Not everyone is taking Dario Amodei's pacing essay at face value. In an episode of his podcast titled "The Great AI Slowdown Cover Story," longtime cloud and AI analyst David Linthicum argues the industry's sudden, coordinated concern about AI safety isn't really about safety at all. His theory: Frontier labs know their capability gains are stalling, and turning a business slowdown into a moral stand is good PR for a problem they'd rather not admit to directly.
Linthicum is not a fringe voice. He's a former chief cloud strategy officer at Deloitte, founder of Linthicum Research, and a regular InfoWorld columnist, and he hosts a podcast built specifically around a skeptical read of AI industry claims. That doesn't make his theory correct. It does mean it's worth taking seriously rather than dismissing outright.
The Core Challenge: Show Your Work
Linthicum's central complaint isn't that AI risk is impossible. It's that the warnings from Amodei and others stay abstract by design. He's asking for the mechanism, not the mood: which system fails first, what access does the model actually have, and what is the concrete, step-by-step path from a model running in a data center to something capable of hurting people at scale. Superintelligence, digital gods, and extinction-level framing, in his view, are doing the work that a specific technical scenario should be doing instead.
The Alternative Explanation: A Plateau in Costume
Linthicum's second argument is about timing. He points to the frontier model market showing real signs of fatigue: releases that demo well and add polish, but where the underlying capability leaps are smaller, more expensive, and harder to sustain than the presentations suggest. His conclusion follows from that observation rather than from the essay itself: if the major labs already expected to slow down for ordinary technical and financial reasons, reframing that slowdown as a voluntary safety sacrifice is a considerably better story to tell investors, employees, and the public than admitting the technology has hit a wall.
He's Not Alone in Asking
Versions of this same question have shown up elsewhere since Amodei's essay published. Asia Times ran commentary asking directly whether the coordinated pause is a safety measure or a strategic retreat, noting that a public commitment to slow down doubles as a convenient explanation for a plateau in model performance. The Register made a related but distinct argument: that the essay reads less like a safety intervention and more like the largest labs writing the regulatory rules themselves before anyone else gets the chance. Linthicum's version is the sharpest of the three, since it doesn't just question the framing, it questions whether there's a genuine capability slowdown underneath the safety language at all.
The Case Against the Theory
The strongest evidence cutting the other way is one WorkplaceAI has already reported: A coordinated slowdown is genuinely costly to the companies proposing it. It would hand slower-moving or smaller competitors time to close the gap, and Amodei says as much directly in his essay, arguing that democracies can only afford to pace themselves as far as their current lead over competitors allows. If the three biggest labs are volunteering a position that works against their own competitive interest, that's at minimum a more complicated story than pure reputational cover.
It's also worth noting that AI capability plateauing and AI capability advancing too fast to control safely are opposite claims, and the industry has had prominent voices arguing both within the same month. Both theories can't be fully right. It's possible neither is, and the truth sits somewhere in the messier middle: real safety concerns, real competitive incentives, and a genuine cooling in the pace of headline-grabbing capability jumps, all operating on the industry at once.
Neither of those readings is resolved by anything Amodei, Altman, Musk, or Linthicum have said so far, since the honest answer requires information none of them have publicly disclosed: how close current models actually are to the next capability tier, and how much of the industry's public conduct this week has been about protecting people versus protecting narratives. Worth watching both threads as they continue to develop.
Sources: AI Pulse · Big Picture · workplaceai.ai. David Linthicum, "Dave Linthicum Is Not AI: The Great AI Slowdown Cover Story," episode description published via Libsyn, September 2026. Background on Linthicum: InfoWorld contributor profile and Linthicum Research. The competing critiques: Asia Times, "Big AI slowdown a safety pause or strategic retreat?" and The Register, "Big AI sets out its terms for regulatory capture and calls it 'Pace the frontier,'" both September 14, 2026. Context on Amodei's essay and industry reaction: previously reported by WorkplaceAI AI Pulse. Every claim and quote above is attributed to its original source; none is a WorkplaceAI study, and WorkplaceAI takes no position on which explanation is correct.