This Is My Favorite AI Chip Stock (By Far) -- and It's Not Nvidia, Intel, or Broadcom

When investors think of chip stocks, three names usually dominate the conversation. Nvidia, of course, sits at the center of the artificial intelligence (AI) build-out. And another name that has been making headlines more recently is Intel, which had an extraordinary month, with shares rising 114% in April alone. Finally, Broadcom has quietly become the custom-silicon partner of choice for hyperscalers.

But my favorite chip stock isn't any of those. In fact, it isn't even classified as a semiconductor company. It's e-commerce and cloud computing giant Amazon (NASDAQ: AMZN) -- and the company's first-quarter results released last week may have been the clearest confirmation yet of why.

Image source: Getty Images.

Hiding in plain sight

Amazon's chip business -- which combines its Graviton (general-purpose central processing units), Trainium (AI training and inference accelerators), and Nitro (network and storage virtualization) silicon -- exited the first quarter of 2026 at an annual revenue run rate above $20 billion.

Further, the business grew nearly 40% sequentially, while year-over-year growth was in the triple digits.

But these headline figures -- as staggering as they are -- are arguably understating the true scale of Amazon's chips business. Speaking on the company's first-quarter earnings call, Amazon CEO Andy Jassy said the run rate "somewhat masks the size."

"If our chips business was a stand-alone business and sold chips produced this year to AWS and other third parties as other leading chip companies do, our annual revenue run rate would be $50 billion," Jassy explained. "As best as we can tell, our custom silicon business is now one of the top three data center chip businesses in the world."

That last point is worth pausing on. In just a few years, Amazon has gone from a customer of the chip industry to a peer of its largest players.

And forward demand looks staggering.

Amazon now holds more than $225 billion in revenue commitments for Trainium alone. Anchor customers Anthropic and OpenAI have signed on for up to 5 gigawatts and approximately 2 gigawatts of Trainium capacity, respectively.

Further, Trainium2 is largely sold out, and "Trainium3, which just started shipping in 2026 and is 30% to 40% more price performant than Trainium2, is nearly fully subscribed, and much of Trainium4, which is still about 18 months from broad availability, has already been reserved," Jassy said during Amazon's first-quarter earnings call.

Driving results for customers and for Amazon's own business

It's not surprising customers are lining up. Amazon's product pipeline is jam-packed with value.

Trainium2 delivers about 30% better price-performance than comparable graphics processing units (GPUs), according to management, and Trainium3 is another 30% to 40% better than Trainium2.

Additionally, every workload that moves to Trainium instead of a third-party GPU reduces what Amazon Web Services (AWS) pays for compute, while preserving the price-performance edge that's attracting customers. Jassy said Trainium will save Amazon "tens of billions of dollars of [capital expenditures] each year and provide several hundred basis points of operating margin advantage versus relying on others' chips for inference."

That tailwind is starting to show up in the cloud business.

But there are risks worth flagging. Amazon's trailing-12-month free cash flow has collapsed to $1.2 billion from nearly $26 billion a year earlier, pressured by an extraordinary capital spending ramp. So, if AI demand softens before this build-out is digested, the math could get uncomfortable. But the gigawatt-scale, multi-year nature of these customer commitments suggests this isn't a typical capital expenditure cycle. And the company's backlog similarly supports the bull case for this cycle to continue. Amazon wrapped up its first quarter with an AWS backlog of $364 billion -- a figure that doesn't include a deal valued at more than $100 announced with Anthropic after the quarter ended.

Of course, shares are no longer the bargain they were earlier this year. Still, a forward price-to-earnings ratio of about 32, as high as it might seem on the surface, may be reasonable given that Amazon has two powerful businesses -- AWS and silicon -- that are seeing accelerating momentum simultaneously.

Nvidia, Intel, and Broadcom each have their own merits. But none looks as compelling to me as Amazon does today. For investors trying to ride the AI hardware wave without paying pure-play chip valuation multiples, this could be one of the best ways to do it -- and last week's results arguably only bolster the bull case further.

Daniel Sparks and his clients have no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Amazon, Broadcom, Intel, and Nvidia. The Motley Fool has a disclosure policy.