Amazon AI Strategy: CEO Andy Jassy Three-Layered Approach

AI Summary: CEO Andy Jassy views AI as the biggest technology transformation since the internet and cloud, outlining the company's "very substantial investment" in a comprehensive three-layered strategy spanning the entire AI stack. This approach includes building foundational tools for model builders (like Trainium chips and SageMaker), providing services for customizing existing models (via Bedrock), and integrating AI into a thousand different applications across Amazon's businesses and customer experiences.

May 13 2025 15:01
In a recent conversation with Harvard Business Review, Amazon CEO Andy Jassy outlined his vision for how the tech giant is positioning itself in the rapidly evolving AI landscape. Far from being late to the AI party, as some analysts have suggested, Jassy revealed a comprehensive three-layered strategy that spans the entire AI stack and represents what he describes as "a very substantial investment in the AI space."

For Amazon, AI isn't just another technology trend—it's a fundamental transformation that will reshape every aspect of business and daily life. "AI is probably the biggest technology transformation since the cloud," Jassy emphasized. "It's probably the biggest technology transformation since the internet."

This perspective helps explain why Amazon is approaching AI with such methodical deliberation. Rather than rushing a consumer-facing chatbot to market in response to competitors, the company has built an extensive AI infrastructure designed to serve multiple levels of the market, from sophisticated model builders to everyday application users.



According to Jassy, Amazon's AI strategy breaks down into three distinct layers, each addressing different market needs and technological challenges:

• Layer 1: Tools for Model Builders - The foundation layer, focused on the computational infrastructure and services that make AI development possible
• Layer 2: Services for Model Customization - The middle layer, designed for those who want to leverage existing frontier models without building from scratch
• Layer 3: Application Development - The top layer, where AI capabilities are integrated into specific customer-facing solutions

This structured approach gives Amazon a comprehensive position in the AI ecosystem, even if some of its investments haven't generated the same headlines as standalone AI applications like ChatGPT.


At the foundation of Amazon's AI strategy lies the infrastructure that makes advanced AI development possible. This layer targets organizations building their own AI models from the ground up—a technically complex and resource-intensive process.

At that bottom layer of the AI stack is for model builders, and what they care about is two things. They care about the compute to do the training and the inference, which is really the chip. And they care about services to make it easier to build models.

On the hardware front, Amazon has developed its own custom AI chips called "Trainium," designed to reduce the substantial costs associated with training large models. This positions Amazon as a key player in the AI chip market alongside established players like NVIDIA.


Meanwhile, Amazon's SageMaker platform has become "the standard way for people building their own models to get the data in, to build a model, to experiment, and to deploy into production," according to Jassy. This combination of specialized hardware and software tools creates a foundation for organizations with the expertise to build customized AI models tailored to their specific needs.


Not every organization has the resources or expertise to build AI models from scratch, which is why Amazon's second strategic layer focuses on making existing "frontier models" accessible and customizable.

At that middle layer is for people that don't want to have to build their own models. They want to leverage an existing frontier model. They want to customize it with their own data.

This approach acknowledges the reality that while organizations may not want to build models from scratch, they still need ways to adapt general-purpose AI to their specific business contexts and data.

Amazon's Bedrock service embodies this middle-layer strategy, offering "the largest selection of those leading third-party frontier models, including our own," along with tools for guardrails, knowledge integration, and task automation—what Jassy calls "agentic capabilities."


By providing access to multiple models through a single interface and adding capabilities for customization and control, Amazon positions itself as a neutral platform in the increasingly competitive AI model landscape. This approach contrasts with companies that are focused primarily on promoting their own proprietary models.


The most visible layer of Amazon's AI strategy focuses on the applications that deliver AI capabilities directly to users. While Amazon has developed Q, which Jassy describes as "the best AI-powered coding assistant," the company's application strategy is more notable for its breadth than for flagship consumer products.

"We have over 1,000 generative AI applications across Amazon that we've built or are building," Jassy revealed. This approach emphasizes integrating AI throughout Amazon's existing product lines rather than creating standalone AI experiences.

One prominent example is Rufus, Amazon's generative AI-powered shopping assistant. Jassy frames Rufus as addressing a fundamental gap in online shopping—the absence of the personalized guidance traditionally provided by in-store sales associates.


Rufus aims to bring conversational shopping experience online, allowing customers to narrow options, compare products, and ask detailed questions. But Amazon's AI application strategy extends well beyond consumer-facing tools like Rufus to include inventory management, supply chain optimization, and even apparel sizing recommendations powered by foundation models.


Amazon's approach to AI represents a distinctly different strategy from companies that have led with flashy consumer applications. While competitors like OpenAI, Google, and Microsoft have emphasized standalone chatbots and creative tools, Amazon has focused on building a complete AI infrastructure stack.

This approach may have initially created the perception that Amazon was lagging in AI development. As Jassy noted:

In the early days of people getting excited about generative AI, people forgot that if you really want to pursue AI in earnest, there are three macro layers of that AI stack, all of which are gigantic, all of which we're investing in.

The numbers suggest this strategy is already bearing fruit. Jassy revealed that Amazon has "a multibillion dollar annual revenue run rate in the AI space," primarily from the infrastructure and platform layers that have received less public attention than consumer applications.

However, Jassy also sees AI as potentially democratizing technical skills. "I think the number of people who are going to be able to be software developers is going to go up exponentially," he predicted, "because you're going to have these coding apps that allow you to use natural language to describe what you want to go build."

The challenge, in his view, is ensuring education systems adapt to prepare people for this new AI-powered economy. He pointed to concerning educational trends in the United States, noting that "the quality of education in the country has really suffered over the last 20 to 30 years," with the U.S. now ranking "30 out of 35 developed countries... in efficacy of education."


Amazon's three-layer AI strategy offers valuable insights for organizations developing their own approaches to AI:

• Think in layers - Different stakeholders have different AI needs, from infrastructure to applications
• Build for integration - AI is most valuable when embedded in existing workflows and products
• Balance innovation and practicality - Focus on solving genuine customer problems rather than chasing technological novelty
• Prepare for organizational change - AI will transform not just products and services but how teams work and collaborate

For Amazon, AI represents both a major technological shift and a return to first principles. The company's approach to AI reflects many of the core values Jassy highlighted as essential to maintaining a startup mindset at scale: customer obsession, builder culture, ownership, speed, and willingness to take risks.

Perhaps most importantly, Jassy's perspective suggests that AI should be viewed through the lens of customer problems rather than technological capabilities. As he put it:

In whatever we build, whatever we spend resources on, we have to make sure we're solving a real customer problem.