By Updated 6 August 2026 — AMD (NASDAQ: AMD) closed at $482.05 on Wednesday 5 August 2026, down 7.04% (−$36.53), per StockAnalysis. Market capitalisation is $786.93 billion on a trailing P/E of 123.0 and a forward P/E of 43.7.
Verdict: AMD reported a record quarter — revenue up 50% to $11.5 billion, Data Center up 107% — beat on both lines, guided Q3 above consensus, and fell 7% anyway. The bar was never the published estimate; it was the price. Jefferies raised its target to $650, the Street average sits at $597.99, and the first real technical support is $430.
KEY FACTS
Spot: $482.05, down 7.04% (−$36.53) on 5 August 2026. Market cap $786.93 billion, trailing P/E 123.0, forward P/E 43.7.
52-week range: $149.22 – $584.73. The stock is 17.6% below its high and 223% above its low.
Q2 2026 actuals: revenue $11.5 billion, up 50% year over year; non-GAAP EPS $1.66; GAAP EPS $1.38. Non-GAAP gross margin 56%, non-GAAP operating income $3.1 billion — AMD Investor Relations
The beat: consensus was $11.31 billion of revenue and $1.62 EPS. AMD cleared both.
Data Center: $6.7 billion, up 107% year over year — 58% of company revenue — on EPYC and Instinct demand.
The other segments: Client and Gaming $3.8 billion, up just 6%; Embedded $977 million, up 19%.
Q3 2026 guidance: about $13 billion ±$300 million — roughly 41% year-over-year and 13% sequential growth — with non-GAAP gross margin near 56%. No full-year outlook was given.
Street: average target $597.99 (23.6% above spot). Jefferies raised to $650 from $640; Truist raised to $594 from $478.
Technical support: $430 is the first major level below spot, with a $460–$480 stabilisation zone — TradingKey
What changed since this forecast was last updated
This page previously carried a $640 bull case and a $385 bear case, with AMD at $429.56 and Q2 earnings still ahead. Three things have resolved since.
The print landed, and it was good. Revenue of $11.5 billion beat the $11.31 billion consensus. Non-GAAP EPS of $1.66 beat $1.62. Data Center revenue more than doubled. Q3 guidance of roughly $13 billion came in above where the Street sat. On every published metric, AMD delivered.
The stock fell 7.04% anyway. This is the part worth sitting with. AMD did not miss Wall Street’s targets; it missed the targets embedded in its own share price. The stock had rallied from $429.56 on 29 July to $518.58 by 4 August — a 20.7% run into the print — which meant the beat had to be enormous simply to hold the line. It was merely good.
And the multiple compressed. Here is the datapoint almost nobody has framed correctly: AMD is up 12.2% since the 29 July close, yet its trailing P/E has fallen from 143.2 to 123.0 — a 14% de-rating. The quarter’s earnings entered the trailing twelve-month denominator faster than the price rose. On a forward basis the number is 43.7. Judged on what it earns rather than what it costs, AMD got materially cheaper in the same week the tape said it got worse.
The insight the selloff buries: Data Center is now the entire company
Run the segment arithmetic, because the headline growth rate hides it. Total revenue of $11.5 billion at 50% growth implies a year-ago base of roughly $7.67 billion — so AMD added about $3.83 billion of revenue year over year. Data Center at $6.7 billion and 107% growth implies a year-ago base of roughly $3.24 billion, so that one segment added about $3.46 billion.
Which means Data Center delivered roughly 90% of AMD’s entire revenue growth. Client and Gaming — the PC and console business that was AMD’s identity for forty years — grew 6%. Embedded grew 19% off a $977 million base.
This is the real story of the quarter, and it cuts both ways. The bull reading: AMD has successfully become an AI infrastructure company, and it did so in about eight quarters. The bear reading: there is no longer a second engine. When a company’s growth is 90% concentrated in one segment selling into a handful of hyperscale buyers, the multiple should compress on any wobble in that segment — and a 7% drop on a beat is the market beginning to price that concentration honestly. It is the same structural question we examine at the memory layer in our Micron bull and bear cases.
The bear case just acquired a named, dated catalyst
Until this week the bear case for AMD was abstract: a high multiple, memory-cost inflation, hyperscaler capex discipline. On 4 August it acquired something specific.
On SpaceX’s first earnings call as a public company — the same evening AMD reported — Elon Musk told shareholders that SpaceX will build its AI data centres exclusively on Nvidia silicon. “We think the Vera Rubin architecture is the best architecture. We think it’s the best AI computer, and we greatly value our close cooperation and partnership on many levels with Nvidia,” Musk said, per Yahoo Finance.
Set that against what SpaceX disclosed in the same report: AI revenue of $2.6 billion, up 213% quarter over quarter, with multibillion-dollar data-centre leasing agreements signed with Google and Anthropic. A new hyperscale-class compute buyer has emerged, is scaling faster than almost anything in the sector, and has publicly committed its silicon budget to a competitor — with the word “exclusively” attached.
