Buy, Sell Or Hold AMD Stock?

Part of our AI selloff guide. Start here: what the Nasdaq correction means for your money.

Follow the whole story: every piece in this series is on our AI storyline page.

Image from Pixabay

What Does Everyone Already Know?

AMD walked into its report as one of the hardest-hit AI chip stocks. It closed Tuesday at $483.90, down 24% in a month, 14% in five days and 25% below its 52-week high. Then it beat. Third-quarter revenue was $13.4 billion, up 45%, against about $13.1 billion expected; adjusted earnings were $2.09 a share against about $1.94. Data center revenue hit a record $8.5 billion, up 97%. AMD guided the fourth quarter to about $17 billion, plus or minus 3%, when analysts expected about $16.2 to $16.3 billion. Even the bottom of that range beats the consensus. CEO Lisa Su told analysts she expects revenue to grow more than 80% in 2027 and server revenue to double. After rising more than 4% in early trading Wednesday, the stock closed up only 1.0%, at $488.78.

The quarter and the guide were real beats. The 2027 claim was barely above what analysts already had.

What Does The Data Say?

Here’s the part the headlines skip. Before Tuesday, analysts already expected AMD’s revenue to grow about 77% in 2027, according to consensus estimates compiled by FMP. Management’s “more than 80%” lifts that by a few points. The report nudged the forecast up; it didn’t have to rescue it.

So what fell? The price investors will pay for the same future. Add up 2026 (three reported quarters plus the $17 billion guide) and AMD is on track for about $52 billion in revenue this year; 80% growth makes about $94 billion in 2027. At last quarter’s adjusted profit margin (about 26% of revenue), that’s roughly $15 a share of 2027 earnings.

The drop took the premium out of AMD, not the earnings. One good night put back only a small part of it.

The 52-week high is AMD’s highest close of the past year ($649.42, on Oct 6). Market value at other prices scales Tuesday’s $790B by price. 2026 revenue: Q1 ~$10.3B, Q2 ~$11.5B, Q3 $13.4B plus the $17B guide midpoint. The ~$15 of 2027 earnings assumes the Q3 adjusted net margin (~26%) holds and the diluted share count stays near 1.66 billion; it is our arithmetic, not a forecast from AMD or analysts. Two things in the report deserve a skeptic’s eye: free cash flow was only $554 million in the quarter, about 4% of revenue (a ramp builds inventory before it ships), and the fourth-quarter gross margin guide of about 55% is a point below the third quarter’s 56%.

Was The Drop Overdone?

For AMD, the numbers say yes, if management delivers. At the high, buyers paid about 44 times 2027 earnings for a business growing about 80%; at Tuesday’s close they paid about 33 times for the same business, with a bigger guide behind it. Nothing in the report justified a 25% haircut on its own. That’s the definition of a price move that tells you about other people, not the company.

But “overdone” isn’t the same as “cheap.” AMD has fallen far more than 25% before: about 65% from late 2021 to late 2022, and more than 90% from 2000 to 2002. A stock that more than doubled in a year can give back a lot without anything going wrong at the company, and AMD is still up 95% over 12 months even after the slide.

Does AMD’s Guide Answer The AI Spending Doubt For Every Chip Stock?

Partly. AMD named the buyers: its first Helios rack systems are running at Anthropic, Meta, Microsoft, OpenAI and Oracle. Several are the same companies whose spending plans set off the selloff, and a guide that steps up 27% in one quarter says they’re still writing orders for the fourth quarter.

The market didn’t read it as a verdict for the group, though. On Wednesday Nvidia fell 3.2% while AMD rose 1.0%. That’s fair. AMD is taking share from a smaller base, so it can grow 80% even if total spending slows; Nvidia, far bigger, needs the total to keep rising. Memory and networking stocks answer to yet another number (why chip stocks answer to memory prices). The group’s real test is Nvidia’s report, set for Nov 17.

What Should You Do This Week?

The report didn’t change AMD’s story, so it shouldn’t change a plan that was sound before it. Here’s how to make one, depending on where you stand.

1. If you want to own it: size from the worst drop, not today’s mood. Decide the most you’d lose on one stock without selling in a panic, then divide by AMD’s 2021–22 fall of about 65%. On a $1 million portfolio where a 3% loss ($30,000) on a single name is your limit, that’s a position of about $46,000. If you’d like more than that, the problem is the size, not the stock.

2. Write down your buy price. If ~33 times 2027 earnings at Wednesday’s price feels rich but 28 times would be fair, that’s about $410, some 37% below the high. Put it in a good-till-canceled limit order now, on a calm day, rather than deciding during the next bad one.

3. Or get paid to wait for that price. A cash-secured $410 put expiring Dec 17, 2027, about 13 months out, paid about $62 a share at Wednesday’s close (midpoint of the bid and ask), or about $6,200 per contract. If AMD falls to $410 and you’re assigned, your real cost is about $348, roughly 29% below Wednesday’s $488.78 close and 46% below the 52-week high. You set aside $41,000 per contract (100 shares). The premium is high because AMD swings a lot, and there’s a real chance you’ll be assigned. That’s what you’re being paid to accept.

4. If you’ve held it through a 95% year: rent the downside. A buyer from a year ago paid about $250. Selling now could mean a big tax bill. A collar keeps the shares: buy a $440 put expiring Dec 17, 2027 (about $76 a share at Wednesday’s close, midpoint of the bid and ask) and pay for most of it by selling a $700 call with the same expiry (about $57). The net cost is about $19 a share, or about $1,860 per 100 shares. You lock in at least $440, about 76% above a $250 cost, and give up gains above $700, about 8% over the old high. Ask your tax adviser how a collar is treated for your account.

What you give up: the limit order and the put miss a rally that never comes back to $410; the collar caps the next leg up at $700. A put doesn’t stop the stock from falling through $410. It only means you chose the price and were paid to choose it. Never sell a put you couldn’t pay for if assigned. This is education, not personal advice; options aren’t right for every account.

What’s The Bigger Idea?

We call it Sell the thesis, not the price. A falling price is information about other people: how scared they are, how much they’ll pay. A changed thesis is information about the company. For a month, AMD’s price said “sell,” while the business kept growing fast. The investors who sold in that slide sold the price. Tuesday night showed the thesis hadn’t moved. Write down what would change it for you (a guide that falls short, a big customer that stalls, cash flow that never catches up with sales) and check those, not the chart. It’s one of the principles in our AI selloff guide.

The drop was about the price of AMD’s future, not the future itself. Analysts expected about 77% growth in 2027 before the report, and AMD now says more than 80%; the stock went from about 44 times estimated 2027 earnings to about 33. If AMD delivers, the selloff overshot. It didn’t settle the question for Nvidia and the suppliers, and it doesn’t make AMD safe: it has fallen 65% before. Size for that, pick your price, and let the market come to you.

What Should You Ask Next?

• Nvidia Stock To $129? How far each AI chip stock could fall in a 2022-style slide.

This story is still moving. Follow it with us: each answer raises the next question, and we’ll keep answering them.

How Much Of Your Portfolio Rides On One Earnings Call?

Corrections test concentration, not conviction. If you have $1 million or more invested, see how exposed your portfolio is to the AI trade in a free 15-minute call with a Trefis partner firm.

For questions about this article or underlying research, email help@trefis.com.