Amazon unifies AGI, AI chips and quantum under Peter DeSantis

Amazon’s reorganization is a clear signal that it wants to compete at the very top of the frontier-model stack, not just sell generic cloud capacity. By putting Nova, the AGI team, custom silicon (Graviton, Trainium, Nitro) and quantum computing under Peter DeSantis, Andy Jassy is explicitly betting that tight integration between models, chips and infrastructure is how you win the next decade of AI. The move also elevates Pieter Abbeel, one of the field’s most respected researchers, to lead frontier model research inside Amazon, giving the company more academic credibility as it chases OpenAI, Google DeepMind and Anthropic. ([aboutamazon.com](https://www.aboutamazon.com/news/company-news/andy-jassy-peter-desantis-amazon-leadership-update/))

For the race to AGI, this is Amazon trying to compress its organizational distance from rivals. Until now, its AI story has felt fragmented: Alexa, Bedrock, Nova, and custom silicon all advanced, but not always synchronized. A single org with a mandate across models, hardware and quantum gives Amazon a chance to close that gap, especially if DeSantis can turn chip advantages into lower training and inference costs. It also marks a passing of the torch: Rohit Prasad, who defined Amazon’s first wave of conversational AI and then its AGI push, is out just as the company enters a more capital‑intensive, high‑stakes phase. Expect more aggressive model roadmaps, deeper integration of Nova into AWS services, and increased pressure on Nvidia as Amazon leans harder on its own Trainium stacks. ([investing.com](https://www.investing.com/news/stock-market-news/amazons-jassy-announces-leadership-changes-for-ai-silicon-development-93CH-4413460?utm_source=openai))