AI Spending Stays Strong Despite Slowdown Warnings

The split is now public and unusually sharp. Dario Amodei of Anthropic called for a three-step slowdown after warning advanced models could become uncontrollable, while Elon Musk backed the idea and Sam Altman said the pace should be calibrated. Yet Donald Trump dismissed the concern, saying the US is ahead of China and must stay there, even as his administration frames data centers as a strategic asset and Jensen Huang echoed support for continued expansion.

The market consequence is that AI spending is likely to remain elevated rather than retrench into a safety-first pause. Nvidia, Microsoft and Alphabet are trading off a policy backdrop in which frontier-model development continues, despite the existential rhetoric. Nvidia shares, after a volatile stretch, were last around $212.17, below a recent $227.73 peak but still well above the $197.42 200-day moving average; Microsoft closed at $497.12, holding comfortably above its $429.94 200-day average; Alphabet ended at $344.98, near its $336.65 200-day line. The tape says investors are still paying for the AI buildout, even as they mark up the risk premium.

Adalytica’s proprietary AI sentiment snapshot shows the tension in blunt form: sentiment sits at 4, or “Extreme Fear,” while awareness remains at 100, “Extreme Greed.” In other words, investors and the public are fully focused on the danger, but capital is not leaving the theme. That is exactly the sort of disconnect that creates asymmetric opportunity in infrastructure names and the chips that sit at the center of the stack.

The bigger narrative is not whether AI is becoming smarter faster than regulators can react. It is that geopolitical rivalry has made slowdown impossible. Washington does not want to concede ground to Beijing, and Beijing is equally unwilling to be boxed out. Chinese state media is already casting US safety arguments as a cover for monopoly power, while Washington treats AI as a national-security race. Once the contest becomes strategic, safety objections matter less than industrial policy.

For investors, that means the winners are not necessarily the companies making the loudest claims about artificial general intelligence. The more durable beneficiaries are the toll collectors: Nvidia on accelerators, Microsoft and Alphabet on cloud and model deployment, and the broader data-center supply chain on power, networking and cooling. If the AI race continues at full speed, demand for compute, electricity and storage remains the real trade.

The near-term catalyst is more hearings, more rhetoric and more selective pauses in frontier training, but not a true halt. If anything, every new warning from insiders reinforces the idea that the technology is moving too fast to stop. That is a policy failure for governments — and a multi-year tailwind for the infrastructure layer of the AI economy. Stay positioned in the enablers, not the headline risk.