Amazon has unified its artificial intelligence (AI) operations under a single leader.
Peter DeSantis is a 27 year AWS veteran who has been appointed to head a newly-unified team that includes AI model design, custom chip development and quantum computing research.
This move is a strategic restructuring aimed at increasing Amazon’s ability to directly compete with OpenAI and Microsoft, as well as Google in the race for AI supremacy.
DeSantis will be reporting directly to Andy Jassy as CEO, highlighting the strategic significance of the project for the future of the company.
This announcement is made in conjunction with Rohit’s departure planned for the end of this year.
Prasad will leave Amazon after leading the most ambitious AI effort for two years. He led the AGI team at Amazon and was responsible for the development of Nova’s foundation models.
Amazon’s departure indicates that the company views consolidation as a way to compete with rivals, who already have integrated AI capabilities into infrastructure and silicon.
Amazon’s Playbook: Combining AI, quantum and chips
Jassy described the reorganization in response to an Amazon “inflection” point for its AI capabilities.
DeSantis is now at the crossroads of three key technical layers, including the Nova line, Annapurna Labs Trainium and Graviton Processors and quantum technologies.
Amazon is attempting to take more control of its AI infrastructure. This integration vertically mirrors Microsoft’s strategy, which integrates OpenAI with Azure Infrastructure.
DeSantis is a man of impressive credentials.
He has led the launch of Amazon EC2 in 2006, spearheaded Annapurna Labs’ acquisition in 2015, and overseen data centers infrastructure spanning global regions encompassing 38 geographical areas and 120 availability zones.
He also appointed Pieter Abbeel as the head of frontier models research in the AGI division. Abbeel is described by the company as “one the world’s foremost AI researchers” and co-founder at robotics startup Covariant.
It is important to be on time.
Amazon announced the Nova 2 model just a few days prior to its re-Invent conference.
This is what it means for the AI Race
Amazon believes that integrating infrastructure gives them a competitive edge.
Amazon hopes to maximize performance by combining model development, custom chips and research into quantum technology under a single executive. This will reduce dependence on Nvidia’s GPUs while controlling margins.
The strategy is similar to that of cloud computing where AWS’s Nitro system was a major differentiator. It had been designed and developed in-house, but it became an essential part of every AWS Server.
The demand for AI infrastructure is outpacing the supply of AI infrastructure.
Observers in the industry believe that DeSantis’s decision could change competitive dynamics, if Amazon is able to bridge its hardware and technology innovation cycles.
AWS commands around 30% of the market for cloud computing and is leading in traditional machine-learning workloads. However, it lags behind when generating AI cases are compared to market share.
This gap could be narrowed by combining DeSantis’s technical strength with Nova’s advantages in cost and Trainium’s growing traction.
Amazon has also been reported to be in discussions about investing at least $10 Billion in OpenAI. This could include OpenAI adopting Amazon’s Trainium chip, an example of a symbolic win that would demonstrate customer confidence in Amazon’s silicon.
The post Amazon appoints new AI chief to compete with OpenAI, Google & Microsoft could be updated as more information becomes available
