Amazon to sink $4B into AI dev Anthropic, become its cloud provider

Amazon is strengthening ties with OpenAI rival Anthropic via an investment of up to $4 billion in the company, as AWS becomes the primary cloud provider and the home for Anthropic's foundation models.

Megacorp Amazon expects this expanded agreement with the AI developer to speed development of Anthropic's future foundation models, and make them more accessible to customers of Amazon's AWS cloud platform.

The $4 billion investment makes Amazon a minority owner of Anthropic, with AWS becoming Anthropic's primary cloud provider for mission critical workloads, including its safety research and future foundation model dev work.

Amazon is following Microsoft's lead in taking a stake in OpenAI to drive its own ambitions. The Redmond giant has poured billions into OpenAI, which gave it an exclusive license to incorporate OpenAI's GPT-3 model into its own tech and applications. Anthropic was founded by former OpenAI employees.

As for Anthropic, it plans to run the majority of its workloads on AWS from now on. It will make use of Amazon's Trainium and Inferentia AI accelerators to build, train, and deploy future foundation models. The two companies will also work together on future Trainium and Inferentia tech.

The AI dev has also agreed to give AWS cloud users access to future generations of its foundation models via Amazon Bedrock, the service it launched earlier this year for building and operating generative AI applications.

Support for Amazon Bedrock is expanding, said Anthropic, because of "significant AWS customer demand for Claude," the large language model (LLM) chatbot that Anthropic is best known for. This will include secure model customization and fine-tuning on the service to enable enterprises to optimize Claude's performance with their knowledge, while limiting the potential for harmful outcomes, it claimed.

"We have tremendous respect for Anthropic's team and foundation models, and believe we can help improve many customer experiences, short and long-term, through our deeper collaboration," said Amazon CEO Andy Jassy in a canned statement.

Amazon claimed that it is offering services at all three layers of the generative AI stack, starting with compute instances backed by Nvidia GPUs as well as its own custom silicon chips at the bottom, a selection of foundation models available via Bedrock in the middle, and generative AI applications and services such as Amazon CodeWhisperer at the top.

• Don't worry, folks. Big Tech pinky swears it'll build safe, trustworthy generative AI
• Despite the hype, generative AI is not a significant chunk of enterprise cloud spend
• Big Tech pumps $235M into AI model depot Hugging Face
• OpenAI snaps up role-playing game dev as first acquisition

Anthropic said that both it and Amazon are committed to the safe training and deployment of advanced foundation models, and that Amazon taking a minority stake in the company does not change its corporate governance structure. ®