tapebrief
AMD · Q4 2025

AMD · Q1 2026 Earnings

Bullish

AMD

Reported May 5, 2026

30-second summary

30-second take: Revenue grew 38% YoY to $10.25B in Q1, beating the $10.1B high end of the prior guide by $153M and the $9.89B consensus by 3.7%, with Data Center +57% (a further 18-point acceleration from Q4's +39%). Q2 guidance of $11.2B at the midpoint implies +46% YoY growth against the $7.68B prior-year base — an 8-point YoY acceleration — with non-GAAP gross margin guided to ~56%, 100bps above Q1's 55%. The sequential -0.2% Q1 print versus the guided -5% means the China-MI308 step-down was more than absorbed by Data Center strength, and the Q2 setup is the cleanest "supply, not demand" framing AMD has issued in this cycle.

Headline numbers

EPS

Q1 FY2026

$1.37

+6.2% vs est.

Revenue

Q1 FY2026

$10.25B

+38.0% YoY

+3.7% vs est.

Gross margin

Q1 FY2026

53.0%

Free cash flow

Q1 FY2026

$2.57B

Operating margin

Q1 FY2026

14.0%

Key financials

Q1 FY2026
MetricQ1 FY2026YoYQ4 FY2025QoQ
Revenue$10.25B+38.0%$10.27B-0.2%
EPS$1.37—$1.53-10.5%
Gross margin53.0%—54.0%-100bps
Operating margin14.0%—17.0%-300bps
Free cash flow$2.57B—$2.08B+23.2%

Guidance

AMD raised expectations significantly: Q1 beat on revenue (+6pts YoY growth above guidance) and near-flat sequential trend (vs. prior -5% guide); Q2 FY2026 guidance of $11.2B implies 46% YoY growth with gross margin expanding 100bps to ~56%.

Guidance is issued one quarter forward. The Prior-guide column references the guide issued last quarter for the period just reported; the New-guide column is for next quarter.

Actuals vs prior guidance

MetricPeriodPrior guideActualΔResult
RevenueQ1 FY2026$9.5B - $10.1B (midpoint $9.8B)$10.253B+$0.45B above the high end of guideBeat
Non-GAAP Gross MarginQ1 FY2026approximately 55%55%in-lineMet
Revenue YoY GrowthQ1 FY2026approximately 32% YoY38% YoY+6pts above guideBeat
Revenue Sequential GrowthQ1 FY2026approximately -5% sequential-0.2% sequential+4.8pts above guide (near-flat vs. guided decline)Beat

New guidance

MetricPeriodGuideYoY
RevenueQ2 FY2026$10.9B - $11.5B (midpoint $11.2B)+46-50% YoY
Non-GAAP Gross MarginQ2 FY2026approximately 56%—
Revenue YoY GrowthQ2 FY2026approximately 46% YoY—
Revenue Sequential GrowthQ2 FY2026approximately 9% sequential—

Segment performance

Q1 FY2026
SegmentQ1 FY2026YoY
Data Center$5.775B+57.0%
Client$2.885B+26.0%
Gaming$0.72B+11.0%
Embedded$0.873B+6.0%
Data Center Operating Income$1.599B—
Data Center YoY Growth57%—
Client & Gaming YoY Growth23%—

Profitability

Q1 FY2026
SegmentQ1 FY2026
Non-GAAP Gross Margin55%
Non-GAAP Operating Margin25%
Free Cash Flow Margin25%
Adjusted EBITDA$2.746B
Operating Cash Flow$2.955B

Management tone

Q2 "Data Center stalls, China cliff" → Q3 "Record everything, Q4 guide sustains" → Q4 "Data Center re-accelerates, MI450 sized" → Q1 "Demand exceeds plan, supply is the constraint"

Three quarters ago Data Center was a $155M operating loss segment with a frozen China line; this quarter it printed $5.78B at +57% YoY and management's framing pivoted from "ramping to meet plan" to "scaling supply to meet demand." Su's prepared-remarks anchor — "Customer engagement around MI450 Series and Helios is strengthening, with leading customer forecasts exceeding our initial expectations" — is the most consequential disclosure shift in this cycle. It is the first quarter AMD has explicitly said customer pull is exceeding the company's own internal forecasts, which inverts the typical hyperscaler-AI dynamic where suppliers chase customer commitments.

The server CPU TAM disclosure in the Buckwalter exchange is a second-order tonal shift that compounds the first. Three quarters ago AMD anchored the server CPU TAM at ~$60B by 2030 growing at 18% CAGR; this quarter Su doubled it to ">$120B by 2030, growing >35% annually" with agentic AI as the driver. A TAM doubling between quarters with no acquisition or new product disclosure is unusual; either the prior framing was conservative beyond credibility or the agentic CPU thesis is genuinely re-rating the host-node ratio (1:4-1:8 → 1:1) faster than the supply chain assumed. The latter is what Su and Arya both anchored to.

