AI stocks, on this site, are not “every company that mentions AI.” They are a curated list of U.S.-listed companies across the AI supply chain: the chips and foundries that make the hardware, the clouds that train and serve models, and software firms whose core product is AI. MarketCapLens ranks them by market capitalization. The live table, charts, and current combined size are on the AI stocks page.
That is a different job from a sector. Technology is an industry label that also includes consumer gadgets, IT services, and plenty of businesses with little AI exposure. The AI list is smaller, editorial, and allowed to overlap other Featured lists.
Key takeaways
AI is not one industry. Nvidia designs the accelerators. TSMC (listed here as an ADR) fabricates leading-edge chips. Microsoft, Amazon, and Alphabet sell the cloud that trains and runs models. Meta builds open models. Broadcom, AMD, Arm, and Marvell sit in silicon and interconnect. Oracle and Palantir are more software than fab.
A Technology sector ranking would mix those names with companies that barely touch this stack. An AI theme that only listed chat-app startups would miss the hardware bottleneck. The MarketCapLens list tries to show the large U.S.-listed pieces of the chain on one page so you can see how concentrated the winners have become — chips next to clouds next to software — without pretending they are one sector.
Membership is curated. It is not “the N largest tech stocks,” and it is not automatically every semiconductor. Micron is a memory company; it belongs on the Memory list (see What are memory and chip stocks?), not on AI unless we independently decide the AI list should include it. Today it does not. That kind of line-drawing is the point of an editorial list: you can disagree with the cut, and you can still see the rule.
On AI stocks, names are ordered by market cap in U.S. dollars, using the same snapshot as the rest of the site. % of group is each company's share of this list's combined cap. Share of tracked market cap is this list versus every company MarketCapLens ranks. Because the AI list overlaps the Magnificent 7, that “share of tracked” is not a piece of a pie you can stack with Mag7's share.
Alphabet may appear as GOOGL or GOOG; share classes are combined so the company is counted once. TSMC and Arm are U.S.-listed (ADRs / foreign issuers) and still have to clear the same exchange and size tests as everyone else — see how the ranking works.
Returns on the hub are share-price changes, not an AI index. Newer listings do not get a fake five-year history. The page flags names that listed recently so a blank return is not mistaken for a 0% return.
Nvidia, Microsoft, Alphabet, Amazon, and Meta are usually on both lists. Apple and Tesla usually are not. The Magnificent 7 question is: how do these seven mega-caps compare with each other? The AI question is: how does the supply chain look by size? Shared companies are a feature of Featured lists, documented on Featured. Adding combined caps double-counts them.
Read What are the Magnificent 7 stocks? for the fixed seven-name list.
The AI hub also has a smaller section for specialized GPU clouds and related infrastructure — names such as CoreWeave — that rent capacity or build data-center power rather than selling a mega-cap platform. They are shown separately because they are younger, smaller, and more volatile than the flagship list. Multi-year returns may be missing. They are still not a recommendation; they are a second table so the main ranking stays focused on the established chain.
These are traits of the AI supply-chain set, not a buy or sell list.
It maps the stack, not a slogan. Accelerators, foundry, cloud, and software are different businesses. Putting them on one ranking shows where the market is assigning value in the chain — chips versus clouds versus applications — without pretending they are one sector.
Demand has a real bottleneck. Training and running large models needs GPUs, high-bandwidth memory, power, and data-center capacity. Companies that sell those pieces can grow with AI spending even if they never ship a chatbot.
Hardware and software economics both show up. A foundry and an ads-and-cloud giant do not share the same margins or cycle. The list lets you see that mix instead of flattening AI into a single “tech” bet.
The rule is visible. You can disagree with who is in or out. You can still see why Micron sits on Memory and Nvidia sits here.
Concentration is the main risk. A handful of mega-caps — several of them also in the Magnificent 7 — can be most of the combined total. This is not a diversified “AI basket” in the way a 50-stock sector fund is.
The spend can reverse. Cloud and chip customers can slow capital spending. When they do, accelerators, foundries, and GPU landlords often weaken together. That is a cycle wearing an AI label.
The story can outrun the financials. AI attracts rich valuations and fast narratives. A ranking by market cap will look largest exactly where the story is hottest, which is useful — and easy to confuse with “best.”
Younger infrastructure names are a different animal. Specialized GPU clouds and data-center names can be smaller, more tied to one chip generation, and missing a full five-year history. They do not behave like Microsoft. Mixing them mentally with the flagship list overstates how uniform “AI stocks” are.
It is not a complete census of AI, not investment advice, and not “the Technology sector with a new label.” If you want every large software and hardware company regardless of AI, open Technology. If you want DRAM, NAND, and disk drives as a cycle of their own, open Memory stocks.
For general education only. Nothing here is investment advice.