Amazon has spent the better part of two years watching OpenAI and Anthropic dominate the conversation around frontier AI models. Now the company is betting that a leadership shakeup, billions in investment commitments, and a unified AI organization can close that gap faster than most observers expect.

A new captain for the AI ship

Rohit Prasad, who previously led the company’s AI efforts, departed Amazon by the end of 2025. In his place, Peter DeSantis now heads a newly unified AI organization designed to eliminate the silos that had slowed Amazon’s progress.

DeSantis’s mandate encompasses both the software and hardware sides of the equation. That includes Amazon’s custom chip efforts, specifically Trainium for training AI models and Inferentia for running inference workloads.

The Anthropic bet keeps getting bigger

Amazon’s investment in Anthropic started at $1.25 billion. That initial check has since ballooned into a commitment that could reach as high as $25 billion, making it one of the largest corporate bets on a single AI company in history.

Anthropic’s valuation has reflected this dynamic. The company reportedly achieved a pre-money valuation of roughly $900 billion, a figure that in some assessments has surpassed OpenAI’s estimated valuation.

Flirting with the competition

Amazon has reportedly been in discussions with OpenAI about potentially substantial commercial partnerships. The estimated investment figures discussed exceed $10 billion, with the goal of integrating OpenAI’s models into AWS Bedrock, Amazon’s managed service for accessing various AI models.

CEO Andy Jassy has also been actively engaged in discussions with US officials regarding AI security matters involving Anthropic.