Anthropic is establishing a dedicated internal team to design custom silicon for its Claude artificial intelligence models, marking a strategic shift toward vertical integration in hardware.
The company confirmed the initiative on Wednesday, aiming to optimize performance and efficiency for its large language models by moving beyond off-the-shelf processors.
The move follows significant capital backing from Advanced Micro Devices (AMD), which announced a partnership and up to $5 billion in investment for Anthropic earlier this year.
This development signals a deepening competition in the AI infrastructure layer, where major software firms are increasingly seeking control over their compute stack.
By building in-house chip capabilities, Anthropic seeks to tailor hardware specifically to the architectural needs of its models, potentially reducing latency and energy consumption compared to general-purpose accelerators.
The move follows significant capital backing from Advanced Micro Devices (AMD), which announced a partnership and up to $5 billion in investment for Anthropic earlier this year.
That deal was structured to support the development of AI chips, suggesting a collaborative approach where AMD’s foundry and design expertise may complement Anthropic’s new internal efforts.