Why Amazon May Be the Smartest Long-Term AI Investment Nobody Is Talking About

Artificial intelligence has produced no shortage of headline-grabbing stories. Every week brings another breakthrough model from OpenAI, Anthropic, or Google, while NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) dominates the conversation around the chips powering it all.

Yet history shows the companies creating the most value aren’t always the ones making the most noise. During the cloud computing boom, Amazon (NASDAQ:AMZN) quietly built Amazon Web Services (AWS) into a business that now generates tens of billions of dollars in operating income each quarter. The same dynamic may be unfolding in AI, where Amazon’s biggest advantage isn’t building the best chatbot. It’s becoming the platform where businesses deploy them.

Bedrock Is the AI Platform Most Investors Overlook

During Amazon’s first-quarter 2026 earnings call, CEO Andy Jassy reported that Bedrock “saw 170% growth in customer spend quarter over quarter and processed more tokens in Q1 than all prior years combined.” That isn’t just a usage milestone. It signals that enterprise AI adoption has moved from experimentation into full production.

Bedrock isn’t a large language model competing with ChatGPT or Gemini. It’s a managed platform that lets businesses access multiple foundation models, including Anthropic’s Claude, Amazon’s Nova, Meta Platforms (NASDAQ:META) Llama, and others, through a single interface while AWS handles security, governance, and infrastructure. Since that Q1 report, Amazon has also made OpenAI’s models available on the platform, including its flagship GPT-6 Astra, which became generally available on Bedrock in September 2026.

Amazon isn’t trying to convince customers that one AI model is best. The bet is that enterprises will want the flexibility to choose whichever model works best for each task, and that Bedrock becomes the easiest place to do that choosing.

That strategy mirrors what AWS did in cloud computing. Companies didn’t choose AWS because Amazon built the best database or operating system. They chose it because AWS became the easiest place to run almost everything. Bedrock is positioning for the same kind of stickiness in AI.

Amazon Is Competing for the Most Valuable Layer of AI

The AI market is separating rapidly into distinct layers.

Unlike OpenAI or Anthropic, Amazon doesn’t need to win the race to build the smartest model. It only needs to become the preferred platform where enterprises deploy AI applications. The numbers suggest that strategy is working. AWS posted $42.2 billion in Q2 2026 revenue, up 37% year over year, the segment’s fastest growth rate in 18 quarters and an acceleration from 28% growth in Q1. The division now runs at a $169 billion annualized pace and contributed nearly 61% of Amazon’s total operating profit in the quarter, despite accounting for only about 21% of company revenue.

The AI piece of that picture is growing even faster than the overall division. By Q2 2026, both Amazon’s AI business and its custom-chip business had each surpassed $25 billion in annualized revenue, more than doubling year over year. That is a meaningful upgrade from the $15 billion annualized AI run rate Jassy had cited on the Q1 call just three months earlier. The contracted backlog underpinning that trajectory stood at $364 billion as of Q1 2026, with Trainium chip commitments alone exceeding $225 billion.

As more companies deploy AI agents capable of completing multi-step tasks, inference demand, the computing required every time an AI model generates a response, keeps expanding. Every inference request creates demand for GPUs, networking equipment, memory chips, and cloud infrastructure, all of which deepen AWS’s ecosystem.

Investors May Be Looking in the Wrong Place

Amazon doesn’t receive the same attention as NVIDIA’s chips or OpenAI’s newest model releases. But enterprise customers typically care less about who built the model than whether their applications run securely, reliably, and at scale. That is precisely where Bedrock fits. By the first quarter of 2026, Bedrock was already serving roughly 80% of Fortune 100 companies and more than 125,000 customers in total, a breadth of adoption that reflects genuine production use rather than pilot-stage curiosity.

Amazon’s willingness to support multiple competing AI models, including rival OpenAI’s, has become one of its most defensible advantages. Businesses gain flexibility without locking into a single vendor, and Amazon collects platform revenue regardless of which model customers ultimately choose. That dynamic became even clearer in early 2026 when Amazon and OpenAI struck a broad partnership valued at more than $100 billion over eight years, bringing OpenAI’s full model lineup to Bedrock and validating the platform’s reach among the AI industry’s biggest players.

Key Takeaway

Amazon may not generate the flashiest AI headlines, but it is building ownership of one of the industry’s most valuable pieces: the enterprise platform where AI applications are built and deployed. The Q2 2026 results, with AWS growing at its fastest pace in 18 quarters and the AI business crossing a $25 billion run rate, suggest that strategy is moving well past the early-innings stage.

The better frame for investors is to think of Amazon less as another participant in the AI race and more as the company building the digital highway that many of the race’s winners will need to travel. If enterprise AI adoption continues accelerating, Bedrock could become as foundational to artificial intelligence as AWS became to cloud computing. At the pace the numbers are moving, that outcome looks less like speculation and more like a developing reality.

Editor’s note: This article has been updated to reflect Amazon’s Q2 2026 earnings results, including AWS revenue of $42.2 billion (up 37% year over year) and the disclosure that both Amazon’s AI business and its custom-chip business each surpassed $25 billion in annualized revenue. It also incorporates the Amazon-OpenAI partnership, the addition of OpenAI’s GPT-6 Astra to Bedrock, and the updated 2026 capital expenditure guidance of $220 billion.

Contact [email protected] for any questions or corrections.