Amazon taps Peter DeSantis to lead new AI organization

Amazon is reorganizing key pieces of its artificial intelligence strategy under a newly formed internal group led by longtime cloud executive Peter DeSantis, according to a staff message from CEO Andy Jassy. The move signals a tighter, more centralized approach to building and operating the company’s core AI models, specialized hardware, and advanced computing initiatives that underpin modern AI systems.

In his note, Jassy said the new organization will take responsibility for Amazon’s AI model portfolio—such as the company’s Nova family—alongside custom silicon development and quantum computing. Those areas increasingly determine how quickly AI products can be trained, deployed, and scaled, particularly for enterprise customers that demand performance, security, and cost predictability.

AWS veteran steps into a broader AI mandate

DeSantis has spent 27 years at Amazon and has served for eight years as an SVP at AWS, the company’s cloud division and a major profit engine for the broader business. AWS’s infrastructure is widely used across the internet and has become a central battleground in the race to provide the computing capacity and developer tools needed for AI applications.

The appointment comes after Amazon emphasized enterprise AI at AWS re:Invent, where it highlighted new model releases and infrastructure updates aimed at companies adopting generative AI. Jassy framed the reorg as a way to concentrate leadership and “invention cycles” on the areas that most directly shape AI competitiveness: models, chips, and the cloud stack that delivers them.

Why Amazon is separating this from AWS leadership

While AWS remains the primary channel through which Amazon sells AI capabilities to businesses, the company’s decision to create a separate AI-focused organization suggests it wants faster coordination across layers that often sit in different teams: model research and development, chip design, and the infrastructure software that schedules and optimizes workloads.

That integration matters because the economics of AI are increasingly driven by hardware efficiency and end-to-end optimization. Companies that can tune models to run better on their own chips, and then deliver them efficiently through their own cloud services, can reduce costs while improving performance—an advantage that can be translated into pricing leverage and better margins.

Nova models, silicon, and quantum: what the new org will oversee

Amazon’s new group will oversee the company’s Nova AI models, which have been positioned as part of its broader generative AI offering for customers building applications through AWS. In parallel, it will manage Amazon’s efforts in custom silicon—chips tailored for AI workloads—and quantum computing, a longer-horizon bet that could eventually influence future approaches to complex optimization and computation.

Jassy pointed to the recent launch of Nova 2 models and the rapid growth of Amazon’s chip initiatives as key reasons for the change. The strategic logic is straightforward: the more Amazon can align model design with the capabilities of its chips and the realities of its cloud infrastructure, the more efficiently it can deliver AI services at scale.

• AI models: Core model development and iteration cadence, including Nova releases.
• Custom silicon: Specialized chips designed to power training and inference more efficiently.
• Quantum computing: Advanced R&D that could complement future AI and optimization needs.

Competitive pressure is reshaping Big Tech org charts

Amazon’s move mirrors a broader trend across major technology companies: reorganizing leadership structures to accelerate AI product cycles and reduce internal friction. The market has shifted from experimenting with AI to operationalizing it, and that transition places a premium on reliability, cost control, and rapid deployment—areas where organizational design can be as important as technical breakthroughs.

For Amazon, the stakes are high. AWS has historically dominated cloud infrastructure, but the generative AI boom has intensified competition among cloud providers and AI platform vendors. Customers increasingly evaluate clouds based on access to models, availability of high-performance compute, and the tools that make AI adoption practical in regulated, security-conscious enterprise environments.

Enterprise AI is the near-term prize

At re:Invent and in subsequent messaging, Amazon has emphasized enterprise use cases—where companies want AI embedded into workflows, customer support, analytics, and developer productivity. That market rewards platforms that can offer not only model choice, but also predictable performance and governance.

By placing models, chips, and advanced computing under one AI organization, Amazon is signaling that it views the “full stack” as a single competitive unit. The company’s ability to optimize across that stack may determine how effectively it can compete on both capability and cost as AI adoption expands.

Investments and infrastructure: Amazon’s parallel AI strategy

The reorganization also lands amid heightened attention on Amazon’s AI investments and infrastructure commitments. AWS recently announced a $50 billion investment tied to U.S. government AI infrastructure, underscoring how demand for compute is becoming a strategic national and commercial priority.

Amazon has also been linked to potential large-scale investments in outside AI players, while maintaining a significant stake in Anthropic. Those moves illustrate a dual-track approach: advancing in-house models and infrastructure while also backing external innovators that can strengthen Amazon’s AI ecosystem and customer offerings.

With DeSantis at the helm of the new AI organization, Amazon appears to be betting that tighter coordination among its model roadmap, silicon strategy, and cloud delivery will help it move faster—and compete more effectively—in a market where AI capability, efficiency, and scale are increasingly inseparable.

For customers and partners, the immediate impact will be watched in the cadence of Nova releases, the pace of chip improvements, and how seamlessly those pieces show up inside AWS products over the coming quarters.