AMD Q2 2026 Earnings — How Much Is Left in a Guide That Was Already Aggressive
The Call
This quarter is not a demand question, it is a sandbag question. AMD already guided Q2 revenue to $11.2B, which was 6.3% above consensus when issued, and separately guided gross margin and opex. Almost every input is management-supplied, so the only thing genuinely being forecast is how far above its own midpoint AMD lands, and whether an unusually aggressive guide left any room to beat.
| Metric | Our Estimate | Company Guidance | Consensus | Actual | Error | vs Cons. |
|---|---|---|---|---|---|---|
| RevenueGAAP Guide is the ±$300M midpoint. Consensus quoted. | 11,540 USD M | 11,200 USD M | 11,300 USD M | 11,536 USD M | +0.03% | 59.0× |
| Gross profitnon-GAAP Both benchmarks derived: revenue × gross margin on the same line. | 6,485 USD M | 6,272 USD M | 6,328 USD M | 6,488 USD M | -0.05% | — |
| Gross marginnon-GAAP Guide quoted as ~56%. Consensus sits on the guide. | 56.2% | 56.0% | 56.0% | 56.24% | -0.07% | 5.8× |
| Operating profitnon-GAAP Derived: gross profit less opex ($3,300M guided, ~$3,250M implied by consensus). | 3,195 USD M | 2,972 USD M | 3,078 USD M | 3,094 USD M | +3.26% | — |
| Operating marginnon-GAAP Derived from the operating-profit line. | 27.7% | 26.5% | 27.2% | 26.82% | +3.28% | — |
| Net incomenon-GAAP Derived: operating profit plus $30M net interest, taxed at 15%. | 2,741 USD M | 2,552 USD M | 2,642 USD M | 2,760 USD M | -0.69% | — |
| Diluted EPSnon-GAAP Consensus quoted. Guide-implied EPS derived on 1,645M shares; the street already sits 3.9% above it. | 1.67 USD | 1.55 USD | 1.61 USD | 1.66 USD | +0.60% | 5.0× |
- → AMD merely meets its guide midpoint. Revenue $11,200M, EPS $1.57, which lands BELOW the $1.61 consensus and inverts the call.
- → The Q1 beat rate repeats in full (+4.6%). Revenue $11,715M, EPS $1.72.
- → Gross margin comes in at the Q1 level of 55% rather than 56.2%, on HBM cost pass-through from the memory upcycle. EPS $1.60, essentially in line with consensus.
- → Centre risk: the whole call rests on assuming a sandbag inside a guide that was already 6.3% above consensus when issued. If management front-loaded its optimism in May, the historical beat rate does not apply and the honest centre is closer to the guide.
Behaviour model, not arithmetic, and deliberately low confidence. Built from the recent sequential guide step (Q2 guided +9% off the Q1 print) applied to our Q2 estimate, cross-checked against TSMC guiding Q3 revenue to $44.6-45.8B (+11% QoQ). This method predicts what management chooses to say rather than what the business does, and a seasonal or historical base rate on management behaviour breaks the moment the narrative changes. Score it separately from the ladder above.
Evidence (5)
Positioning Suggestions
→ The call is 2.1% above consensus on revenue and 3.7% on EPS, which is a modest edge. Size accordingly: this is not a high-conviction dislocation, it is a small directional lean on operating leverage.
→ The asymmetry sits in the guide, not the quarter. A Q3 guide materially above $12.6B matters more to the multiple than any plausible Q2 beat.
→ TSM already reported and already re-rated on the same AI demand. Consider whether AMD exposure here duplicates TSM exposure already held.
Predictions
8Q2 2026 revenue exceeds the $11.30B consensus
If revenue beats but the Q3 guide is at or below $12.3B, treat the print as spent and do not add
Q2 2026 non-GAAP diluted EPS exceeds the $1.61 consensus
Non-GAAP gross margin is at or above the 56% guide
Data Center segment revenue exceeds $6.5B
If Data Center misses $6.5B while TSMC HPC grew 20% QoQ, the share-loss-to-NVDA thesis needs re-opening
Q3 2026 revenue guidance midpoint is above $12.0B
Revenue exceeds the top of AMD's own guide range ($11.5B)
AMD discloses a standalone MI450 revenue figure
Management explicitly names HBM or memory cost as a gross-margin headwind on the call
Key Questions
- Did AMD beat its own guide midpoint, and by how much relative to the +4.6% it managed in Q1?
- Is non-GAAP gross margin holding above 56% while the MI-series ramps, or is HBM cost inflation from the memory upcycle already visible in the bridge?
- Does Data Center growth track the +20% QoQ that TSMC's HPC line implies, or does AMD lag its own foundry's read-through?
- Does management quantify MI450 revenue or keep it inside the Data Center aggregate?
- How large is the Q3 guide step, and does it rest on committed customer volume or on pipeline language?
Monitoring Checklist
Post-Event Results
Actual Outcomes
- • Revenue $11,536M against our $11,540M and a $11,300M consensus. Non-GAAP gross margin 56.24% against our 56.2%. Non-GAAP EPS $1.66 against our $1.67 and a $1.61 consensus.
- • Data Center $6,718M against our $6,700M; Client $3,062M, Gaming $779M, Embedded $977M, together $4,818M against our $4,840M. The volume-versus-mix double edge did not fire: the beat came from Data Center, which is where the main line said it had to come from.
- • The single free variable was exact. The beat over the guide midpoint was +3.00% against our +3.00%.
- • Three inputs tagged `anchored` were each wrong. Opex $3,394M against our $3,290M, so AMD landed ABOVE its own opex guide, which the anchoring rationale had explicitly ruled out. Tax rate 11.65% against our 15.0%. Diluted shares 1,659M against our 1,645M. Correcting any single one moves EPS further from the print; the +0.60% came from the three netting out.
- • Q3 2026 guidance $13.0B +/- $0.3B against our $12.6B call, -3.08%. Second consecutive guidance call low in an up-cycle after STX FQ1 FY2027 at -9.3%.
- • MI450 was not given a standalone revenue figure, so the accelerator-versus-EPYC split inside Data Center stays inferred rather than observed, exactly as the limits section stated.
- • First-session price reaction not recorded here: press coverage reported the shares lower after hours, but no verified close-to-close figure was captured, so marketReaction is left empty rather than filled with an invented number.
Trade Recommendations
→ No action from this print alone. The call was 2.1% above consensus on revenue and it landed, but the edge was small by design and is now spent.
→ The Q3 guide at $13.0B is the part that matters and it came in above our $12.6B. Any position sizing should key off that, not off the Q2 beat.