AMD’s problem has never been demand in aggregate; it is share of the marginal new buyer. Meta’s commitment of up to 6 gigawatts of Instinct GPUs and the 2-gigawatt Anthropic agreement worth up to $5 billion prove AMD can win these deals. SpaceX proves it can also lose them before the bidding starts. The counterweight worth watching: Anthropic is now leasing SpaceX data-centre capacity while separately buying AMD silicon — the buyer relationships are becoming layered rather than exclusive, whatever the headline says. We cover the other side of this trade in our SpaceX SPCX stock prediction.
The bull case: Helios, MI450 and a Q3 already guided to $13bn
The bull case is no longer about AMD being a cheaper proxy for the Nvidia trade. It is about a specific product cycle with named customers and, now, a guided number attached.
Q3 guidance of roughly $13 billion at the midpoint is 41% year-over-year growth and 13% sequential — an acceleration on a base that has already doubled in the Data Center segment. AMD has said customer engagements for the MI450 series and Helios rack-scale systems are running ahead of expectations, with initial volume in Q3, a significant ramp in Q4 and further scaling into Q1 2027. That Q4 ramp is the crux: it is the quarter in which the AI GPU line is supposed to stop being a promise.
Lisa Su framed the quarter directly: “We delivered an excellent quarter, with record revenue and profitability as Data Center revenue more than doubled year-over-year,” said the AMD Chair and CEO in the results release.
The analyst response was notably unbothered by the tape. Jefferies’ Blayne Curtis reiterated Buy and raised his target to $650 from $640, arguing the growth story is intact with AI GPUs ramping significantly in Q4. Truist’s William Stein went further in relative terms, lifting his target to $594 from $478 on expectations that AMD’s 2027 data-centre revenue can more than double, per TipRanks. When the stock falls 7% and the sell side raises targets, the disagreement is about entry price, not about the business.
The bear case: concentration, exclusivity and a 123x trailing multiple
Three things carry the bear case, and none of them requires AMD to miss a number.
The concentration. Ninety per cent of growth from one segment, sold to a small set of buyers with enormous negotiating leverage and their own custom-silicon programmes. Client and Gaming growing 6% means there is no cyclical ballast if AI capex pauses.
The exclusivity risk. Musk’s Nvidia commitment is one data point, but it is the loudest available signal on how the newest large buyers are choosing. The Philadelphia Semiconductor Index has already seen a maximum drawdown of around 30%, and Wall Street is openly scrutinising whether hyperscaler AI capex is converting into revenue — the question that decides the 2027 order book.
The multiple. At 123x trailing and 43.7x forward, AMD still prices the Helios ramp as substantially delivered. AMD also declined to give a full-year outlook, which in a quarter this strong reads as prudence about H2 rather than confidence. The first real technical support is $430 — roughly the 29 July close — with a $460–$480 stabilisation zone above it. A retest of $430 would be an 11% drawdown from here and would not require a single piece of bad operational news.
Scenario table: bear, base and bull from $482.05
Case
Level
vs $482.05 spot
Anchor and what has to happen
Bear$430−10.8%First major technical support, and roughly the 29 July close. Requires only continued multiple compression — no operational miss needed. Triggered faster by any softening of Q4 MI450 ramp language.
Stabilisation$460–$480−4.6% to −0.4%The zone TradingKey flags for a base to form on declining volume. This is where the stock sits now; holding it keeps the uptrend structurally intact.
Street mean$597.99+24.0%Twelve-month consensus, close to Truist’s freshly raised $594. Needs the Q4 Helios ramp to arrive on schedule and at target margin.
Bull$650+34.8%Jefferies’ raised target, above the $584.73 high. Needs the Meta 6-gigawatt and Anthropic 2-gigawatt commitments converting into disclosed revenue, plus 2027 data-centre revenue tracking toward a double.
The bear and stabilisation levels are technical reference points for the weeks after the print, not twelve-month forecasts. The Street mean and Jefferies’ $650 are published twelve-month targets. This page previously carried a $640 bull and $385 bear anchored to the pre-earnings options band; both have been replaced now that the event has resolved and the options premium has collapsed.
A note on this page’s last call
Before the print, this page put AMD’s base case at $482 — the upper half of the options-implied move from $429.56. AMD closed at $482.05. The level was right to within five cents, and the reasoning was wrong: we expected the stock to grind up into that band on a solid quarter. Instead it overshot to $518.58 first and fell back through 7% of downside to land in the same place. Same destination, different road — a reminder that options-implied bands describe the distribution of outcomes, not the path, and that a level being reached is not evidence the mechanism behind it was sound.
What decides the next quarter
The Q4 ramp, quantified. AMD has guided Q3 to roughly $13 billion. The number that matters is what Q4 does, because that is when MI450 volume is supposed to be material rather than initial.
