Advanced Micro Devices (AMD)

Investment Review After Q1 FY2026 Earnings (2026.05.05)

TL;DR

AMD is the credible #2 in AI accelerators (Instinct MI series) and the #1 server CPU vendor (EPYC), assembling full-stack AI rack systems via Helios + Venice + Pensando. Just reported: Q1 FY2026 revenue $10.3B (+38% YoY), GM 55%, Data Center +57% YoY, FCF tripled to a record $2.6B — strong, scorecard 6/6. Q2 FY2026 guide $11.2B (+46% YoY at midpoint), GM ~56% — accelerating, very strong. Stock at $355 with 95x P/E(ttm), up from 66x at quarter close — extreme valuation that prices in flawless MI450 ramp and TAM doubling.

Summary

Finding	Assessment
5-Year Stock Return	Stock currently $355, up >350% from ~$78 (5 years ago). TTM OP grew from $1.85B to $5.17B (+179%). P/E(ttm) range 32x–95x. FCF/OP consistently above 100% shows strong cash conversion despite GAAP earnings depressed by Xilinx amortization.
Business Model	Full-stack compute platform: CPUs (EPYC, Ryzen), GPUs (Instinct, Radeon), FPGAs (Versal), adaptive SoCs. Fabless model with TSMC foundry leverage. $35B+ revenue across Data Center (~50%), Client+Gaming, Embedded. ZT Systems acquisition adds rack-scale integration.
Competitive Moat	NARROW. x86 duopoly in CPUs, chiplet architecture lead, growing ROCm AI ecosystem. NVIDIA CUDA still dominates AI training; Arm expanding; hyperscaler custom silicon threatens GPU addressable market.
Past 5-YR Drivers	AI GPU revenue from zero to $5B+ run rate, EPYC server share gains (single digits to ~30%+), margin mix-up from data center growth. Xilinx amortization declining as share of revenue.
Recent Stock Move	Stock currently $355, up >190% from $121 (Dec 2024); +75% post Q1 print vs $203 quarter close. TTM OP $558M→$5.17B in 9 quarters. P/E(ttm) re-expanded 66x→95x at $355. R40 above 40% for 8 consecutive quarters.
Revenue Trend	ACCELERATING Rev YoY: 38% (Q1 2026) → 46% (Q2 2026E).
Margin Trend	EXPANDING OM%: 14% (Q1 2026) → 17% (Q2 2026E).
TTM OP Trajectory	ACCELERATING TTM OP QoQ: 15% (Q1 2026) → 25% (Q2 2026E). TTM OP $5,165M.
Key Opportunities	Agentic AI is "largely additive" to TAM — server CPU TAM revised $60B→$120B by 2030. Meta + OpenAI 6 GW each; MI450 sampling, Helios production H2 2026; "tens of billions" DC AI revenue 2027; LT EPS >$20.
Key Risks	P/E 95x prices in flawless MI450 ramp + TAM doubling on schedule. Helios slip / yield issue / Meta-OpenAI deployment delay would compress multiple sharply. NVIDIA CUDA still dominant; MI450 margins below corp avg pressures H2 GM; H2 PC+gaming weak on memory inflation; OpEx +42% YoY.
Catalysts	Q2 FY2026 print (~Aug 2026, guided $11.2B ±$300M, +46% YoY); Advancing AI event July 2026; Helios initial volume Q3, ramp Q4; Venice (Zen 6, 2nm) launch later 2026; MI500 (CDNA6, 2nm, HBM4e) 2027.
P/E(ttm) Valuation	ELEVATED at 95x (range: 32x–95x over last 20 quarters). P/E(ttm) distorted by ~$4B/year Xilinx intangible amortization. On non-GAAP P/E (~50-55x at $355), AMD trades at a steep premium. Sustained >35% revenue + clean MI450 ramp could justify; execution miss compresses sharply.

