Anthropic has hired Amir Salek, a former Google executive who helped build Alphabet's custom AI chip programme, as the Claude maker steps up work on silicon of its own.
The hire was reported by Bloomberg, according to Outlook Business, which framed the move as Anthropic "preparing for a possible push into its own semiconductor business." BusinessWorld described the same appointment as part of a "custom silicon push."
The detail that matters is Salek's background. Alphabet's in-house chip effort is one of the few examples of a major technology company successfully designing AI processors for its own use rather than buying them off the shelf. Hiring someone who worked on that programme is the kind of move a company makes when it is serious about the option, not merely curious about it.
For now, that is the extent of what the sources establish. There is no announced chip, no timeline, no manufacturing partner, and no stated budget in any of the reports. Anthropic has hired a person with relevant experience; the rest is inference.
Still, the direction is worth understanding. Training and running large AI models is enormously expensive, and most of that cost flows to specialised processors. Companies that design their own chips can tune the hardware to their specific models and reduce dependence on outside suppliers. Companies that don't are subject to someone else's prices, priorities, and production schedules.
Why it matters: if one of the leading AI labs starts building its own processors, the economics of who profits from the AI boom — chipmakers or the model developers themselves — could begin to shift.