Amazon brings AI, chip and quantum teams under Peter DeSantis
Amazon CEO Andy Jassy has reshaped senior engineering leadership, placing AI models, chips and quantum research under Peter DeSantis, as Alexa and Nova leader Rohit Prasad prepares to depart.
Andy Jassy, Amazon’s chief executive, has announced a significant reshuffle of the company’s technical leadership that places long-time executive Peter DeSantis in charge of an expanded AI remit covering foundation models, custom chips and quantum research.
In a message posted on the company blog, Jassy said he had asked DeSantis to lead a new organisation responsible for Amazon’s most expansive AI models, including the Nova family, its silicon efforts such as Graviton and Trainium, and quantum computing.
The note highlighted DeSantis’s 27 years at Amazon and recalled his role in launching Amazon EC2 and later acquiring and running the team from Annapurna Labs that builds custom silicon.
The changes also named Pieter Abbeel as leader of the frontier model research team within what Amazon has described internally as its AGI work.
The blog post confirmed that Rohit Prasad, who helped lead Amazon’s Alexa efforts and more recently directed the creation of the Nova models and the AGI organisation, will leave the company at the end of the year.
Amazon framed the move as a response to an inflection point in which models, chips and cloud infrastructure must be optimised together to capture technical and commercial advantage. The company pointed to the recent launch of Nova 2 models at re:Invent and to rapid progress on custom silicon as drivers of the reorganisation.
“At Amazon, we often start new businesses in parts of the company where there’s an initial customer need, and as they grow and get momentum, we assess where they’re best situated to maximise potential for customers and Amazon over the long term. I believe we are at this inflection point with several of our new technologies that will power a significant amount of our future customer experiences,” Jassy wrote.
The restructure sits within a wider industry trend where major technology firms concentrate AI authority in a small number of senior leaders to speed decision making and knit together research, product and infrastructure.
Google has, in recent months, repositioned key DeepMind and Google Brain figures and created senior roles to accelerate product integration of large models. Microsoft has placed AI near the top of its executive agenda, with senior technical leaders overseeing both research and cloud-based deployment. Apple has also recently moved its machine learning and AI leadership, with John Giannandrea exiting a senior role as the company rethinks its approach to foundational models and product road maps.
Organisations that control chips and cloud platforms, as well as model research, can iterate faster and lower costs by co-designing hardware and software. Amazon’s explicit linking of Nova models, Graviton and Trainium chips and its global cloud infrastructure under a single leader makes that strategy plain.
Concentrated power over model design and deployment heightens the importance of governance, safety protocols and external oversight. It also raises questions about talent continuity when senior researchers move on.
Amazon’s note thanking Prasad for building teams and models that are already used by thousands of firms suggested both recognition and the need to manage transitions carefully.
The changes are likely to be felt through tighter integration between AI services and Amazon Web Services, potentially faster access to new chips and cost improvements when training or running large models.
Firms that can align research and infrastructure at scale will likely be better placed to compete in both enterprise and consumer AI markets.
Edited by Jyoti Narayan