How Amazon’s $20B Chip Business and $200B AI Bet Toppled Walmart on the Fortune Global 500

For the first time in more than a decade, the company with the world’s highest revenue is no longer a retailer with physical aisles. In 2026, Amazon ended Walmart's 12-year grip on the Fortune Global 500 No. 1 spot—and its 13-year streak atop the Fortune 500—by posting $716.9 billion in annual revenue, narrowly beating Walmart’s $713.2 billion . But the headline figure only hints at the deeper story. Amazon’s rise was powered by a $20 billion custom-silicon business, a potential breakthrough into merchant chip sales, and a nearly $200 billion capital-expenditure plan aimed squarely at AI infrastructure. Each of those developments is reshaping not just Amazon’s fortunes but the entire technology landscape.

The shift at the top of the Fortune Global 500

Amazon’s ascent to No. 1 on both the Fortune 500 (U.S.) and Fortune Global 500 (worldwide) marks a generational change in corporate rankings. Walmart had held the global top spot for 12 consecutive years and the U.S. top spot for 13 years. Amazon’s $716.9 billion in fiscal 2025 revenue—driven by 12% year-over-year growth—edged past Walmart’s $713.2 billion . The company debuted on the Fortune 500 in 2002 at No. 492; its climb to the summit in just over two decades is one of the fastest ascents in the list’s 72-year history . Only four companies have ever held the No. 1 Fortune 500 title: General Motors, ExxonMobil, Walmart, and now Amazon .

Custom silicon hits a $20 billion run rate

Amazon’s in-house chip division—spanning Graviton CPUs, Trainium AI accelerators, and Nitro DPUs—surpassed a $20 billion annualized revenue run rate in the first quarter of 2026, growing triple digits year over year . CEO Andy Jassy disclosed the milestone on the Q1 2026 earnings call (April 29, 2026), adding a startling comparison: if the chip business operated as a standalone merchant semiconductor company selling to outside buyers, its annual revenue run rate would be closer to $50 billion—larger than Broadcom’s AI chip franchise ($34 billion) and AMD’s entire data center segment ($23 billion) .

Demand for Trainium chips alone is staggering. The backlog of committed orders stands at $225 billion, with Anthropic and OpenAI signing up for gigawatts of capacity . Trainium2 is completely sold out, Trainium3 is nearly fully subscribed, and Trainium4 is already taking reservations roughly 18 months before it ships .

Amazon may sell Trainium chips to third parties

For a decade, Amazon’s custom chips were available only through Amazon Web Services (AWS), with customers paying for cloud-based usage rather than owning the hardware. That exclusive model is now being tested. AWS AI chief Peter DeSantis confirmed to Bloomberg (June 18, 2026) that Amazon is in early-stage talks to sell Trainium chips and full server racks directly to third-party data centers . “There’s so much underconsumption in AI,” DeSantis said at the VivaTech conference in Paris, framing external sales as a natural response to overwhelming demand rather than a sacrifice of cloud revenue .

The discussions are still exploratory and no agreements have been reached, but the strategic break is unmistakable. Selling Trainium as a merchant product—rack-level systems that customers install in their own facilities—would put Amazon in direct competition with Nvidia’s dominant GPU line . CEO Andy Jassy had previously hinted at the possibility in his April 2026 shareholder letter, calling it “quite possible” that Amazon would sell racks of chips to third parties “over the next couple of years” .

A nearly $200 billion AI-infrastructure bet

Amazon announced a capital-expenditure plan of approximately $200 billion for 2026, a more-than-50% increase from the roughly $130 billion it spent in 2025 . The bulk of the spending is directed at AI infrastructure: data centers, custom chips, servers, networking gear, and robotics. The figure exceeded Wall Street estimates by roughly $50 billion, and shares dropped 11.5% in after-hours trading on announcement day, reflecting investor unease about the pace of spending .

Jassy defended the scale by emphasizing that the capital is backed by committed customer demand rather than speculative bet-placement. “We’re not investing about $200 billion in capital expenditures in 2026 based on a mere guess,” he told CNBC, noting that Amazon has secured large, multi-year, multi-gigawatt Trainium commitments from customers . The company reinvests nearly 90% of its operating cash flow into AI infrastructure, a rate far exceeding competitors like Microsoft and Google, who typically reinvest 40–60% .

AWS AI services are accelerating

While the specific claim of a “$15 billion AI services annualized run rate” was referenced in the shareholder letter , the most directly confirmed metric comes from Amazon’s own SEC filing: AWS is growing at 28%—its fastest rate in 15 quarters—on a very large base . AWS segment sales grew 24% year over year to $35.6 billion in Q4 2025 alone . Amazon’s massive AI infrastructure spending is monetizing faster than many analysts expected, with the company installing capacity as quickly as it can build it .

What the combination means

Amazon’s 2026 Fortune Global 500 victory is not a one-dimensional story of e-commerce growth. The company is being rewarded across three parallel engines: its core retail business hit $716.9 billion in revenue; its custom silicon operation became a top-three data-center chip franchise globally; and its AWS AI buildout—funded by a record capex cycle—is producing the kind of accelerating revenue growth that justifies the investment. Together, those forces ended Walmart’s 12- to 13-year grip on the top of both the Fortune 500 and the Fortune Global 500. As Fortune itself noted, “the biggest headline: Amazon dethrones Walmart at No. 1, ending a 13-year reign” .