LONDON, Sept 14 – Global artificial intelligence stocks came under pressure on Monday after Anthropic CEO Dario Amodei called for a slower pace of AI capability development, intensifying concerns among investors over the potential impact on chip demand, computing infrastructure and future technology spending.
The sell-off spread across major AI and semiconductor companies, with investors weighing whether a slower development cycle could eventually affect the enormous capital investments supporting the AI industry.
In Asia, SK Hynix and Samsung Electronics fell more than 6% and 4%, respectively, while SoftBank, one of the largest investors in OpenAI, dropped about 10% in Japan.
European technology and semiconductor stocks also declined sharply in early trading. ASML fell more than 4%, while Nokia declined around 5% and Infineon dropped more than 6%. Companies supplying equipment and infrastructure for data centres, including Siemens Energy and Schneider Electric, also traded lower.
The pressure extended to U.S. markets before the open, with Micron down about 5%, Intel nearly 6% lower and Nvidia falling more than 2%. Microsoft, Amazon and Alphabet also edged lower.
AI Safety Debate Intensifies
The market reaction follows a growing debate within the AI industry over the risks associated with increasingly capable models.
The discussion intensified last week after Jacob Coxon, an Anthropic researcher who previously worked at OpenAI, resigned and said the companies were “gambling with our lives.”
Anthropic safety researcher Evan Hubinger subsequently said he believed there was a greater than 10% probability that AI could “kill all humans” within the next decade. The comments generated significant discussion across social media and prompted responses from prominent figures in the technology industry.
Amodei subsequently published an essay calling for the industry to reduce the speed at which frontier AI capabilities are developed.
“We must slow the pace at which we improve the capabilities of AI models,” Amodei said.
However, the Anthropic chief did not call for AI development to stop altogether, arguing that progress could remain rapid even under a slower development approach.
His position received support from several major technology leaders. OpenAI CEO Sam Altman said he agreed that the industry needs to “pace the frontier,” while SpaceX CEO Elon Musk responded on X: “Dario is right.”
Altman later clarified that pacing development does not amount to stopping it.
“Pacing” does “not mean ‘stopping’,” Altman said on Monday.
“Progress has been rapid and will continue to be. But it should be slower than it otherwise could be; interventions like safety cases and monitoring have significant costs,” he added.
Investors Assess AI Spending Risks
The debate has direct implications for the financial markets because the AI investment boom has driven enormous spending on semiconductors, data centres and computing capacity.
A sustained slowdown in AI development could potentially reduce the pace at which technology companies purchase advanced chips and computing infrastructure, while also affecting expectations for future earnings growth across the sector.
“The kind of equity market rally has been based on AI growth and productivity gains … so if we do see that start to derail, then it could have an impact on equity performance going forward,” Zoe Gillespie, senior director at RBC Brewin Dolphin, told CNBC.
“Certainly, a lot of what we are looking into with equity returns is baked into the future earnings growth of these companies, and if that comes under threat then we may see this destabilize,” she added.
Other analysts argue that a moderation in frontier-model development would not necessarily translate into an immediate collapse in AI-related demand.
Ben Barringer, global head of technology research at Quilter Cheviot, said the pace of technological change could remain substantial even if AI training and deployment were slowed.
“While things may slow somewhat, the pace of change is still going to be vast,” Barringer told CNBC.
“Even if training and rollout is slowed, inference is still the area that the industry is short in supply. Demand still far outstrips supply, so even if things are to slow a little, company revenues are unlikely to be impacted,” he said.
Inference refers to the process of running AI models after they have been trained, while training involves using large quantities of data and computing power to improve the underlying models.
The contrasting views highlight a key question for markets: whether growing calls for stronger AI safety measures will materially slow the enormous investment cycle behind the technology, or whether demand for computing and AI applications will continue to expand even as frontier-model development becomes more cautious.
