Amazon announced plans on 25 September to invest up to US$4 billion in Anthropic, taking a minority stake. AWS will become the AI company’s main cloud provider for critical workloads. Anthropic plans to use Trainium and Inferentia chips and help develop future versions.
Composition: Sulat. Logos: Anthropic and AWS, Amazon’s cloud subsidiary.
The full sum is not an immediate payment. Reuters reported, after interviewing both companies’ chief executives, that the initial investment is US$1.25 billion, with either company able to trigger a further US$2.75 billion. Amazon will not receive a board seat, according to the report. Anthropic’s existing Google relationship will continue.
The companies also promise future Anthropic models through Amazon Bedrock, plus early access to customisation features. Anthropic says its governance remains unchanged. Neither announcement declares Bedrock generally available.
AWS introduced Bedrock in limited preview in April as a managed way to access foundation models. Customers could build applications through an API rather than operate the underlying model infrastructure themselves.
Claude 2 is already a product: Anthropic launched it in July through an API and a public chat beta in the US and UK. Its announced input limit is 100,000 tokens per prompt, enough to work with lengthy documents. Anthropic also cautioned that the model can produce inappropriate responses. This investment announcement is not a new Claude release.
Discover more from Sulat AI & Tech News
Subscribe to get the latest posts sent to your email.
