The Engine That Beat Walmart Is Now Powering Amazon’s Bet on AI
AWS built the profit engine that dethroned Walmart. Now it’s the infrastructure behind Amazon’s biggest bet yet, and it affects every seller.
Most people know Amazon as the company that out-delivered, out-priced, and out-convenienced Walmart into second place. What fewer people appreciate is that the real weapon was never the retail side. It was a cloud computing division that quietly became one of the most profitable businesses in corporate history, and is now the foundation for Amazon’s AI strategy.
Amazon Web Services turned 20 years old in 2025 and marked the occasion by pushing Amazon to the number one spot on the Fortune 500, ending Walmart’s 13-year run at the top. The symbolism is hard to miss: the same infrastructure business that subsidized Amazon’s retail expansion for two decades is now the company’s most visible bet on the next era of computing.
Ready to Start Growing Your Amazon Brand?
Canopy’s Partners Achieve an Average 84% Profit Increase!
Find out moreAmazon Just Became the World’s Largest Company by Revenue
For the first time in decades, Walmart is not the biggest company in the world by annual sales. Amazon reported approximately $717 billion in 2025 revenue, narrowly topping Walmart’s roughly $713 billion, ending a run at the top that Walmart had held for 13 consecutive years. When the 2026 Fortune 500 publishes, Amazon will be at number one.
The headline number is significant, but the way Amazon got there is the real story. Walmart built its dominance on one thing: retail, executed at massive scale across 10,500 physical stores.
Amazon reached $717 billion by combining its e-commerce marketplace with three faster-growing, higher-margin businesses: AWS cloud computing, advertising, and Prime membership. Each of those layers compounds the others. Prime drives purchase frequency. Advertising monetizes the traffic. AWS generates the profits that fund everything else.
That structure is what makes this milestone different from a simple revenue race. Amazon did not beat Walmart by out-retailing it. It beat Walmart by building businesses that Walmart cannot replicate, and using the profits from those businesses to compete in retail at margins no traditional retailer can sustain.
The division most responsible for that structural advantage is Amazon Web Services. And now it is also the center of Amazon’s AI strategy.
AWS Is Now Amazon’s AI Engine
The same division that gave Amazon retail dominance is now the center of its AI strategy, and the capital commitment behind that strategy is staggering.
Amazon spent roughly $75 billion on infrastructure in 2024, the overwhelming majority directed toward AWS. For 2025, the company committed to $100 billion. Andy Jassy announced that Amazon foresees $200 billion in capital expenditures for 2026, again concentrated on AWS AI infrastructure. Jassy has also stated plans to double computing capacity again by the end of 2027.
For context: AWS’s 2025 AI-specific revenue is already running at a multi-billion-dollar annual rate and growing at triple-digit percentages year over year.
From Chips to Chatbots: What AWS Actually Builds
The AI services AWS offers span the full stack. At the infrastructure level, Amazon has invested in proprietary chips (Trainium2 for training, Inferentia for inference) to reduce dependence on expensive Nvidia hardware. Jassy has been direct about this: “Chips are the biggest culprit. Most AI to date has been built on one chip provider. It’s pricey.” Building custom silicon is how Amazon controls cost at scale.
Above the hardware layer, AWS offers Amazon Bedrock, which allows companies to access and deploy foundation models, including Amazon’s own Nova models, along with models from Anthropic (in which Amazon has invested $4 billion), Meta, and others. Amazon SageMaker provides the environment to train and manage custom models. Amazon Q handles enterprise automation use cases, including migrating legacy applications from Windows to Linux and moving workloads from mainframe to cloud.
For ecommerce sellers and operators, this matters in concrete ways. The AI features Amazon has introduced in its retail environment, including Rufus (its AI shopping assistant), AI Shopping Guides, and automated listing optimization tools, run on the same AWS infrastructure. What Jassy describes as “building the key primitives for AI development” at the infrastructure layer flows downstream into the tools that touch Amazon customers and sellers directly.
What This Means for the Competitive Landscape
The AWS-as-AI-engine story has direct implications for anyone competing or collaborating within Amazon’s orbit.
First, Amazon’s AI investment is not speculative. It is backed by $45+ billion in annual operating income from a division that has never lost its growth trajectory. When Jassy describes AI as “a once-in-a-lifetime business opportunity,” he is making that statement from a position of structural financial advantage that no traditional retailer can replicate.
Second, the inference cost curve matters for sellers. Jassy has noted that while model training accounts for a large share of current AI spend, inference (the cost of running AI models in production) will represent the overwhelming majority of future AI cost. As AWS drives efficiency improvements in this area through model distillation, prompt caching, and hardware advances, AI-powered tools across Amazon’s ecosystem become cheaper to run. That means more AI features, deployed faster, at lower cost.
Third, AWS’s position as the world’s largest AI infrastructure provider creates a significant moat. Enterprises that store their data in AWS and build AI workflows there will not move easily. That stickiness compounds over time, generating recurring revenue that supports continued infrastructure investment.
The company that turned cloud computing into a retail weapon is now turning AI infrastructure into a cloud weapon. For operators building businesses on Amazon’s platforms, that trajectory shapes the tools, policies, and opportunities that define the next several years.
How Canopy Management Can Help
Ready to partner with a team that has the systems and expertise to scale your brand?
Canopy Management delivers end-to-end eCommerce growth, leading the industry in Amazon marketplace strategy while powering expansion through Shopify, Meta, and Google. Our full-funnel approach — from marketplace optimization to customer acquisition — has generated over $3.3 billion in partner revenue and made us the trusted growth engine for brands worldwide.
Schedule a strategy session with our team to discover exactly how our proven frameworks can accelerate your growth.
Frequently Asked Questions
AWS generates roughly half of Amazon’s total operating profit despite representing a fraction of its total revenue. Those profits funded the sustained investment in logistics, pricing, and Prime benefits that gradually eroded Walmart’s retail advantages. Walmart has no equivalent high-margin business to cross-subsidize its retail operations at the same scale.
AWS provides AI capabilities across three levels. At the infrastructure level: proprietary training chips (Trainium2) and inference chips (Inferentia) plus high-performance networking. At the platform level: Amazon Bedrock for accessing and deploying foundation models, SageMaker for building and training custom models. And finally at the application level: Amazon Nova (Amazon’s own foundation models), Amazon Q for enterprise automation, and Rufus for consumer-facing shopping assistance.
CEO Andy Jassy has described AI as a “once-in-a-lifetime” business opportunity and pointed to demand outpacing available capacity. AWS revenues would have been higher in recent quarters if not for supply constraints, both in power availability and specialized server components. The $200 billion commitment is Amazon’s attempt to get ahead of demand rather than catch up to it.
Yes, though the timeline varies. Amazon’s AI-powered tools, including listing optimization, ad targeting, and shopping features like Rufus, run on AWS infrastructure. As inference costs fall due to hardware and model efficiency improvements, Amazon can deploy more of these tools more broadly. Sellers who understand how these systems work, and optimize for them, will have a structural advantage.
Thinking About Hiring an Amazon Management Agency?
Canopy’s Partners Achieve an Average 84% Profit Increase!
Let’s talk
