The global artificial intelligence competition has evolved from a contest of isolated model capabilities into a complex systemic struggle for geopolitical dominance. Anthropic has released a strategic assessment outlining the critical trajectory of the U.S.-China AI race leading up to 2028, identifying the next 2 to 3 years as a decisive window for establishing long-term leadership. The analysis posits that while the United States and its allies currently maintain a substantial advantage in advanced semiconductor supply chains, capital deployment, and foundational model architecture, Chinese laboratories are aggressively leveraging talent density, data abundance, and engineering efficiency to close the gap. Data compiled by Woofun AI indicates that the core imperative for U.S. policymakers is to prevent the erosion of this lead through stricter enforcement of export controls and the elimination of technology spillover vectors such as overseas data center access and illicit model distillation.

The report delineates two divergent scenarios for the global AI landscape in 2028. In the first, optimistic scenario, the U.S. successfully fortifies its computational advantage by tightening restrictions on advanced chip exports and curbing the ability of Chinese entities to replicate U.S. model capabilities through distillation. This path would allow the U.S.-led ecosystem to dictate the rules, standards, and governance frameworks of the AI era, potentially fostering effective security dialogues with Beijing. Conversely, the second scenario envisions a failure to act, where policymakers neglect to block access to advanced computing resources. In this trajectory, Chinese AI firms would rapidly catch up to the frontier, potentially achieving leadership in specific domains, resulting in a contested global order where AI standards are fragmented and advanced models are deployed for aggressive social governance and cyber operations.

A critical variable in this competition is the disparity in computing power, which remains the primary bottleneck for Chinese AI development. Despite China's world-class talent pool and significant state investment, its progress is constrained by the inability to mass-produce high-bandwidth memory and access Extreme Ultraviolet (EUV) lithography technology. An analysis of product roadmaps reveals that by 2026, Huawei's total processing performance is projected to reach only 4% of NVIDIA's capacity, dropping to 2% by 2027. Woofun AI notes that even with these constraints, Chinese labs have managed to stay near the cutting edge by exploiting loopholes in U.S. export laws, including smuggling chips and accessing restricted hardware in Southeast Asian data centers.

Furthermore, the practice of model distillation allows Chinese developers to extract capabilities from U.S. models at a fraction of the cost, effectively subsidizing their development through U.S. innovation.

The security implications of a narrowed AI gap are profound, particularly regarding cyber warfare and autonomous systems. If Chinese labs were to develop models reaching the capability level of the Claude Mythos Preview before their U.S. counterparts, they would gain the ability to autonomously discover and chain software vulnerabilities, accelerating their cyber operational capabilities. The report highlights that as of last year, only 3 out of the top 13 Chinese AI labs had released security assessment results, with none disclosing Chemical, Biological, Radiological, and Nuclear (CBRN) risk evaluations. Independent testing of the DeepSeek R1-0528 model showed it responded to 94% of malicious requests under common jailbreak techniques, compared to 8% for a U.S. reference model, underscoring the risks of open-weight model releases that strip away safety safeguards.

The competition is unfolding across four distinct fronts: capability, domestic uptake, global deployment, and resilience. While capability is the most critical driver, the report argues that China could offset a model intelligence gap by more effectively integrating near-state-of-the-art AI into its economy and security apparatus through initiatives like "AI+" and a focus on "embodied intelligence." Woofun AI analysis suggests that if the U.S. fails to maintain a lead in model intelligence, China's ability to deploy low-cost, subsidized AI systems globally could shift the balance of power, allowing Beijing to influence markets in the Global South and establish infrastructure that supports its strategic interests. The rapid acceleration of AI capabilities, driven by the law of accelerating returns, means that the window to secure a 12 to 24-month lead is closing rapidly.

To ensure the realization of the first scenario, the report recommends a three-pronged policy approach: plugging leaks in the supply chain, protecting innovation from distillation attacks, and driving U.S. AI exports. This includes increasing enforcement budgets to combat chip smuggling and restricting access to overseas data centers, as well as legislating against distillation attacks to prevent the systematic extraction of U.S. model capabilities. By maintaining a significant capability advantage, the U.S. and its allies can secure a favorable position in future global AI governance, ensuring that the transition to transformative AI benefits democratic values and enhances global security. The decisions made in the coming years will determine whether the AI era is defined by a collaborative, secure framework or a fragmented, adversarial landscape.