Q1 FY2026 Key Data

Revenue $10.3B (+38% YoY), beat high end of guide. Data Center $5.8B (+57%), Client+Gaming $3.6B (+23%), Embedded $873M (+6%). GM 55% (+170bps). EPS $1.37 (+43%).
Record FCF $2.6B (25% of revenue, more than tripled YoY). OI $2.5B (25% OM); cash $12.3B; OpEx $3.1B (+42% YoY). Buyback: 1.1M shares / $221M; $9.2B authorization remaining.
Q2 guide: revenue $11.2B ±$300M (+46% YoY at midpoint, +9% QoQ). GM ~56%; OpEx ~$3.3B; tax 13%. Server CPU expected +70% YoY; DC AI returns to growth post China step-down.

Key Takeaways from Earnings Call

Server CPU TAM revised UP $60B → $120B by 2030 (>35% CAGR vs prior 18%). Agentic AI "largely additive"; CPU/GPU ratio shifting 1:8 → 1:1. Server +50% YoY Q1, +70% YoY Q2 guide. Verano (AI-optimized EPYC) ships in Venice 2026.
Meta 6 GW multi-gen Instinct partnership joins OpenAI as second mega-anchor. Includes custom MI450. Sampling now; Helios initial volume Q3, ramp Q4. "Tens of billions" DC AI revenue targeted 2027 — visibility "down to which data center."
H2 consumer demand softens on memory inflation; LT GM 55-58% reaffirmed. Gaming -20%+ H1→H2. MI450/Helios margins below corp avg pressure H2 GM but offset by server/embedded mix-up. R&D growth to outpace SG&A.

Key Segments and Revenue Mix (FY2025)

Segment	Revenue	% of Total	Key Details
Data Center	$16,635M	48%	EPYC server CPUs + Instinct AI GPUs (MI350 series). +32% YoY. Record share gains in server CPUs; 8 of 10 top AI companies using Instinct.
Client	$10,640M	31%	Ryzen desktop/mobile CPUs. +51% YoY. 31% unit growth + 15% ASP increase. Record desktop share for 4 consecutive quarters.
Gaming	$3,910M	11%	Semi-custom console SoCs (Sony/Microsoft) + Radeon GPUs. +51% YoY. RDNA 4 launched. Console cycle maturing (year 7).
Embedded	$3,454M	10%	FPGAs, adaptive SoCs (Xilinx). -3% YoY. Recovery underway from post-COVID inventory normalization. $17B design wins (record).

Business Model

SUPPLIERS

TSMC (primary foundry, advanced nodes)
Samsung Foundry (select products)
Substrate & packaging suppliers
EDA tools (Synopsys, Cadence)

Wafers / Packaging →

AMD

Chip design (CPUs, GPUs, FPGAs)
Chiplet architecture (Infinity Fabric)
ROCm software ecosystem
ZT Systems rack-scale integration
Fabless: ~$35B revenue, ~28K employees

Chips / Systems → $ Revenue

CUSTOMERS

Hyperscalers (OpenAI, Microsoft, Meta)
Cloud providers (AWS, Azure, GCP)
PC OEMs (Dell, HP, Lenovo)
Console makers (Sony, Microsoft)
Enterprise & sovereign AI

Moat Assessment: NARROW

x86 CPU duopoly with Intel is a structural advantage: Only AMD + Intel hold x86 licenses — a de facto PC/server duopoly. Chiplet architecture + TSMC partnership give a sustained perf lead. Arm (Qualcomm, Graviton, Apple) expanding.
AI GPU ecosystem is a growing but unproven moat: ROCm gaining traction (vLLM, millions of models); NVIDIA CUDA still dominates with 10+ yrs developer lock-in. Moat depends on execution — OpenAI + Meta + Helios build switching costs.
Embedded/FPGA franchise has deep design-in stickiness: 5-10 yr cycles in aerospace, automotive, industrial. $50B+ cumulative design wins since Xilinx = committed future revenue. Faces Altera, Lattice, ASIC competition as volumes scale.

