Mustafa Suleyman, chief of Microsoft AI, told EuroHerald that the industry must pause to avoid creating systems beyond human control. His comments come as investors reacted sharply to calls for a slower AI race, with chip makers sliding while cloud providers rose.
Market reaction to AI safety calls
On Monday, shares of semiconductor firms fell after leading AI researchers and politicians urged a more cautious approach. Nvidia dropped more than three percent, while Intel, AMD and Marvell each slipped between five and six percent, pulling the Philadelphia Semiconductor Index down almost six percent.
In contrast, the biggest spenders on AI infrastructure, Alphabet (Google), Microsoft and Meta, posted gains of roughly two, 1.6 and 1.4 percent respectively.
Why the split matters for Europe
The divergence highlights a strategic dilemma for European investors and policymakers. If AI development slows, the demand for high-performance chips could wane, hurting companies that supply the hardware. At the same time, cloud operators may see higher cash flows as they can halt new data-centre construction while still monetising existing capacity.
Gil Luria, head of technology research at D.A. Davidson, likened the scenario to Amazon's post-pandemic reduction in warehouse building, which boosted cash flow despite lower capital outlays. He warned that a slowdown would hurt the "picks and shovels" firms more than the "hyperscalers" that buy the tools.
Microsoft's new AI code of conduct
"We shouldn't be trying to design models that can recursively self-improve beyond our control," Suleyman said.
In response to mounting safety concerns, Microsoft unveiled a code of conduct that bans AI systems from simulating feelings, pursuing intrinsic motivation or assisting in the creation of chemical, biological or nuclear weapons, among other prohibited uses.
Suleyman added that coordination across labs is essential and called for transparent disclosure of model capabilities to responsible third parties.
Regulatory pressure in the United States
Across the Atlantic, the Midas Project, an AI watchdog, accused OpenAI of breaching California's Transparency in Frontier AI Act, a law that obliges large AI developers to publish risk assessments and adhere to their own safety frameworks. OpenAI disputes the claim, asserting compliance and ongoing investment in risk mitigation.
These developments suggest that regulatory scrutiny will intensify, potentially influencing European AI strategy as the EU finalises its own AI Act and related safety standards.
What comes next?
Analysts expect semiconductor firms to monitor AI policy signals closely, adjusting production plans if a slowdown materialises. Meanwhile, cloud providers may continue to reap cash-flow benefits from existing infrastructure while awaiting clearer guidance on AI safety.
European regulators are likely to watch the unfolding debate, balancing innovation incentives with the need to prevent uncontrolled AI systems. The next few months could see further alignment between industry self-regulation and government policy, shaping the continent's role in the global AI race.