A line of public code that should not have exposed commercial secrets has revealed Anthropic's next chip layout. According to the chip industry analysis firm SemiAnalysis, the code published on GitHub by AMD's senior director of AI business clearly leaves traces of Anthropic as an AMD customer. This non-official, non-media leak sends a clear signal: Anthropic is actively expanding its chip sources, freeing its compute suppliers from a single list.

On Monday, SemiAnalysis posted an analysis on the social platform X, and the conclusion came from a professional dissection of the public code by AMD's senior director. The code content directly involves information related to Anthropic as an AMD customer. The institution stated that it will further explain the code details and background. If this judgment is correct, AMD will gain a new position in the high-end AI training market previously dominated by NVIDIA and Google, and Anthropic's computing infrastructure will thus accelerate its move towards diversification.

The flow path of this information is quite special. It did not come through official announcements or news reports, but rather through code pushed to GitHub by an internal executive of AMD. SemiAnalysis deciphered this code and then publicly shared the conclusions on X. In the technology industry, such cases of accidental leaks through public code repositories are not uncommon—GitHub is inherently searchable, and the publisher was the senior director of AMD's AI business, which combined makes the credibility of the information quite high. In the post, SemiAnalysis promised to continue breaking down the code logic layer by layer, providing the market with a more complete interpretation.

In fact, before this line of AMD was exposed, Anthropic's computing procurement had already been far from a single path. Google's TPU (Tensor Processing Unit) is one of its important computing sources, which is closely related to Google's strategic investment in Anthropic; additionally, Samsung has also been included in its supplier system. The existing layout indicates that Anthropic has never bet on a single point, but has deliberately created a diversified supplier matrix. The addition of AMD will only make this network have another high-performance computing option outside of NVIDIA GPUs.

For Anthropic, bringing AMD into the alliance carries multiple strategic weights. As model training and inference scale continue to expand, the supply chain risks, suppressed bargaining power, and the drawbacks of being locked into a specific technical path brought by a single chip supplier have become increasingly evident. By embracing Google TPU, Samsung, and AMD at the same time, it can re-find a better balance between cost control, supply stability, and technological flexibility. For AMD, winning over Anthropic is a landmark victory—this company has been continuously investing in data center AI chips, and its MI series GPUs are considered one of the main competitors to NVIDIA's H series products; once Anthropic officially joins, AMD will not only enhance its presence among top AI laboratories, but also bring real incremental revenue for its data center AI chip business.

Anthropic's move reflects a systemic shift in the entire industry. As OpenAI, Google DeepMind, Meta AI, and other leading institutions continue to increase their demand for computing power, AI companies are collectively evaluating and expanding their chip supply sources, just to avoid the risk of putting all eggs in one basket. This collective turn objectively provides AMD, Intel, and other companies chasing NVIDIA, an opening to enter the core AI customer supply chain.