From model to silicon: Anthropic’s next big bet could be custom AI chips

Anthropic, the Claude AI-maker, is exploring a plan to build custom Artificial Intelligence (AI) chips amid the growing shortage of computing hardware. According to a Reuters report, the AI company wants to reduce dependence on third-party chip suppliers as AI chip demand grows to power and develop advanced AI systems.

The report noted three unnamed sources of Anthropic plans, revealed that it is in early-stage exploration, and noted that the company also relies on an AI chip supplier. Rivals, including Google, Amazon, and Microsoft, are already using custom hardware, bringing enhanced security directly into chips, rather than just relying on software.

Also read: Anthropic signs multi-gigawatt AI compute deal with Google, Broadcom

As of now the Anthropic has not decided on a design yet or has plans to build a team to work on it, as per the report. Currently, the company relies on companies like Google and Amazon for a range of chips to run its AI software and Claude AI chatbot.

Earlier this week, Anthropic signed a long-term deal with Google and Broadcom to focus on tensor processing units (TPU) design. This aligns with  Anthropic’s $50 billion commitment to expanding U.S. computing infrastructure.

Also read: Anthropic releases preview of new AI model Mythos for cyber defence

Although designing its own chip will require a massive amount of investment, roughly half a billion dollars, as per the report. Anthropic will have to spend a massive amount on employing engineers and hardware architects, who are currently in high demand by Apple, Nvidia, and Google. Furthermore, the company will have to design a chip from scratch and also have to invest in manufacturing processes.

As of now, information about Anthropic’s custom AI chip remains scarce, but we may have to wait until the company formalises its plans or begins assembling a dedicated team before clearer details emerge.