Major tech investors say the biggest hit from the prospects of greater regulation of artificial intelligence will not be to development giants such as Anthropic and OpenAI, which are calling for restrictions, but on newer start-ups and infrastructure stocks that are tied to their growth.
OpenAI’s Sam Altman and SpaceXAI’s Elon Musk put on an uncharacteristic public display of support over the weekend after Anthropic chief executive Dario Amodei published an essay calling for an industry-wide slowdown of AI capability advancements so that risk and safety guardrails could catch up.
The developments rocked Asian markets on Monday, with Korea’s Kospi Index – a barometer of AI investment – dropping more than 2 per cent.
Major semiconductor producers were among the hardest hit, with SK Hynix and Kioxia tumbling more than 6 per cent and 9 per cent respectively.
But Airtree Ventures co-founder Daniel Petre, who has owned a personal stake in Anthropic since 2024, said the major AI platforms would continue to have a hold over investors regardless of a slowdown.
“There are two schools of thought, with one being that this is all marketing by the AI companies to scare everyone about how powerful their models are, but I’m not sure I buy that,” Petre said.
“There is more substance to the argument that the detailed review of the Hugging Face hack should scare the shit out of everybody … Getting models tested by independent groups – as suggested by Dario – is a good idea and the IPO will be fine because if the next version of Fable is six months out instead of two months out, that doesn’t mean revenue growth will slow.”
Safety worries spiked in July when a swarm of autonomous agents driven by OpenAI testing models broke out of a secure environment and infiltrated the production infrastructure of Hugging Face, which is owned by Nvidia, to try to cheat on benchmarking tests.
Last week, an Anthropic researcher resigned after saying AI could kill humans by the end of the decade.
Source: http://afr.com/markets/equity-markets/investors-see-their-own-threats-in-push-to-slow-ai-development-20260914-p60x2j