Gross margin at volume. Non-GAAP margin held at 56% and is guided to 56% for Q3. Holding that through the MI450 ramp — against memory-cost inflation — is the single cleanest test of the bull case.
Named-customer conversion. Whether Meta’s 6 gigawatts and Anthropic’s 2 gigawatts start appearing as disclosed revenue or backlog rather than as press releases.
Segment breadth. Any sign that Client and Gaming can grow faster than 6% would materially de-risk the concentration problem. Nothing in this quarter suggests it will.
How AMD sits against the rest of the AI complex
The concentration question and the capex-scrutiny question are hitting the whole chain, not one name. FinanceFeeds tracks the comparable setups in Nvidia’s bull and bear cases, on the memory side in Micron, and at the system-builder layer in Super Micro.
Quick Take
At $482.05, AMD is 17.6% off its high after beating on revenue, EPS and guidance and falling 7% regardless. Data Center more than doubled to $6.7 billion and supplied roughly 90% of all revenue growth; Client and Gaming grew 6%. The trailing multiple actually compressed from 143x to 123x on the print, and the forward multiple is 43.7x. Jefferies raised to $650 and Truist to $594 into the drop. The risk is not the current quarter — it is concentration, and a new hyperscale buyer in SpaceX publicly committing to Nvidia exclusively. Watch the Q4 ramp language and the 56% margin, not the EPS line.
FAQ
What is the AMD stock forecast for 2026?
The consensus twelve-month target is $597.99, about 24% above the 5 August close of $482.05. Jefferies raised its target to $650 from $640 after Q2 earnings and Truist raised to $594 from $478. The first significant technical support below spot is $430.
Why did AMD stock fall 7% after beating earnings?
AMD beat consensus on revenue ($11.5 billion versus $11.31 billion) and EPS ($1.66 versus $1.62) and guided Q3 to roughly $13 billion, above expectations. But the stock had already run 20.7% from $429.56 on 29 July to $518.58 by 4 August, so a merely good quarter was not enough to hold that price. AMD cleared Wall Street’s published targets and missed the expectations embedded in its own valuation.
What were AMD’s Q2 2026 results?
Revenue of $11.5 billion, up 50% year over year. GAAP EPS $1.38 and non-GAAP EPS $1.66. GAAP gross margin 54%, non-GAAP 56%. Data Center revenue of $6.7 billion, up 107%, representing 58% of the company. Client and Gaming $3.8 billion, up 6%. Embedded $977 million, up 19%.
What is AMD’s Q3 2026 guidance?
Roughly $13 billion in revenue, plus or minus $300 million — about 41% year-over-year growth and 13% sequential — with non-GAAP gross margin near 56%. AMD did not provide a full-year outlook.
What is the bear case for AMD stock?
Roughly 90% of AMD’s revenue growth now comes from Data Center alone, with Client and Gaming growing just 6% — leaving no cyclical ballast if AI capex pauses. Add a 123x trailing multiple, the absence of a full-year outlook, and SpaceX publicly committing to Nvidia exclusively for its AI data centres. First technical support is $430, an 11% drawdown that requires no operational miss.
Is AMD cheap after the drop?
It depends which multiple you use. AMD trades at 123x trailing earnings, which is expensive by any conventional measure, but at 43.7x forward earnings — a large gap that reflects how fast earnings are currently growing. Notably, the trailing multiple fell from 143x to 123x across the print even as the share price rose 12.2% from the 29 July close, because the quarter’s earnings entered the trailing calculation.
How does SpaceX’s Nvidia commitment affect AMD?
On its first earnings call as a public company, SpaceX said it will build its AI data centres exclusively with Nvidia chips, with Musk citing the Vera Rubin architecture. SpaceX’s AI revenue grew 213% quarter over quarter to $2.6 billion, making it one of the fastest-scaling new compute buyers. It does not affect AMD’s existing contracts, but it removes a large prospective customer from AMD’s addressable pipeline.
How far is AMD from its 52-week high?
At $482.05 the stock is 17.6% below the $584.73 high and 223% above the $149.22 low.
Sources: price, market-cap, P/E and 52-week range data from StockAnalysis and Nasdaq (5 August 2026 close); Q2 2026 actuals, segment detail, Q3 guidance and the Lisa Su quote from AMD Investor Relations; analyst target changes from TipRanks; technical levels and SOX drawdown from TradingKey; Musk’s Nvidia remarks from Yahoo Finance and SpaceX segment data from CNBC. Segment growth-contribution figures are FinanceFeeds calculations derived from AMD’s reported revenue and year-over-year growth rates. Analyst targets as published by the named firms.
This article is for information purposes only and is not financial advice. FinanceFeeds does not recommend buying or selling any security. Technical levels describe market structure, not predictions, and semiconductor equities are highly volatile around earnings. Always do your own research and consider consulting a licensed financial adviser.