The customer concentration narrative also continued to evolve. Last quarter Su moved from "OpenAI plus Oracle plus multiple unnamed" to "multiple customers working on MI450 ramp"; this quarter she added supply visibility "down to specific data center locations" in the O'Malley exchange. The progression from "we have customers" to "we have customers at named sites with power and capacity locked" is a meaningful credibility step — though specific anchor names beyond OpenAI and Oracle are still absent.

The one place tone got more cautious is memory. Su's Arya answer was the first time this cycle AMD has acknowledged shared cost inflation with customers ("sharing some cost increases while prioritizing unit volume") — a concession that semis pricing power has limits and that the memory cost cycle is real. Management framed the impact as concentrated in consumer PC and gaming in H2 2026, but the gross margin expansion to 56% in Q2 implies the data center side is absorbing it through mix.

Q&A highlights

Joshua Buckwalter · TD Cowen

How has the server CPU TAM inflected from $60B to $120B+ so quickly, particularly driven by agentic AI? Can AMD achieve its >50% share target given improving x86 competitor supply and ARM momentum?

Lisa explained the TAM expansion is driven by faster-than-expected AI adoption, inferencing, and agentic AI requiring significant CPU orchestration and data processing. AMD feels confident in >50% share target given broad CPU portfolio optimization across general purpose, head nodes, and agentic AI workloads, with Venice well-positioned and strong customer traction.

Server CPU TAM now >$120B by 2030, growing >35% annuallyPreviously guided ~$60B TAM at 18% CAGRVenice CPU family spans throughput, power, and cost-optimized variants plus Verano for AI infrastructureAMD targeting >50% server CPU share

Thomas O'Malley · Barclays

How has the cadence of data center GPU growth changed from prior back-half weighting? What are the supply constraints (power, data center build-outs) that could limit 2027 growth?

Lisa clarified Q1 data center AI was down modestly due to China revenue timing, not weakness. MI450/Helios ramp begins Q3 with initial volume, significant Q4 ramp, then Q1 2027 continuation. Management expressed confidence in supply visibility down to specific data center locations, with tight but manageable supply chain and increasing data center power capacity coming online.

Data center AI segment grew double-digit in Q2 guidance (sequential)Server CPU expected to grow >70% YoY in Q2, continuing strong through H2 2026 and 2027Helios ramp: initial volume Q3, significant ramp Q4, continues Q1 2027MI450 forecasts now exceeding initial plans; visibility extends to specific data center locations

Ross Seymour · Deutsche Bank

How does AMD differentiate in server CPUs given ARM competition and x86 competitor supply improvements? What is the gross margin trajectory as Helios ramps with below-corporate-average margins offset by strong server CPU and reduced gaming contribution?

Lisa emphasized broad CPU portfolio differentiation across general purpose, head nodes, and agentic AI workloads rather than single point product competition. Gene outlined gross margin tailwinds: server CPU >70% growth, gaming decline, client mix shift upward, and embedded accretion offsetting MI450 below-average margins in Q4, with confidence in 55-58% long-term range.

AMD pursuing broad portfolio optimization vs. ARM point productsServer CPU, client, and embedded are gross margin tailwinds in H2 2026MI450 ramps Q3 with significant Q4 volume at below-corporate-average marginsLong-term gross margin target: 55-58% range

Vivek Arya · Bank of America

Is agentic CPU growth incremental or cannibalistic to GPU TAM? Has AMD raised its AI TAM estimate with this CPU expansion? How is memory inflation being managed and what is AMD's memory supply security vs. competitors?

Lisa stated CPU growth is largely additive to AI TAM, with CPU-to-GPU ratios moving from 1:4-1:8 (host node only) toward 1:1 or higher as agentic workloads spawn more CPU tasks. On memory inflation, AMD has secured adequate supply via deep vendor partnerships and is sharing some cost increases while prioritizing unit volume and long-term relationships; larger capex impact falls on consumer markets.

Agentic AI CPU growth is largely incremental, not cannibalisticCPU-to-GPU ratio shifting toward 1:1 configuration from 1:4-1:8 previouslyAMD has secured sufficient memory supply to meet/exceed targetsMemory inflation mainly impacting consumer PC and gaming in H2 2026

Timothy Arcuri · UBS

How much of server CPU growth is unit-driven vs. ASP-driven in Q2 and H2? Are price increases being captured primarily in Q2 or distributed across the year?

