AI Safety Warnings Collide With Geopolitical Race for Dominance

A wave of stark warnings about the dangers of advanced artificial intelligence has run headlong into a geopolitical contest that appears to make any coordinated slowdown improbable.

Over the weekend, concerns voiced by leading figures in the technology industry reached a crescendo. A departing employee at AI lab Anthropic warned that the technology could pose an existential threat by the end of the decade. By Sunday, a broad consensus had emerged among Anthropic's Dario Amodei, OpenAI's Sam Altman, Elon Musk of SpaceX, Apple's Tim Cook and others on the need to slow development, increase human oversight and set clear limits on the autonomy of AI agents. Microsoft published its own draft code of conduct.

The warnings included indications that the most sophisticated models could be used to develop biological weapons, hack financial systems or even seize control of the internet.

Yet the response from major governments suggests that such cautions may only intensify the race. Both Washington and Beijing signalled on Monday that they intend to press ahead with AI development, viewing the capability as essential to national security and economic competitiveness.

US President Donald Trump dismissed the safety push, writing on social media that a "SICK conspiracy" against AI and data centres was underway and that only China would benefit. An editorial in China's state-backed Global Times characterised the warnings as a "Cold War playbook" aimed at China. Trump is scheduled to meet Chinese President Xi Jinping next week in Washington, with lower-level officials previously reported to be planning talks on AI risks.

Even if the two powers agreed on protocols, other countries or rivals would not necessarily be bound by them. The history of nuclear non-proliferation offers a parallel: 23 years passed between the first atomic weapon and the 1968 treaty, and the number of nuclear-armed states has since grown from five to nine.

Financial markets are weighing two interpretations. One dismisses the scare stories as hype ahead of upcoming initial public offerings. The other holds that AI development is now too strategically important to fail, leaving governments no choice but to proceed.

Materially, Altman said OpenAI's long-awaited IPO would likely be delayed to next year, news that knocked shares in stakeholder SoftBank by about 13 per cent on Monday. Anthropic still appears to be proceeding with a listing as soon as next month.

The bigger question for investors is whether calls for a slowdown will affect the frenzied investment in data centres and computing capacity. Goldman Sachs expects cumulative global AI spending through the end of the decade to reach as much as 5 per cent of global GDP. Much of that is funded by debt and equity, and the timing of the eventual payoff remains uncertain.

Morgan Stanley strategists estimated that almost $200 billion of data centre projects were cancelled or delayed last year and through the first quarter of 2026 amid rising domestic political opposition in the US. The bank noted that AI-related capital expenditure is set to account for roughly one-third of all US GDP growth this year and next. It also pointed out that demand for computing power still far exceeds supply, meaning any slowdown in the buildout could strengthen the position of hyperscalers by increasing the scarcity value of their installed base and enhancing their pricing power.

Ultimately, go-slows would only significantly alter the long-term direction of the AI boom if there were suggestions that development could be reversed or stalled. The political imperative to press ahead, made clear by both Washington and Beijing, indicates that governments will seek to drive the capability to its fullest extent — precisely because of, not despite, the risks now being flagged.