Amazon has launched a restructuring of its artificial intelligence operations, folding artificial general intelligence, in-house chip development, and quantum computing into a single business group under Peter DeSantis, a 27-year AWS veteran.
Citing Bloomberg, Reuters, CNBC, and CRN, Amazon Chief Executive Andy Jassy said in an internal memo that DeSantis will oversee the Nova foundation models, the AGI team, and custom silicon programs, including Graviton, Trainium, and Nitro, alongside quantum computing research.
Jassy said the growing convergence of generative AI, in-house silicon, and cloud infrastructure marks a critical inflection point for Amazon, demanding more centralized leadership and faster decisions to sustain technological momentum and improve customer experience.
DeSantis has spent 27 years at Amazon and AWS, previously running AWS infrastructure and utility computing, overseeing global data centers and core cloud services, and playing a central role in the 2015 acquisition of Annapurna Labs that underpinned Amazon's in-house chip strategy.
Market observers say DeSantis's experience in cloud-scale operations and hardware integration could tighten coordination across AI models, computing power, and system architecture—an area where execution, not ambition, will matter most.
Following the reorganization, Rohit Prasad, long responsible for Alexa and AGI research, is set to leave by the end of 2025. Since joining Amazon in 2013, he has led Alexa development and, in recent years, overseen the Nova and Nova 2 foundation models and the buildout of the AGI organization.
Jassy credited Prasad with turning Alexa from an experimental product into a service reaching hundreds of millions of users daily, and with delivering 12 cost-effective foundation models adopted by tens of thousands of enterprises over the past two years.
AI researcher Pieter Abbeel, who joined Amazon in 2024 through the acquisition of robotics software startup Covariant, will lead an advanced model research group within the AGI team, focusing on generative AI and reinforcement learning in collaboration with Amazon's robotics units.
Analysts view the move as an effort to dispel perceptions that Amazon trails rivals in AI. With cloud services, models, and silicon now under one roof, execution will determine whether Amazon can narrow the gap and influence the global AI landscape in the years ahead.
Article edited by Jerry Chen