Stock vs. Earnings — 5-Year Performance

Stock currently $355, up >350% from ~$78 (5 yrs ago); TTM OP $1.85B→$5.17B (+179%). AMD’s shift from CPU underdog to full-stack AI compute platform drove earnings and stock gains. Each AI inflection (MI300/350/450) re-rated the multiple.
Path not smooth: $164 (Jun 2022) → $63 (Sep 2022) on PC/gaming downturn, then $183 (Q1 2024) on AI hype → $103 (Q1 2025) on export controls. Each downcycle was segment-specific weakness, not fundamental competitive deterioration.
P/E(ttm) ranged wildly 32x → 340x+ over the period due to ~$4B/year Xilinx intangible amortization suppressing GAAP earnings. Current 95x P/E reflects re-expansion post Q1 2026 print + premium pricing of sustained AI GPU growth.

Past 5-Year Key Drivers

AI GPU emergence as a major revenue driver: Instinct MI300 launched late 2023; AI DC GPU revenue grew from near-zero to $5B+ run rate by Q4 2025. MI350/450 roadmap positions AMD as the only credible NVIDIA alt; OpenAI + Meta validate it.
EPYC server CPU share gains drive margin mix-up: EPYC went single-digit → ~30%+ server share over 5 yrs on TSMC perf-per-watt edge. Server CPUs carry higher margins; each share point lifts mix. Agentic AI accelerates CPU demand alongside GPU.
Xilinx ($49B, 2022) added embedded/FPGA franchise: Brought $3.5B embedded revenue + $17B/yr design wins, but ~$4B/yr amortization depresses GAAP OM. As amortization declines and embedded recovers, contribution lifts profitability.

Quarterly Key Metrics

	Q1 2024	Q2 2024	Q3 2024	Q4 2024	Q1 2025	Q2 2025	Q3 2025	Q4 2025	Q1 2026	Q2 2026E
Quarter End	Mar 2024	Jun 2024	Sep 2024	Dec 2024	Mar 2025	Jun 2025	Sep 2025	Dec 2025	Mar 2026	Jun 2026
Stock Price	$183	$158	$164	$121	$103	$142	$162	$214	$203	$355
Stock QoQ	+32%	-14%	+4%	-26%	-15%	+38%	+14%	+32%	-5%	+75%
Rev YoY	2%	9%	18%	24%	36%	32%	36%	34%	38%	46%
OM%	1%	5%	11%	16%	11%	9%	14%	17%	14%	17%
R40 (Rev YoY + OM%)	3%	13%	28%	40%	47%	40%	49%	51%	52%	63%
TTM OP ($M)	$558	$855	$1,365	$2,272	$3,042	$3,440	$3,986	$4,495	$5,165	$6,448
TTM OP QoQ	+52%	+53%	+60%	+66%	+34%	+13%	+16%	+13%	+15%	+25%
FCF/OP ttm	210%	159%	114%	106%	90%	118%	137%	150%	166%	157%
Debt/EBITDA	0.5x	0.4x	0.3x	0.4x	0.8x	0.6x	0.6x	0.4x	0.4x	0.4x
P/E ttm	275x	193x	147x	108x	69x	77x	76x	79x	66x	95x

Recent Stock Performance & Drivers

Stock currently $355, vs. $121 (Dec 2024) and $203 (Q1 2026 close). The +75% post last quarter end rip reflects a structural shift: TAM raise, Meta anchor, FCF tripling, 6/6 scorecard — all in one print. Pre-print, range-bound $200–$240 since late 2025.
TTM OP grew $558M → $5.17B over 9 quarters as all four segments improved: DC +57%, client +26%, gaming +11%, embedded back to growth. GAAP OM expanded 1% → 14% as Xilinx amortization shrank as share of revenue.
P/E(ttm) re-expanded 66x → 95x post last quarter end. Stock prices in clean MI450 ramp, TAM doubling, AI GPU returning to sequential growth Q2. R40 above 40% for 8 consecutive qtrs — longest streak since 2021 boom.