Lisa and Gene confirmed growth is primarily unit-driven, with ASP increases mainly from product mix (higher core counts in newer generations) and modest cost passthrough due to supply-chain inflation, while prioritizing volume and long-term customer relationships. ASP increases are not major driver.

Q1 and forward server CPU growth predominantly unit-drivenQ1 shipped mix includes high-end Turin plus significant Genoa Zen 4 volumeASP increases primarily from mix (higher core counts), not pricing powerCost inflation shared with customers, but unit volume is priority

Answers to last quarter's watch list

Whether the Q1 ~$9.8B guide is another conservative setup — Q1 printed $10.25B, $153M above the high end and $453M above the midpoint (+4.6%). The Q2 $11.2B midpoint should be read as a more honest centroid than prior guides given the explicit supply-constrained framing. Status: Resolved positively
Q1 Data Center sequential growth disclosure — Data Center printed $5.78B in Q1 vs. $5.38B in Q4, +$395M sequential (+7.3% QoQ, confirmed by Jean Hu as "up 7% sequentially"). Su's commitment in the Q4 call that Data Center would be up sequentially was met cleanly, and the magnitude was stronger than the segment-level guide implied. Status: Resolved positively
Q1 gross margin actual vs. ~55% guided — Non-GAAP gross margin landed exactly at 55%, in-line with guide. The China-MI308 mechanical step-down framing held; underlying product mix did not deteriorate. The Q2 guide to ~56% confirms the 55% Q1 was a trough, not a new ceiling. Status: Resolved positively
MI325 China license decision — Management did not disclose any update on MI325 licensing, and Q2 guidance does not include incremental China revenue beyond the MI-308 baseline. The China optionality remains live but unresolved. Status: Continue monitoring
Named MI450 anchor customers beyond OpenAI and Oracle — Su referenced "leading customer forecasts exceeding our initial expectations" and added supply visibility "down to specific data center locations," but no new anchor customers were named on the call. The concentration narrative continues to soften through framing rather than disclosure. Status: Continue monitoring
Embedded sequential durability — Embedded printed +6% YoY in Q1 vs. +3% YoY in Q4 — second consecutive positive print and an acceleration. The Q4 inflection was not a low-base artifact. Status: Resolved positively

What to watch into next quarter

Whether Q2 prints at or above the $11.5B high end — would confirm the supply-constrained framing Su anchored to and reset 2H 2026 expectations upward. Anything below $11.2B midpoint with no acquisition or one-off undercuts the "demand exceeding plan" narrative.

Q2 Data Center segment operating margin — Q1 compressed 490bps QoQ to ~27.7% despite 7% QoQ revenue growth. Watch whether Q2's revenue acceleration restores margin toward Q4's 32.6% or whether the compression continues ahead of MI450's H2 dilution. A Q2 Data Center op margin below 28% with revenue accelerating signals structural mix erosion, not just MI-308 step-down.

Server CPU YoY growth disclosure — Su committed to >70% server CPU YoY growth in Q2. This is the most specific segment-within-segment guide AMD has issued and is directly checkable. Anything below +60% YoY puts the broader Data Center re-acceleration story under pressure.

MI450 initial revenue contribution and rack-scale Helios ASP commentary — Su confirmed initial MI450 volume in Q3 2026 with rack-scale Helios as the dominant form factor. Watch the Q2 call for any pre-shipment revenue recognition framework, Helios ASP anchors, or first-customer naming. The H2 ramp magnitude depends entirely on these data points.

Memory cost passthrough impact on Client and Gaming gross margin — Su flagged memory inflation as concentrated in consumer PC and gaming in H2 2026. Watch whether Q2 Client gross margin compresses sequentially as inflation flows through ASP — would validate Su's framing as conservative or aggressive.

MI450 customer names beyond OpenAI and Oracle — H2 2026 production starts in Q3. By Q2 results, AMD will be 90 days from first volume shipment. A Q2 call that closes with only OpenAI and Oracle as named anchors keeps concentration risk live going into the ramp quarter.

Sources

  1. AMD Q1 FY2026 press release (SEC filing): https://www.sec.gov/Archives/edgar/data/2488/000000248826000072/q12026991.htm
  2. AMD Q1 FY2026 earnings call Q&A
  3. Tapebrief AMD Q4 FY2025 brief
  4. Tapebrief AMD Q3 FY2025 brief
  5. Tapebrief AMD Q2 FY2025 brief

Get the next brief, free.

We publish analyst-grade earnings briefs the same day or morning after every call — headline numbers, segment KPIs, Q&A highlights, and tone analysis. Free during beta.

This is not investment advice.