Musk Admits He Was Wrong About Anthropic in Rare Reversal

Elon Musk has publicly admitted he was "clearly wrong" about Anthropic, the AI safety company behind Claude, and pledged that he would not use SpaceX AI's compute access as leverage against the firm. The statement, reported by Yahoo Finance, represents one of the more candid reversals from a figure who has spent years making pointed criticisms of AI companies he views as ideologically opposed to his own ventures.

What Musk Said and Why It Matters

Musk's admission touches on a genuinely complicated history. His relationship with Anthropic is not straightforward: he was an early backer of OpenAI, which shares several former employees with Anthropic, and he has at various points dismissed safety-focused AI labs as too cautious or politically motivated. The pledge not to weaponize compute access is notable because SpaceX AI controls significant infrastructure that could, in theory, be used to squeeze competitors or partners in the increasingly tight market for GPU time. Musk's framing, "not my style," stops short of a formal commitment but carries public weight given how closely the AI industry watches his moves. This comes at a time when Anthropic's revenue has surpassed a $30 billion run rate and the company is expanding its compute partnerships aggressively.

The timing is worth examining. Musk's tone toward Anthropic has been shifting for some months. Earlier this year he described Claude as "woke" and implied the company's safety focus was a kind of ideological gatekeeping. Those comments drew significant attention at the time, and by July he had already begun walking back that framing, suggesting the reversal announced this week has been building for a while rather than arriving out of nowhere.

"I was clearly wrong about Anthropic. That's not my style, to use compute as a weapon against them."Elon Musk, via Yahoo Finance

The Compute Question Looming Over the Industry

Compute access has become one of the central pressure points in AI competition. Controlling GPU clusters gives large infrastructure players meaningful influence over which models get trained, at what speed, and at what cost. Anthropic has worked to diversify its compute base, with Amazon committing $25 billion and 5 gigawatts of compute capacity to support Claude. Against that backdrop, any suggestion that a rival infrastructure holder might restrict or manipulate access would be taken seriously by investors and partners alike. Musk's pledge, whatever its legal weight, removes one potential variable from an already complex equation.

For those following the deeper history between Musk and Anthropic's founders, this moment carries extra texture. Several of Anthropic's senior team members came through organizations Musk has been publicly critical of, and the competitive dynamics between xAI, OpenAI, and Anthropic have made for an unusually fractious corner of the tech industry. A public acknowledgment of being wrong, even a brief one, is not common currency in that world.

What this means practically for Anthropic's day-to-day operations is likely limited. The company has secured major infrastructure deals and continues to scale its enterprise offerings. But perception matters in a sector where partnerships, talent recruitment, and regulatory relationships all hinge partly on reputation. A reduction in public hostility from one of the industry's most prominent figures is not nothing, even if the underlying competitive dynamics remain unchanged.

Whether Musk's reversal holds, and whether it translates into any form of cooperation or simply a cessation of open criticism, remains to be seen. For now, it stands as an unusual moment of candor in an industry that rarely sees public admissions of error from its most vocal players.