Anthropic joins the AI chip race: the company behind Claude is building its own hardware

Anthropic, the developer of the AI assistant Claude, is taking a significant step toward technological independence by forming a specialized team to design its own chips for artificial intelligence, according to TechCrunch.

The decision marks a major strategic shift for the company, which has so far relied exclusively on external hardware suppliers to train and run its language models. According to the source, Anthropic plans to develop hardware and AI models in parallel, through a co-design process aimed at boosting both speed and energy efficiency.

Why hardware-software co-design matters

The approach is no accident. Companies that manage to optimize chips and algorithms simultaneously can gain major competitive advantages in terms of operating costs and performance. Training large language models involves enormous computing infrastructure costs, and a chip designed specifically for the architecture behind the Claude models could significantly cut those expenses.

In practice, rather than adapting its models to the limitations of hardware already available on the market, Anthropic could build chips designed from the ground up to serve the specific needs of its own AI algorithms.

Anthropic is following an established industry trend

Anthropic's move is not unique in the tech industry. Companies such as Google, with its TPU processors, and Amazon, with its Trainium and Inferentia chips, have already demonstrated the benefits of having greater control over computing infrastructure. OpenAI has also explored similar options to reduce its dependence on third-party suppliers such as Nvidia.

For Anthropic, which is financially backed by giants like Amazon and Google, this move could also serve as a way to diversify its hardware sources, reducing the risks tied to reliance on a single supplier in a market where demand for AI chips frequently outstrips available supply.

Implications for the future of the AI market

If the strategy proves viable, Anthropic could become an even stronger player in the global competition for AI supremacy, alongside companies investing heavily in their own infrastructure. It remains to be seen how quickly this project will progress and what impact it will have on the operating costs of the Claude models in the medium and long term.