Anthropic Building Custom Silicon Team for Claude AI
AI giant aims to co-design hardware and models to boost performance through in-house chip development.
1Custom Silicon Team Formation
Anthropic is establishing a new team to develop custom chips for its Claude AI model. In an August 5, 2026 announcement, the company revealed plans to co-design hardware and models to enhance Claude's speed and efficiency. The team will oversee the entire process from chip design to production. This move forms part of Anthropic's strategy to gain greater control over AI hardware infrastructure, while maintaining a multi-chip approach that still incorporates third-party components.
2Job Listings Reveal Team Structure
Two job postings on Anthropic's careers portal detail the custom silicon team's composition. The company is recruiting for Silicon Engineer and Technical Program Manager positions in San Francisco, New York, and Seattle. The listings seek candidates with experience working from the chip level up with partners. Notably, they emphasize the ability to make consequential decisions without relying on large organizational support structures - suggesting significant autonomy for team members.
3TPM Role's Critical Responsibilities
The Technical Program Manager position will own decisions defining the chip program's economics. Responsibilities include managing technology and IP selection processes, and making critical build/buy/license decisions with engineering leadership. Candidates need 8+ years of silicon program management experience and must have directly owned at least one complete chip program from design to production. The role will coordinate external engagements across ASIC design services, IP vendors, foundries, and packaging/test contractors.
4Engineering Profile and Expectations
The Silicon Engineer listing targets professionals experienced with machine learning accelerators or other large, high-bandwidth designs. Deep expertise in memory subsystems, on-die interconnects, numerics, or low-power design is required. Engineers must demonstrate capability to review partner work and maintain technical standards. Anthropic notes it's 'opening searches across key domains,' with each hire covering 'a large surface' - indicating broad responsibility per team member.
5Operating Model: Lean Team, Extensive Partner Network
Rather than building a full-scale design organization, Anthropic is adopting a model combining a small internal team with external contractors. The TPM role is explicitly described as 'a partner-heavy program by design.' While maintaining design authority in-house, the company will collaborate with external partners for ASIC, IP, foundry, and packaging capacity. Reports suggest Anthropic has discussed Samsung as a potential manufacturing partner for this initiative.
6AI-Assisted Engineering Focus
Both listings require candidates to bring informed perspectives on AI-assisted engineering. The TPM role expects hires to 'push hard on what AI-assisted program management can look like.' Engineers need 'clear ideas about where AI helps and where it doesn't' - particularly regarding using Claude to help design the silicon that runs Claude. This recursive approach to AI development represents an innovative application of the technology to its own infrastructure creation.
7Salary Bands Reveal Program Investment
The public job postings provide insight into Anthropic's talent investment for this program. The engineer position spans eight domains including front-end design, pre-silicon verification, physical design, and packaging with signal integrity. Published salary ranges for these specialized roles indicate significant financial commitment to building custom silicon capabilities. Anthropic emphasizes it's conducting searches across all key technical domains simultaneously to accelerate development.
8Strategic Shift in AI Hardware Approach
Anthropic's move reflects a broader trend of AI companies taking greater control over their hardware stack. The hybrid model - combining a lean internal team with contracted specialists - aims to balance cost control with design authority. The company's statement that Claude's hardware represents 'one of the most direct levers on capability, cost, and reliability' underscores the strategic importance of this initiative. This development may mark a new phase in the AI infrastructure race, where model quality becomes increasingly tied to custom hardware optimization.
Original article: Unite.AI — Anthropic, Claude için Özel Çip Geliştirme Ekibi Kuruyor