Anthropic Confirms In-House AI Chip Development to Power Claude Models

On Wednesday (5th), Anthropic formally confirmed it has initiated a custom AI chip program and established an internal chip team to design specialized chips for its Claude models, addressing the continuously growing demand for AI computing power. This marks the first time Anthropic has publicly acknowledged its efforts in developing proprietary chips.

A spokesperson for Anthropic told Business Insider that the company aims to enhance the speed and energy efficiency of Claude through co-design of hardware and AI models, while better supporting the increasing deployment needs of enterprise customers.

However, Anthropic emphasized that its in-house chips will not replace existing partners, adopting a 'multi-chip strategy.' Beyond its own chips, the company will continue using AI computing hardware from suppliers such as AWS, Google, NVIDIA (NVDA-US), and AMD (AMD-US).

In fact, there were already signs in the market.

Reuters previously reported that Anthropic was evaluating the feasibility of designing its own AI chips; The Information disclosed last month that Anthropic had contacted Samsung Electronics to discuss potential future chip manufacturing collaboration.

Meanwhile, Anthropic’s recently released job postings further reveal its strategic direction. The company is recruiting engineers to join its 'custom chip team,' seeking talent with experience in semiconductor design, verification, and product development to advance chip R&D.

According to the job listing, the position offers an annual salary between $320,000 and $485,000, requiring candidates to have directly participated in the full lifecycle of chip development—from design finalization to mass production and delivery—along with strong decision-making skills and experience driving projects to completion under resource-constrained conditions.

Anthropic is not the only player investing in in-house AI chip development.

As large language models rapidly increase their computational demands, more AI companies are beginning to develop their own chips to reduce reliance on external suppliers.

For instance, OpenAI has partnered with Broadcom (AVGO-US) to develop a dedicated AI chip and announced its in-house chip project codenamed 'Jalapeño' this past June.