Anthropic Forms Internal Chip Design Team to Optimize Claude
Anthropic will form an internal chip design team to create custom chips for the Claude model and will recruit engineers with hardware and software stack experience to collaboratively design chips and AI models, making Claude run faster and more efficiently.
The company stated it will continue to adopt a diversified hardware strategy, including AWS, Google, Nvidia, and AMD.
Event-driven AI companies are accelerating funding for self-developed chips, intensifying competition for high-end hardware talent, and relatively easing reliance on external chip suppliers.
Source: Public Information
ABAB AI Insight
Anthropic has previously collaborated with several chip suppliers and explored self-development possibilities; the formal establishment of a team and public recruitment marks a shift from reliance on external computing power to collaborative design of hardware and software. Custom chip design is costly, typically requiring hundreds of millions in investment.
On the capital front, reducing dependence on a single supplier through an internal team and optimizing inference efficiency is motivated by the need to address computing power shortages and scaling customer demands; resources are shifting from purchasing external chips to parallel self-development and multi-source strategies.
Similar cases can be seen in OpenAI's collaboration with Broadcom for custom chips, Google TPU, and Meta MTIA's approach; current AI labs are transitioning from purely software models to hardware-software collaborative optimization.
This essentially represents a technological substitution: AI inference is shifting from general-purpose chips to model-specific custom silicon, with the mechanism being performance and energy efficiency improvements through collaborative design, thereby reducing large-scale deployment costs.
ABAB News · Cognitive Laws
Model scale ultimately drives hardware self-development
Hardware-software collaboration is the shortest path to efficiency improvement
Diversified chip strategy is a necessary choice for risk hedging