No page can tell you which AI stock to buy for your money. This one shows what the data says about ten large, well-known AI-linked US stocks today, what separated past AI winners from losers, and the checks worth running before you buy.
No honest page can name the best AI stocks to buy for you, because the answer depends on what you already own, the price you pay and how large a fall you can sit through. What the data can show is how the biggest AI stocks have behaved, and what separated the AI winners of the past 25 years from the losers.
The theme is real; the stock results are uneven. Even among ten large, well-known AI-linked US companies, returns in the year to Sep 30, 2026 ranged from −50.8% to +537.5%, against +15.7% for an S&P 500 index fund. Owning “AI” did not decide the outcome; in the examples below, the price paid and the size of the falls along the way mattered more.
The general method for any stock idea, with the evidence for each step, is in how to pick stocks to buy now. This page applies it to AI.
Most rose (7 of 10), but not together, and every one of them fell at least 20% from a peak inside the year. The table lists ten large, well-known US stocks tied to AI chips, cloud computing or AI software, ordered by market value. They are the obvious names, not Exent picks and not a recommendation.
| Company | Market value (Oct 22, 2026) | 1-year return to Sep 30, 2026 | Worst fall within that year | Fall in 2022 | Analysts rating Buy (Oct 22, 2026) |
|---|---|---|---|---|---|
| Nvidia (NVDA) | $5.23T | +22.4% | −20% | −62% | 47 of 49 |
| Alphabet (GOOGL) | $4.19T | +41.7% | −21% | −35% | 35 of 43 |
| Microsoft (MSFT) | $3.96T | −0.2% | −34% | −33% | 50 of 53 |
| Amazon (AMZN) | $2.83T | +13.5% | −22% | −40% | 52 of 54 |
| Meta Platforms (META) | $1.80T | −1.1% | −33% | −62% | 44 of 50 |
| Broadcom (AVGO) | $1.61T | +6.9% | −32% | −36% | 34 of 38 |
| Micron Technology (MU) | $1.02T | +537.5% | −30% | −50% | 31 of 33 |
| AMD (AMD) | $902B | +278.1% | −28% | −62% | 34 of 41 |
| Palantir (PLTR) | $447B | +2.5% | −38% | −64% | 18 of 26 |
| Oracle (ORCL) | $392B | −50.8% | −58% | −32% | 36 of 43 |
How to read it. Returns are total returns from month-end prices for the year to Sep 30, 2026; the S&P 500 index fund returned +15.7%. “Worst fall within that year” is the largest peak-to-trough drop inside those 12 months, not the distance from today’s high. “Fall in 2022” is each stock’s peak-to-trough drop in the 2022 selloff. Market value and analyst ratings come from Exent’s data snapshot of Oct 22, 2026.
Two things stand out. First, 7 of the 10 names with five years of history beat the S&P 500 over five years (+90.0% for the index), but the one-year picture was split 4 to 6. Second, analysts made Buy the majority rating on 10 of the ten, so the rating count tells you little; see did analysts see the biggest winners coming.
The winners sold something customers were already paying for at scale, and their owners survived falls of 50% or more on the way. The losers were often priced for a future that arrived late or went to someone else. The table mixes both on purpose.
