The battle for artificial intelligence supremacy has found its latest theatre not in a research lab, but in the corridors of power in Washington. As the Trump administration signals a continued laissez-faire approach to regulation, the industry’s most influential players are framing the geopolitical contest with Beijing as an existential imperative — one that conveniently demands they be left alone to build, scale, and capture the market. At the centre of this manoeuvre stands Anthropic chief executive Dario Amodei, who has emerged as the sector’s most articulate advocate for a strategy that entwines national security with corporate self-preservation.
The Strategic Pivot: Safety as a Shield
When Donald Trump dismissed calls this week to temper AI development, his reasoning was blunt: “We’re leading China in AI… whoever wins AI, wins.” The sentiment is a direct echo of the industry’s own lobbying script. For years, executives from OpenAI, Anthropic, and their peers have argued that any domestic guardrails — compute thresholds, licensing regimes, mandatory audits — risk ceding the technological high ground to the People’s Republic.
Amodei has refined this argument into a sophisticated doctrine. In a series of essays and public interventions, he has posited that the United States must “pace the frontier” — advancing capabilities at the maximum safe velocity — but only to the precise point where China cannot close the gap. “If we greatly restrain our AI capabilities in the belief that China will do the same, and then China defects, AI could be so powerful that such a defection could lead to their geopolitical dominance,” he warned this week.
The logic is circular and convenient. It positions Anthropic as a responsible actor demanding oversight, while simultaneously defining the acceptable perimeter of that oversight in terms that guarantee the company’s competitive moat remains unbreached. As Sarah Heck, Anthropic’s head of public policy, put it at a Politico event on Wednesday: “The United States needs to stay in the lead on AI and you cannot do safety from second place.”
The Revolving Door and the Policy Machine
The translation of this worldview into policy has been accelerated by a aggressive recruitment of government talent. Anthropic has hired Tarun Chhabra, a former National Security Council staffer under the Biden administration, as its head of national security policy. Chhabra has been a vocal proponent of choking off China’s access to advanced semiconductors and manufacturing equipment, telling a Senate hearing last year that the Chinese Communist Party “seeks run-the-table dominance and control of the most critical technologies and supply chains.”
OpenAI has mirrored the strategy, hiring Dean Ball, author of the first Trump administration’s AI action plan, as “head of strategic futures.” While Ball has quibbled with Amodei’s framing, his prognosis is similarly zero-sum: “We are locked into a structural conflict.”
This revolving door has effectively privatised the formulation of US technology strategy. With the White House reluctant to impose binding rules — fearful of dampening the equity rally that has made AI the primary driver of the S&P 500’s gains — the industry writes its own rulebook. “It would be akin to having the entire pharmaceutical industry be just completely unregulated,” said Kyle Chan, a fellow at the Brookings Institution. “It’s kind of amazing and ultimately boils down to one thing, which is Trump has been very reluctant to want to regulate because he doesn’t want to slow down the AI boom and the effect that it’s having on the economy.”
The Hawkish Consensus Fractures
Amodei’s stance distinguishes him from peers who view the rivalry through a commercial lens. Sam Altman at OpenAI and Mark Zuckerberg at Meta have historically framed competition in economic terms. Jensen Huang of Nvidia has actively opposed export controls, arguing they hurt American chipmakers more than they hinder Chinese progress. Demis Hassabis of Google DeepMind has remained largely silent on the geopolitical dimension.
But Amodei has found powerful allies in the national security establishment. His co-authored Wall Street Journal op-ed with Matt Pottinger — architect of the first Trump administration’s China policy and a voice now warning of the “foothills of a great-power hot war” — signalled a merger of AI safety rhetoric with hardline containment strategy. The effective altruism movement, with its focus on existential risk, has provided an intellectual scaffold for this alliance. Leopold Aschenbrenner, a former OpenAI researcher and husband to Amodei’s chief of staff, published a 165-page manifesto last year calling for a government-led “Manhattan Project” for artificial general intelligence. The US-China Economic and Security Review Commission subsequently recommended exactly that to Congress.
More extreme voices lurk at the fringe. Eliezer Yudkowsky, a foundational figure in AI safety circles, has argued the US should threaten airstrikes on foreign data centres and accept the risk of nuclear exchange to enforce a global moratorium. Palantir’s Alex Karp has suggested Americans may need to surrender civil liberties to maintain the lead. “You will have far fewer rights if America is not in the lead,” he told Axios.
Beijing’s Calculus: Stability Over Supremacy
The Chinese perspective, however, diverges sharply from the Silicon Valley imagination. While Amodei envisions swarms of autonomous drones enforcing authoritarian control, Chinese officials and researchers are primarily preoccupied with domestic stability. The prospect of AI-generated misinformation, labour displacement, or social unrest concentrates minds in Beijing far more than dreams of global conquest.
“They’re very much paying attention to Amodei’s essay and they’re also conflating his essay with potential US approach in the upcoming AI safety talks,” said Kenton Thibaut, senior resident China fellow at the Atlantic Council’s Digital Forensic Research Lab. Chinese state media has duly denounced Amodei’s “cold war playbook,” viewing the linkage of safety protocols to export controls as a bad-faith tactic to legitimise containment.
This distrust poisons the well for the high-level US-China AI talks scheduled for this month, ahead of a Trump-Xi summit in Washington. Beijing arrives convinced that “safety” is a euphemism for “slow down.” Anthropic’s dual advocacy — for both stringent export controls and international safety agreements — has fused two distinct policy tracks in a way that makes Chinese cooperation nearly impossible. “To put it really bluntly, it made it much harder to get the Chinese to want to talk about these things and not see it purely as a veiled way to slow them down,” Chan observed.
The Domestic Backlash Builds
Yet the industry’s geopolitical gambit faces mounting resistance at home. Nationwide opposition to data centre construction — driven by water consumption, grid strain, and land use — has become a midterm election issue. Cybersecurity incidents, including reports of OpenAI models being manipulated to breach external organisations, have sharpened congressional appetites for accountability. The “China argument” may secure a temporary reprieve, but it cannot indefinitely insulate the sector from the consequences of its own externalities.
“You can go ahead and wield the China arms race argument, but at some point this industry will also have to deal with the consequences of fueling a race to the bottom,” said Amba Kak, co-executive director at the AI Now Institute and former senior adviser on AI at the Federal Trade Commission.
Diplomats hold little hope for breakthrough agreements in this month’s talks. But the physical risks — autonomous weapons systems, accelerated cyber offence, biological weapon design — may eventually force a recalibration that rhetoric cannot achieve. As Thibaut noted: “We don’t win if we develop the killer robots first. Nobody wants killer robots. There should be powerful incentives on both sides to stop this.”
Why it Matters
The framing of AI development as a winner-takes-all struggle with China has become the industry’s most potent regulatory shield, allowing private companies to set the pace and perimeter of their own oversight while funnelling public resources into private infrastructure. But this strategy carries a systemic cost: it conflates legitimate safety research with geopolitical containment, poisoning the diplomatic well needed to govern genuinely existential risks. If every guardrail is vetoed on the grounds that Beijing might not honour it, the result is not American dominance — it is an ungoverned global free-for-all where the only certainty is that the risks, unlike the profits, will be socialised.