Anthropic assembles in-house chip design team for Claude
Anthropic confirmed on August 5 that it is building an internal team to design custom silicon for its Claude AI models, joining a growing list of AI labs that have concluded off-the-shelf chips are no longer sufficient for the scale they need.
The company is recruiting engineers with chip design experience for what it calls a "custom silicon team," according to a job listing. The group will co-design hardware and AI models in tandem, aiming to make Claude run faster and more efficiently at production scale.
Following a familiar path
Anthropic is not the first to go this route. OpenAI unveiled its Broadcom-built Jalapeño inference chip in June. Google DeepMind has long relied on Alphabet's TPUs. Meta has been developing its MTIA accelerators for years. What Anthropic brings to the table is its stated approach of designing the chip and the model together, rather than adapting an existing architecture to a pre-trained model.
Last month, The Information reported that Anthropic was scouting Samsung as a potential manufacturing partner, though no deal has been confirmed.
Why now
The move comes as demand for Claude strains available compute. Anthropic has existing deals with AWS, Google, Nvidia, and AMD to access AI hardware, and earlier this week signed a reported $10 billion compute agreement with Norwegian cloud startup Volta. Building its own chip team signals those deals are not enough to close the gap between supply and demand.
The broader context: Nvidia's AI chips remain supply-constrained, and every major AI lab is racing to lock down compute. Custom silicon represents a bet that vertical integration -- designing chips purpose-built for specific model architectures -- will deliver cost and performance advantages that general-purpose GPUs cannot match.
The competitive picture
Anthropic's chip effort arrives at a moment of rapid infrastructure expansion across the industry. Google moved $35 billion in Anthropic chip risk off its balance sheet through a special purpose vehicle this week. AMD's data center revenue more than doubled in its most recent quarter. SpaceX reported that Anthropic and Google compute deals contributed to a doubling of its own revenue.
Designing competitive AI chips is a multi-year, multi-billion-dollar undertaking. Anthropic is at the starting line -- hiring a team and finding a fabrication partner. Production silicon is likely years away, but the direction of travel for the largest AI labs is now unmistakable.
