Morgan Stanley sees AI safety push as tailwind for tech spending

AI adoption is already showing measurable gains. Morgan Stanley found that 25% of S&P 500 .SPX companies report quantifiable AI benefits, up from 14% a year ago. Washington could also increase support for domestic frontier-model developers while tightening restrictions to protect U.S. AI technology and address risks from open-weight models, a policy shift Morgan Stanley said could further boost computing demand.

Lastly, a stronger focus on AI safety could fuel public opposition to data centers, while the U.S. midterm elections could bring a regulatory shift toward tighter restrictions on model providers and data-center developers.

Morgan Stanley remains bullish on demand for computing power, arguing that greater investment in AI safety could accelerate, rather than constrain, the broader AI spending boom.

“A focus on AI safety and security can be a tailwind to capex, but it pays to understand potential policy variables and pathways,” Morgan Stanley said in a note on Sunday.

The brokerage expects spending on model monitoring, cybersecurity, and bio-defense to increase as AI capabilities advance, while hyperscaler capital expenditure is forecast to reach $1.5 trillion in 2027.

Last week, calls for stronger AI safeguards gained momentum after Anthropic CEO Dario Amodei urged the industry to slow development, with other technology executives and political leaders backing a greater focus on AI safety. Against that backdrop, Morgan Stanley outlined five outcomes that could support strong demand for computing power and capital spending.

First, government intervention could slow releases of advanced frontier models, potentially benefiting open-model developers. Still, it sees long-term AI capex concerns as misplaced, as enterprise AI adoption remains low and demand for inference computing continues to outstrip supply.