Amazon is about to further expand its partnership with AI start-up Anthropic with a new investment of several billion dollars. The tech company wants to strengthen its position in the global competition surrounding AI.
The Financial Times reports that the proposed investment is in addition to the $8 billion Amazon previously committed to Anthropic. That earlier deal enabled Amazon to become Anthropic’s primary cloud and training partner. The collaboration primarily focuses on the development of large-scale data centers and the utilization of Amazon’s own Trainium2 chips for training language models.
According to insiders, both companies now consider the collaboration to be strategically crucial. With the new investment, Amazon would also consolidate its stake in Anthropic without exceeding the one-third ownership threshold. Google, which has invested more than $3 billion in Anthropic, remains a player, but Amazon wants to maintain a dominant role.
Anthropic wants to operate independently
Anthropic develops advanced AI models, including Claude, which has now been integrated into Amazon products such as the improved Alexa assistant and Prime Video. At the same time, the company continues to operate independently and emphasizes that a single tech giant does not control it.
An important part of the collaboration is Project Rainier, an ambitious data center project in Indiana that accounts for 2.2 gigawatts of power consumption. The facility will meet Anthropic’s growing computing power needs. Further expansions are already in the works.
Although Amazon is also working on its own AI models, it has deliberately opted for a dual strategy. On the one hand, it is investing in external players such as Anthropic. On the other hand, it is continuing to develop internally. In addition to Anthropic, Amazon has previously invested in other AI companies, such as Hugging Face and Scale AI; however, this partnership is on a larger scale.
Amazon believes that the structural nature of the collaboration, in which Anthropic functions as a ‘public benefit corporation’ and investors hold real shares, is more stable than other large AI partnerships, such as that between Microsoft and OpenAI.