByteDance’s Doubao is a closed-source consumer product fighting for domestic market share. Zhipu AI generates over 60% of revenue from enterprise deployment services. MiniMax earns roughly 70% of revenue from international API sales. Alibaba open-sourced Qwen to drive cloud adoption; DeepSeek did so to attract research talent. Framing this diverse, commercially motivated ecosystem as a centrally planned strategic threat produces policy responses that are either too blunt—restricting all Chinese AI—or too narrow, focused on chip exports while ignoring the deployment gap (how models are trained and used in practice).
The U.S. is now chasing artificial “superintelligence” (ASI) in pursuit of permanent dominance, relying on chaotic and unsustainable private investment. Meanwhile, China is building the industrial AI infrastructure with a consistent regulatory approach that will shape how roughly 150 countries deploy this technology for decades.
The consequences of this mismatch are profound. U.S. technology firms have committed over $500 billion annually in AI capital expenditures for 2025–2027, while job openings in the U.S. have declined sharply. Data from the World Bank indicate 60% of the U.S. workforce is at risk of being displaced due to AI without a compensatory social safety net.
The impact in the defense sector is similar. Proponents of the current posture often argue that if China gets AGI first, they’ll weaponize it. But the US military does not need the latest or the best frontier model. It needs models that are fit to task—certified, tested, and integrated into operational systems.
The decisive military advantage may lie less in which country trains the most capable model than in who can field AI-enabled systems fastest across its force. By that metric, the current U.S. acquisition system is at a structural disadvantage. The U.S. military’s vendor and model certification process can take over a year. The Chinese government reviews AI models even before their public releases to streamline their deployment.
AI does pose genuine security risks. AI-enabled cyber weapons, the proliferation of autonomous weapons, and malicious use of AI by bad actors all pose significant hazards. But these threats are best addressed through narrowly scoped controls and shared intelligence with key allies (Australia, Japan, the European Union, and South Korea) to provide safety standards and semiconductor supply chain resilience. This strategy should address misuse of AI by malicious actors, potential instability from mass displacement of workers, undue market concentration. and inadvertent military escalation. Washington should take a posture of allied industrial policy for AI diffusion, targeted safety agreements with enforcement mechanisms, and serious domestic investment in workforce transition. The precedent to replicate is not the Manhattan Project that sought first deployment of nuclear weapons, but Cold War arms control agreements that stabilized relations with the Soviets and allowed the U.S. economy to boom.
Additionally, we must enable a soft landing for the looming workforce displaced by AI. We should be creating workforce legislation modeled on the post-WWII GI Bill and educational, housing, and living assistance programs to help the economy adapt. We should be building with likeminded global partners an architecture that nurtures international AI standards, polices compliance, and provides guardrails for open-source AI capabilities for civil applications.
If Washington continues fighting the last war as AI’s promise matures, it may win battles over benchmarks and chips ,but lose the campaign that actually matters—safely diffusing AI to remake the global economy for the next century. Rethinking the parameters of today’s competition is the first step to ensuring AI strengthens rather than erodes American security and prosperity.
The Cipher Brief is committed to publishing a range of perspectives on national security issues submitted by deeply experienced national security professionals. Opinions expressed are those of the author and do not represent the views or opinions of The Cipher Brief.
Have a perspective to share based on your experience in the national security field? Send it to Editor@thecipherbrief.com for publication consideration.
Read more expert-driven national security insights, perspective and analysis in The Cipher Brief
