Key takeaways

  • Chip and memory stocks sold off over expectations that safety concerns could lead to a slowdown in AI investment.
  • The selloff comes after Anthropic CEO Dario Amodei called for a slowing in the pace of development and increasing AI safety controls, receiving backing from OpenAI and SpaceX.
  • Analysts suggest increased AI guardrails need not significantly derail the AI buildout.

AI hardware stocks took a hit last week after pioneers of the burgeoning technology called for a slowdown in its development amid rising safety concerns.

Chip and memory stocks saw some of the sharpest declines as investors fretted over how a pullback could hamper a recent surge in infrastructure investment. In the US, shares of Intel INTC, Micron MU, AMD AMD all fell about 5%. In Europe, ASM International ASM dropped nearly 9%, while Infineon IFX and BE Semiconductors BESI fell almost 8%. Meanwhile in Asia, South Korean memory chip heavyweights SK Hynix 000660 and Samsung 005930 closed down 4% and 6%, respectively.

Software stocks, previously seen as among those most vulnerable to AI disruption, rallied in tandem on hopes of a slower-than-feared AI buildout. CrowdStrike CRWD added 15%, and ServiceNow NOW gained 6%, with SAP SAP, RELX REL, and Wolters Kluwer WKL rising similarly in Europe.

The selloff comes after Anthropic CEO Dario Amodei on Saturday published a blog post calling for AI developers to slow the pace of development, warning that the tech’s growth is rapidly surpassing safety controls. His views were broadly backed by SpaceX’s SPCX Elon Musk and OpenAI’s Sam Altman. Altman said the company was holding off on its much-anticipated IPO over the weekend, saying it would not happen this year due to safety risks.

SpaceX rebounded from this morning’s decline to trade 0.4% higher Monday, while Microsoft, which owns a large stake in OpenAI, rose 2.5% after it posted a provisional code of conduct that would impose restrictions on its AI models.

Last week’s trading “marks a change for markets, with the negative on software, positive on hardware, trade being turned on its head,” says Morningstar strategist Michael Field, “on the back of warnings that AI development might need to slow down, which the market is viewing as a negative for hardware firms, as it means the buildout may also slow. For software firms that many feared would be disrupted, this means a stay of execution.”

The growing fears around AI safety come as some investors are already skittish about the global macroeconomic outlook. Higher oil prices have revived inflation concerns, pushing up global bond yields and bolstering expectations of an interest rate hike ahead of the Federal Reserve’s meeting on Wednesday.

Neil Wilson, Saxo UK investor strategist, says last week’s selloff suggests that market fears could now shift to equity markets, and whether slower and lower capex spending will weaken global growth and corporate earnings. “The nexus of worry has been oil, but now AI frontier development slowdown risks threaten investment cases and will have analysts scrabbling around to assess likely impact on earnings and valuations if the likes of Anthropic, OpenAI, SpaceX, Google, and Meta coordinate a material slowing in development and erect guardrails.”

Will there be an AI investment slowdown?

Investors have been confronting growing questions over the pace and extent of AI development even as it drives record returns. Shares of Samsung, SK Hynix, Intel, and Micron are among those up 100% or more so far this year, while global AI capex is seen topping $1 trillion in 2026, according to Goldman Sachs.

The debate hit a groundswell last week after Anthropic researcher Jacob Coxon resigned, warning that the race for AI supremacy could destroy humanity by the end of the decade. It follows several security breaches by sophisticated AI models circumventing their developers’ controls. US President Donald Trump has nevertheless urged firms to push ahead with AI progress to fend off Chinese competition. China has said that AI should be used for the advancement of all nations.

Even with increased AI guardrails, analysts say they don’t see AI investment slowing in the near term, particularly as firms require more compute to iterate models according to new guidelines. “Stronger safeguards need not imply the end of the AI capex cycle, in our view, with model training continuing, and expanding adoption requiring higher computing capacity,” Mark Haefele, chief investment officer of UBS Wealth Management, wrote in a client note last Monday.

“More testing does not mean technology companies will suddenly stop building data centers or buying computing equipment. Many projects are already underway, while testing advanced models also requires substantial computing power,” adds Charu Chanana, chief investment strategist at Saxo.

Tighter safety controls could, however, favor larger, incumbent providers that are “better equipped to meet oversight requirements,” according to Haefele. He notes that UBS retains its AI capex forecast of $1.2 trillion in 2027, with the bank favouring diversified AI exposure across semiconductors, networking, power, and cloud infrastructure, alongside select platforms and software companies positioned to “monetise AI adoption.”

Morningstar’s Field says that despite the news, “We don’t expect this to be a lasting theme. AI development slowing down is a sign that the potential for the technology is even more powerful than the market had given it credit for, which over time could lead to share price rises. For software stocks, we believe many of these are undervalued, so today’s rally is much welcomed.”

Bank of America semiconductor equity analyst Vivek Arya said in a note that investors should tune out the “noise” driving last Monday’s rout and focus on the market forces that he says could cause AI capex to triple to $3 trillion by 2030. “Demand signals remain robust: 100% network utilization, rising rental rates even for older-gen chips, and a global AI arms race,” he wrote. The federal government, for which AI has become a strategic priority, is unlikely to intervene, he wrote. “We therefore expect any eventual outcome to resemble industry-led self-regulation,” wrote Arya, “rather than intervention capable of materially slowing deployment or AI capex.”

Still, Chanana struck a note of caution if AI development is hampered more broadly over the longer term. “Investors should not completely rule out a slowdown later. If companies release fewer major models or run fewer very large training programs, future demand for the most advanced processors, memory chips, and chip-packaging capacity could be lower than currently expected.”

Get Morningstar’s insights in your inbox