Anthropic has released a detailed paper that outlines how the balance of power in artificial intelligence could shift by 2028, warning that the United States and its allies face a limited window to secure a lasting advantage over China.
The company argues that AI is approaching a stage where it can reshape global power structures and enable large-scale societal control. It states that decisions made in the next few years will determine whether democratic systems or authoritarian regimes set the rules for how the technology develops and operates.
We've published a paper that explains our views on AI competition between the US and China.
— Anthropic (@AnthropicAI) May 14, 2026
The US and democratic allies hold the lead in frontier AI today. Read more on what it’ll take to keep that lead: https://t.co/TgJBeodWYK
Compute controls remain central to US advantage
At the core of Anthropic’s argument lies access to advanced semiconductors, often referred to as “compute.” These chips power the training and deployment of advanced AI models. The company states that the US currently holds a significant lead due to domestic innovation and export controls that restrict China’s access.
Those controls, first introduced in 2022 and expanded under successive administrations, limit the sale of high-end chips from companies such as Nvidia and AMD. However, Anthropic notes that these measures have not fully prevented Chinese firms from accessing compute.
Chinese AI labs continue to operate close to the frontier by exploiting loopholes. The paper highlights two main methods: illicit chip acquisition, including smuggling and offshore data center usage, and “distillation attacks,” which involve extracting knowledge from advanced US models to train domestic systems.
Anthropic writes that “if the US and its allies act now to address both issues, it may be possible to lock in a 12-24 month lead in frontier capabilities,” but adds that “the window of opportunity to lock in that lead will not necessarily remain open for long.”
Two scenarios for 2028
The paper presents two contrasting scenarios for how the global AI landscape could look by 2028.
In the first scenario, US policymakers tighten export controls, close enforcement gaps, and curb distillation practices. This leads to a sustained advantage, with American AI systems remaining 12 to 24 months ahead of Chinese counterparts. In this case, democratic nations shape global standards, and AI adoption accelerates across industries such as healthcare, cybersecurity, and finance.
Anthropic states that such a lead would also improve the chances of meaningful cooperation with China on AI safety, as a clear capability gap reduces competitive pressure.
In the second scenario, the US fails to act decisively. Export control loopholes persist, and Chinese firms continue to access advanced chips and models. As a result, China reaches near parity with US systems.
This outcome carries broader implications. The company warns that a “neck-and-neck race between American and Chinese AI labs could make industry and government-led safety and governance efforts more difficult.” It adds that intense competition could push companies to release systems faster, without adequate safety testing.
Concerns over authoritarian use of AI
Anthropic raises concerns about how advanced AI could be used under authoritarian systems. It points to existing uses of AI in China, including censorship, surveillance, and cyber operations, as early indicators of how more powerful systems might be deployed.
The company states that future AI systems could automate repression at scale, reducing reliance on human enforcement. It also highlights the role of AI in military applications, noting that systems can be deployed quickly once developed, especially in areas such as cyber warfare and autonomous coordination.
The report references the People’s Liberation Army’s focus on “intelligentization,” which aims to integrate AI into military strategy. It also cites the use of AI models in cyber operations and unmanned systems.
Ongoing policy debate in Washington
Anthropic’s recommendations align with ongoing policy discussions in the US. The company calls for stricter enforcement of chip export controls, increased oversight of overseas data center access, and legal clarification around distillation attacks.
Recent enforcement actions reflect these concerns. US authorities have charged individuals involved in attempts to smuggle advanced chips into China, sometimes by mislabeling hardware. At the same time, debates continue over how strict export policies should be, especially after partial rollbacks that allowed limited chip sales under specific conditions.
The company also emphasizes the importance of promoting global adoption of US-built AI systems. It argues that widespread deployment of American technology can limit the influence of alternative AI ecosystems in international markets.
Diverging views on China’s AI progress
While Anthropic argues that China is closing the gap through workarounds, some voices within the Chinese tech community present a different view. Zhang Chi, a former ByteDance engineer and now a researcher at Peking University, said in April that Chinese AI development faces constraints due to limited access to high-quality data and advanced chips.
These differing perspectives reflect uncertainty around the true state of the competition. However, Anthropic maintains that even a small gap could have significant consequences, especially as AI capabilities accelerate.
A narrow window for action
Anthropic’s paper frames 2026 as a critical turning point. The company highlights rapid progress in AI capabilities and points to recent internal developments as evidence of accelerating change.
It argues that the US and its allies already hold structural advantages, including stronger chip supply chains, greater capital resources, and leading AI firms. The challenge, according to the report, lies in maintaining that position.
“Our past success means that our present task is largely to avoid squandering our advantage: to decide not to make it easier for the CCP to catch up,” the company wrote.
Gunel Ismayilova
Disclaimer: All materials on this site are for informational purposes only. None of the material should be interpreted as investment advice. Please note that, despite the nature of much of the material created and hosted on this website, HODL FM operates as a media and informational platform, not a provider of financial advisory services. The opinions of authors and other contributors are their own and should not be taken as financial advice. If you require advice, HODL FM strongly recommends contacting a qualified industry professional.