| Example | When | What happened | What it showed |
|---|---|---|---|
| Cisco (CSCO), the dot-com “picks and shovels” leader | 2000 to 2025 | Fell nearly 90% from its March 2000 peak; regained its $80.06 record close only on Dec 10, 2025 | A real leader bought at a peak price can take 25 years to break even |
| Amazon (AMZN) in the dot-com bust | 1999 to 2001 | Lost more than 90% before becoming one of the largest companies in the world | A future winner can look like a loser for years |
| Nvidia (NVDA) | 2022 to 2024 | −50.3% in 2022, then +239.2% in 2023 and +171.3% in 2024 | The biggest AI winner first lost half; only holders who sized for that kept the gain |
| Meta Platforms (META) | 2022 to 2023 | −64.2% in 2022, then +194.1% in 2023 | A profitable leader can lose most of its value in a year and then recover it |
| C3.ai (AI) | 2021 to Sep 30, 2026 | −77.5% in 2021 and −64.2% in 2022; −76.3% over five years | An AI name and a strong story did not make the stock a winner |
| UiPath (PATH) | 2021 IPO to Sep 30, 2026 | −70.5% in 2022; −75.6% over five years | Buying near a hot listing proved costly |
| Super Micro Computer (SMCI) | 2023 to 2026 | +246.3% in 2023, then a fall of 85% from peak to trough in the three years to Sep 30, 2026 | Booming AI server demand did not prevent the fall |
| Intel (INTC) | 2024 | −59.6% in 2024, the year Nvidia gained +171.3% | Being in the right industry is not the same as winning the orders |
Calendar-year returns are from the same data set, as of Sep 30, 2026. The Cisco figures are from Sherwood News (Dec 11, 2025) and Amazon’s from CNBC (Dec 18, 2018). The whole tech index fell with them: the Nasdaq-100 fund QQQ lost 83% from March 2000 to October 2002, according to Yahoo Finance.
Two lessons repeat. A real technology shift does not make every stock in it a good buy at any price. And the winners were not smooth: Nvidia fell 62% from peak to trough in 2022 and lost 17% in a single day on Jan 27, 2025, a $589 billion drop in market value, according to Bloomberg. More on the supplier side in picks-and-shovels investing.
Nobody can know, because a past rally on its own says neither that a stock is overpriced nor that it will keep rising. What the record does show is that buying after a big run changes the risk you take, not the direction. Nvidia gained +239% in 2023 and +171% in 2024, then +39% in 2025: still a gain, but a smaller one, bought at a far higher price.
The useful questions are about price and size. How much growth does today’s price already assume? What would the stock have to fall to before you would sell? And how much of your portfolio already rides on the same theme through index funds? The data on stocks that doubled is in is it too late to buy a stock after it has doubled, and how far AI stocks fell in past selloffs is in what happens to your portfolio if the AI bubble bursts.
Nobody knows, and lists that claim to know are guesses. In Exent’s Oct 22, 2026 snapshot, Nvidia was the largest US-listed company by market value at $5.23 trillion, with Apple next at $5.02 trillion. The top spot has changed hands several times since 2000.
The odds for any one small company are low: a few stocks create most of the market’s long-run gains, and most do worse than an index fund. What early big winners tended to share, and how often they halved first, is in how to find the next Nvidia. Whether AI and chip stocks as a group beat the index is in did AI and chip stocks beat the S&P 500.
Check the business behind the story, the price, and your own ability to hold it. These six questions cover what separated the examples above.
To see how much of a portfolio already rides on AI, read how much AI exposure is in your portfolio. For a full single-stock review, use how to evaluate a stock idea.
No list fits everyone, and the big AI names had very different years: in the year to Sep 30, 2026, ten large AI-linked US stocks returned between −50.8% and +537.5%. Judge each one on real AI revenue, the price you pay, customer concentration and whether you could hold through a fall of half.
Nobody can know. A past rally alone does not show a stock is overpriced or that it will keep rising; it does mean you pay more for the same business, so the price and the position size matter more.
Nobody knows. Big winners are rare, most stocks trail an index fund over time, and past winners such as Nvidia and Amazon fell by half or more on the way. What to check is real revenue growth, a price that does not assume perfection, and whether you could hold through a deep fall.
No one can say. In Exent’s Oct 22, 2026 data snapshot Nvidia was the largest US-listed company at $5.23 trillion, with Apple next at $5.02 trillion. The order at the top has changed several times since 2000.
Parts of the theme are priced for a lot of future growth, but no measure tells you when or whether a bubble ends. In the dot-com bust the Nasdaq-100 fell 83% and Cisco took 25 years to regain its 2000 high, so size positions for a fall of that kind.
Mostly yes: Buy was the majority rating on 10 of ten large AI-linked stocks in Exent’s Oct 22, 2026 snapshot, so the rating count says little. Their past accuracy says more.
Exent tracks 4,000+ analysts, keeps only the elite few with proven records, and surfaces the Gems where they agree.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.