Amazon has committed to invest up to $4 billion in Anthropic, an AI startup founded by siblings who formerly worked for OpenAI in the development of ChatGPT.

Anthropic has created its own AI named Claude, which was designed to handle large requests involving exacting data analysis.

Claude also differentiates itself by being able to check and revise its answers to questions without humans needing to tell it to. That makes Claude safer and more reliable than other AIs, many of which are known to make stuff up, Anthropic says.

Anthropic has endorsed Amazon’s Bedrock programming platform that enables companies to customize AIs from various developers using a single interface.

Amazon has developed two AI-dedicated chips. Inferentia is designed to speed machine learning in the cloud; Trainium speeds training of AIs. Trainium in particular will compete with chips from Nvidia, which now owns 80 percent of the market for AI chips but is struggling to keep up with demand.

Under the deal, Amazon is taking a minority ownership share of Anthropic. Google already holds 10 percent of the company in exchange for a $300-million investment made earlier this year.

Amazon will build Anthropic’s technology into several of its products, including its Bedrock service, while Anthropic will use Amazon’s chips to design, train, and deploy future Claudes and other products. 

The partnership will strengthen Amazon as a power in the cloud services market and will help it gain yardage in its quest to become a player in AI.