Meta has officially decided to halt plans to lease its AI computing power to Anthropic. The initiative was discussed internally after Anthropic CEO Dario Amodei reached out to Meta's chief AI officer, Alexandr Wang, earlier this year to request access to the company's hardware.
The decision stems from Meta's own rapidly increasing demand for computing resources. As the company continues the rollout of its Muse series of proprietary models, it has shifted focus toward prioritizing internal needs over external partnerships that could potentially benefit a direct competitor.
Industry experts note that the allocation of computing power is becoming a critical factor in the AI labor market. Limited access to high-performance infrastructure has already influenced the career decisions of top researchers, including the departure of Transformer architecture co-author Noam Shazeer from Google.
The broader AI sector currently faces challenges as many large-scale promises regarding computing capacity remain unfulfilled. These infrastructure constraints are increasingly viewed as a primary bottleneck for the development of the entire artificial intelligence industry.
