Editorial illustration for Anthropic is building a team to design its own AI chips
Anthropic Builds Custom AI Chip Team for Claude
Anthropic is building a team to design its own AI chips
Anthropic is putting together a team to design its own AI chips, according to Business Insider. The company behind Claude is looking to hire engineers for what it calls a "custom silicon team," building hardware alongside its models so the whole system runs faster and cheaper. The Information reported last month that Anthropic had already been talking to Samsung about a potential manufacturing partnership for the effort.
The timing tracks with how stretched Anthropic's compute situation has gotten. Claude usage keeps climbing, and Anthropic has spent the past year locking in deals with AWS, Google, Nvidia and AMD just to keep pace. Renting capacity from four different partners only goes so far when demand keeps outrunning supply, which appears to be pushing Anthropic toward building its own silicon instead of leaning entirely on outside vendors.
That puts Anthropic in company with rivals who made the same call earlier. OpenAI rolled out its Broadcom-built Jalapeño chip in June for inference work, Google DeepMind has run on Alphabet's TPUs for years, and Meta has its own MTIA accelerators in development. Anthropic hasn't responded to a request for comment on its plans.
Anthropic’s decision to design its own chips comes as demand for Claude rises while AI companies snatch up as many AI infrastructure deals as they can.
Why this matters
Google, Amazon, Microsoft and Meta already design their own silicon; Anthropic joining them tells us renting compute from Nvidia and cloud partners no longer looks sustainable at Claude's scale. If Samsung ends up fabricating these chips, that's a multi-year commitment, not a side project, and it signals Anthropic expects demand to keep climbing long enough to justify the engineering overhead of co-designing hardware and models together.
For developers building on Claude, custom silicon could eventually mean cheaper inference and faster response times, but that's years off, not a near-term pricing shift. Founders should read this as another data point in the infrastructure land grab: whoever controls their own chip supply controls their own margins and roadmap, and everyone else is left negotiating for scraps of Nvidia capacity. Researchers watching the hardware-software co-design trend should note this isn't unique to Anthropic anymore. It's now table stakes for any lab serious about competing past the next funding round, which raises the bar for smaller labs without the capital to follow suit.
Common Questions Answered
Why is Anthropic building its own AI chips instead of relying on Nvidia?
Anthropic is designing custom silicon because renting compute from Nvidia and cloud partners is no longer sustainable at Claude's scale as demand for the AI model continues to rise. By building hardware alongside its models, Anthropic can make the entire system run faster and cheaper while maintaining better control over its infrastructure needs.
What is Anthropic's custom silicon team planning to do?
Anthropic's custom silicon team will design proprietary AI chips specifically optimized to work with Claude and other Anthropic models. This approach allows the company to co-design hardware and models together for improved performance and efficiency across the entire system.
Who is manufacturing Anthropic's custom AI chips?
According to The Information, Anthropic has been in talks with Samsung about a potential manufacturing partnership for producing the custom chips. If Samsung becomes the fabricator, it would represent a multi-year commitment rather than a temporary project, signaling Anthropic's confidence in sustained long-term demand.
How does Anthropic's chip design effort compare to other major tech companies?
Anthropic is joining Google, Amazon, Microsoft, and Meta in designing their own silicon for AI infrastructure. This move by Anthropic indicates that the company has reached a scale where custom hardware design is necessary to remain competitive and cost-effective in the AI market.
Further Reading
- It's official: Anthropic is building an in-house chip team for Claude - Business Insider
- Anthropic weighs building its own AI chips: Reuters - CNBC
- Anthropic in Talks With Samsung to Manufacture Custom AI Chip - The Information
- OpenAI's custom chip program lead Clive Chan joins Anthropic - Data Center Dynamics
- Anthropic Takes First Step Toward In-House AI Chip - Yahoo Finance / Reuters