Anthropic discussed acquiring AI chip startup MatX for roughly $7 billion but ultimately abandoned the deal, Reuters reported on August 27, revealing the most detailed picture yet of the Claude maker's custom silicon ambitions.
The acquisition would have given Anthropic in-house expertise in designing chips optimized for large language models. MatX, founded by former Google engineers, builds high-throughput chips specifically for LLM inference and training.
MatX is now seeking to raise new capital at a valuation of about $4 billion — roughly half what Anthropic had discussed paying. The startup's reduced valuation reflects both the difficulty of the deal and the broader reshuffling in the AI chip landscape.
Anthropic confirmed on August 5 that it is building an internal Custom Silicon Team to co-design hardware and models, helping Claude run faster and more efficiently. While the MatX acquisition didn't close, Anthropic continues to pursue chip independence from Nvidia, whose GPUs currently dominate AI training and inference.
The company's run-rate revenue has now surpassed $30 billion, up from about $9 billion earlier in 2026, giving it significant resources to invest in custom hardware. Anthropic is also separately expanding its silicon team through hiring, with salaries for chip design engineers reaching up to $485,000.
