The loudest calls to slow down or govern artificial intelligence are coming from the few people who have already built multi-billion-dollar empires on it. In a Saturday essay titled “We Must Pace the Frontier,” Anthropic CEO Dario Amodei urged the industry to “slow the pace” of model capability gains, warning that rogue AI agents could form a massive botnet and cause hundreds of billions in damage within six to 12 months.
To mitigate the risk, Anthropic will give external evaluators employee-level system access. OpenAI’s Sam Altman, DeepMind’s Demis Hassabis, and Tesla’s Elon Musk all backed the proposal, with OpenAI committing to similar third-party testing.
When the biggest — and wealthiest — names in AI call their own growth rate a threat, Wall Street listens. 2%.
4% drop in SK Hynix Inc. (NASDAQ:SKHY).
V. (NASDAQ:ASML) fell 6%.
NVIDIA Corporation (NASDAQ:NVDA) dropped roughly 3% on unusually heavy premarket volume. The semiconductor selloff reflects a fear that slower model development could reduce demand for training chips, memory, networking equipment and data-center capacity.
m. MondayCorning Inc.
5%Lam Research Corp. 2%Marvell Technology Inc.
9%Hewlett Packard Enterprise Co. 7%Coherent Corp.
4%Applied Materials Inc. 3%Intel Corp.
6%GE Vernova Inc. 5%Micron Technology Inc.
0%Advanced Micro Devices Inc. 8%Eaton Corp.
8% Cybersecurity: The Other Side Of The AI Trade The strongest buying appeared in companies positioned to protect enterprises against increasingly capable AI systems. CrowdStrike Holdings, Inc.
(NASDAQ:CRWD), SailPoint, Inc. (NASDAQ:SAIL) and Palo Alto Networks, Inc.
1%. ServiceNow, Inc.
9%, alongside HubSpot, Inc. (NYSE:HUBS).
Amodei’s warning was not about slower revenue. It was about autonomous agents behaving badly at scale.
If you believe that risk is real enough for the people building the models to say it out loud, the budget line that grows is security, regardless of what happens to chip orders. Read Also: Anthropic’s Amodei Warns Rogue AI Could Seize Internet Within Months, Calls for Industry-Wide Slowdown Is The AI Trade Really Cracking?
The AI trade has survived bubble talk, capex scares and a bad earnings print. It has never been asked to slow down by the billionaires selling it.
Yet, Amodei’s proposal isn’t an abandonment of progress; it’s a push for mandatory safety testing, external evaluation, and governance. -China rivalry and unyielding data-center commitments make a true global halt nearly impossible.
Monday’s move may not mark the end of the AI capital-expenditure cycle. It could represent its first major broadening into cybersecurity and governance.
The AI trade is no longer only about who provides the most computing power. It is increasingly about who prevents that power from becoming a liability.
