- Anthropic is considering building its own AI chips to cater to the increasing demand for AI processing.
- The shift towards in-house chip manufacturing comes as the company seeks to decrease reliance on third-party providers like Nvidia.
What happened
Anthropic is exploring building its own AI chips as demand for computing power surges and its operational revenue rapidly increases. The company’s AI model Claude has seen accelerating adoption in 2026, with Anthropic’s operating revenue exceeding $30bn, compared with about $9bn at the end of 2025 — underscoring how quickly the need for high‑performance hardware is rising.
Sources told Reuters the plan remains at an early exploratory stage. Anthropic may still choose to purchase commercial AI chips rather than develop end‑to‑end chip products itself, and the company has not finalised specific design plans or formed a dedicated chip‑building team.
Currently, Anthropic uses a range of third‑party chips — including Google‑designed Tensor Processing Units (TPUs) and chips from Amazon — to train and run its models, particularly for the Claude family of AI systems.
Earlier this year, Anthropic also signed a multi‑year agreement with Broadcom related to TPU design work. That deal is reportedly based on Broadcom’s commitment to spend up to $50bn to strengthen U.S. semiconductor and infrastructure capabilities.
The discussions at Anthropic reflect a broader trend in the tech industry, where major companies such as Meta and OpenAI are also seeking to design their own AI chips. According to industry sources, designing an advanced semiconductor can require hundreds of millions of dollars and deep engineering expertise.
Why it’s important
Building its own AI chips would offer Anthropic greater control over its hardware and enable it to better meet the demanding requirements of AI models. As AI systems become more complex, the need for bespoke, high-performance chips is becoming crucial. This move could allow Anthropic to bypass the bottlenecks associated with relying on external chipmakers, especially as the global AI chip market becomes more competitive.
The development of in-house AI chips marks a vertical integration strategy for AI companies, significantly enhancing their control over computational power, which is emerging as a core element of future competition. By manufacturing its own chips, Anthropic would have the ability to fine-tune hardware performance to match the specific needs of its AI models, potentially offering advantages in processing speed and cost efficiency. Moreover, this development would signal a shift toward more integrated hardware-software strategies in AI companies, which could set a new precedent for innovation in the sector. With Nvidia already seeing strong demand for its AI chips, including in high-profile partnerships with firms like OpenAI, Anthropic’s venture into custom hardware development could make it a key player in the future of AI infrastructure.
Also read: Huawei AI Chips Gain Traction as Orders Grow
