Amazon’s AI valuation puzzle
Wall Street debateingly values Amazon like a mature retailer even as parts of the business may be accelerating on AI — one commentator contrasted Amazon trading around 26x forward earnings with Walmart at 45x while flagging Amazon’s growing AI semiconductor and infrastructure revenue. That’s important because if Amazon’s AI segment scales as claimed, conventional peer comparisons may underprice the company’s optionality. The discussion frames Amazon as a conglomerate with a hidden, high‑growth engine rather than a simple e‑commerce comparitor. (youtube.com)
Amazon is being priced like a big store chain even though one of its divisions now throws off more operating profit than many stand-alone tech companies. As of April 1, 2026, Amazon traded at about 26.97 times forward earnings, while Walmart’s forward price-to-earnings ratio was about 43.10 as of April 9, 2026. (gurufocus.com) (finance.yahoo.com) That comparison looks odd once you split Amazon into pieces instead of treating it like one business. In 2025, Amazon booked $716.9 billion in total sales, and Amazon Web Services alone produced $45.6 billion of operating income on $128.7 billion of revenue. (aboutamazon.com) The retail side is still enormous, but it is not the whole machine anymore. Amazon’s North America segment generated $426.3 billion of 2025 sales and $29.6 billion of operating income, while the international segment added $161.9 billion of sales and $4.7 billion of operating income. (aboutamazon.com) Then there is the cloud division, which is where the artificial intelligence argument starts. Amazon Web Services grew 24% year over year in the fourth quarter of 2025 to $35.6 billion in quarterly sales, and its quarterly operating income reached $12.5 billion. (aboutamazon.com) Wall Street usually values retailers by asking how fast same-store style profits can grow. That lens misses what happens when the same parent company also owns a cloud landlord renting out the computing power needed to build large artificial intelligence models. (aboutamazon.com) (finance.yahoo.com) Amazon has been building its own artificial intelligence chips so it does not have to buy every engine from Nvidia. In February 2026, Amazon said Trainium and Graviton had reached a combined annual revenue run rate of more than $10 billion, with triple-digit year-over-year growth. (ir.aboutamazon.com) Those chips are not just lab projects sitting on a slide deck. Amazon also said Trainium2 was fully subscribed, with 1.4 million chips landed, and that Bedrock, its managed artificial intelligence service, was being used by more than 100,000 companies. (ir.aboutamazon.com) Amazon’s bet gets easier to see when you look at Anthropic, the artificial intelligence company behind Claude. Amazon has invested up to $8 billion in Anthropic, and Anthropic said it was working closely with Amazon Web Services on future Trainium accelerators. (anthropic.com) That partnership turned into a giant physical build-out called Project Rainier. In October 2025, Amazon said the cluster was live with nearly 500,000 Trainium2 chips and expected to expand to more than 1 million chips by the end of 2025. (aboutamazon.com) So the puzzle is not whether Amazon has a retail business. The puzzle is whether investors should keep valuing the whole company mainly like a retailer when one division is a fast-growing cloud platform and another is becoming a seller of the picks and shovels for the artificial intelligence boom. (gurufocus.com) (aboutamazon.com) (ir.aboutamazon.com) If Amazon’s custom chip sales, Bedrock usage, and Anthropic-linked infrastructure keep scaling, the current multiple may be pricing the grocery aisle and the shipping boxes while giving too little credit to the data center in the back. Amazon plans to spend about $200 billion in capital expenditures in 2026, which is the kind of number companies use when they think demand is arriving, not fading. (rallies.ai)
