Three years ago, Bill Gates wrote an essay about artificial intelligence that read like a sales pitch for the future. His new one does not. In a nearly 6,000 word piece titled “The turbulent AI era is here. The choices we make now are critical,” the Microsoft cofounder warns that AI “will either be the greatest equalizer ever invented, or the worst source of injustice,” and argues society is unprepared for either outcome.

The reversal is notable because of who is making it. Gates spent much of the last two years publicly quiet on AI after his 2023 essay cast the technology as a historic force for good, from curing disease to closing education gaps. The new essay does not retract that hope entirely, but it subordinates it to a warning: “we are not preparing for it.”

Gates’s central claim is about scale and speed, not just risk. He argues people are underestimating the transition because they compare it to past technology shifts like the arrival of PCs, and because early stumbles in AI systems tend to get “fixed quickly,” creating a false sense that the technology is maturing gradually. “This time really is different,” he writes.

That framing matters for how readers should weigh the rest of the essay. Gates is not describing a new category of risk. He is arguing the familiar risks, job loss, concentrated power, uneven access, are compressed into a shorter timeline than prior technological disruptions allowed for. He goes as far as saying that if a credible plan existed to slow global AI development, he would likely back it, a notable position from someone whose own company, Microsoft, has spent the past three years accelerating AI deployment across its product lines.

On jobs, Gates predicts something closer to structural rupture than a normal labor-market adjustment: widespread, permanent unemployment and social upheaval, not the temporary displacement that accompanied earlier automation waves. The essay’s other flagged idea, taxing robots and reserving categories of work for humans, is offered as a partial response to that prediction rather than a fully worked policy.

What the essay does not do is offer a specific mechanism for any of it. According to reporting on the essay, it largely retreads ground already familiar in AI governance debates, systemic and deeply entrenched problems, without proposing solutions that differ meaningfully from what policymakers and researchers have already floated. Gates closes by insisting it is not too late “to ensure that the good outweighs the bad,” but the essay itself supplies little evidence for why that outcome is more likely than the one he now fears.

Gates’s platform for making this argument is not neutral, and naming that is not the same as dismissing the argument. He remains Microsoft’s cofounder, whose fortune and continued relevance are tied to the company’s AI-driven growth, and he runs a foundation whose global health and development grantmaking increasingly intersects with AI-adjacent tools. A warning from that vantage point carries real signal about where the industry’s own insiders see danger, even without a new roadmap attached.

For operators, the practical takeaway is not the robot tax proposal, which has no legislative traction attached to it. It is that one of the industry’s most prominent boosters now says the labor disruption could be structural rather than cyclical, a signal worth weighing against your own workforce and product roadmaps over the next two quarters.

Reported by The Verge on 26 August 2026.