Opportunities

Agentic AI is "largely additive" to AMD's TAM, not zero-sum vs NVIDIA. CPU/GPU ratios shifting 1:8 → 1:1 as agents spawn CPU tasks. Server CPU TAM $60B→$120B by 2030 (>35% CAGR); Verano AI-optimized EPYC inside Venice.
MI450/Helios H2 2026 ramp with TWO mega-anchors: OpenAI + Meta 6 GW. Custom MI450 co-designed for Meta. Sampling now; Helios initial volume Q3, ramp Q4. "Tens of billions" DC AI revenue targeted 2027; LT >80% growth path.
Server CPU +70% YoY guided Q2; Turin crossed 50% of server mix. Venice (Zen 6, 2nm) launches later 2026 with "strongest pull" of any prior gen. Commercial Ryzen PRO sell-through +50% YoY. Embedded record design wins 2025.

Challenges

Valuation risk: P/E 95x prices in flawless MI450 ramp + TAM doubling on schedule. A Helios slip, yield issue, or anchor deployment delay would compress the multiple sharply (pre-print P/E 66x). MI450/Helios margins below corp avg pressure H2 GM.
NVIDIA CUDA still dominant; OpEx +42% YoY is an aggressive bet that has to convert. ROCm closing inference gap (MLPerf strong) but training defaults to CUDA. AMD wins via hyperscaler TCO, not broad enterprise. Multi-year AI payback.
Memory inflation hits consumer H2 — gaming -20%+ H1→H2; PC sub-seasonal. DC compute supply secured via memory partnerships, but consumer softens as pricing flows through. Console year 7; semi-custom declines until next-gen (2027).

Next 12 Months

Q2 FY2026 guide $11.2B ±$300M (+46% YoY, +9% QoQ); GM ~56%. Server CPU +70% YoY, DC AI returns to sequential growth, embedded double-digit, client modest. A beat >$11.5B with GM ≥56% validates the inflection.
Helios ramp: initial volume Q3, significant ramp Q4, continuing into Q1 2027. MI450 sampling at lead customers now. "Visibility down to which data center" for 2027. Qualification or yield slip is the single highest-impact negative catalyst.
Advancing AI event in July; Venice (Zen 6, 2nm) launches later 2026. Watch Verano AI-optimized SKU details, MI500 (CDNA6, 2nm, HBM4e) 2027 cadence, ROCm progress. Server CPU continues >70% YoY pace into H2 2026.

Next 12-24 Months

Execution test: can AMD reach "tens of billions" DC AI revenue 2027 + EPS >$20? 2027 visibility "down to which data center." MI500 (CDNA6, 2nm) maintains annual cadence vs NVIDIA. Failure to broaden beyond OpenAI/Meta compresses the multiple.
Hyperscaler custom silicon (TPU, Trainium, Maia, OpenAI in-house) is the structural threat. Custom inference could shrink AMD’s addressable share. Counter: open ecosystem + TCO + rack-scale flex. CPU TAM doubling helps offset.
LT GM target 55-58% reaffirmed; OpEx leverage hinges on revenue scaling. Helios margins below corp avg in early ramp; server/embedded mix-up offsets. As Instinct scales, ASP + cost leverage drive GM toward target. R&D outpaces SG&A.

Core Metrics Scorecard

	Q3'21	Q4'21	Q1'22	Q2'22	Q3'22	Q4'22	Q1'23	Q2'23	Q3'23	Q4'23	Q1'24	Q2'24	Q3'24	Q4'24	Q1'25	Q2'25	Q3'25	Q4'25	Q1'26	Q2'26E
Quarter End	Sep21	Dec21	Mar22	Jun22	Sep22	Dec22	Mar23	Jun23	Sep23	Dec23	Mar24	Jun24	Sep24	Dec24	Mar25	Jun25	Sep25	Dec25	Mar26	Jun26
R40 > 40%	✅	✅	✅	✅										✅	✅	✅	✅	✅	✅	✅
R40 accelerate			✅						✅	✅		✅	✅	✅	✅		✅	✅	✅	✅
TTM OP QoQ > 10%	✅	✅									✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
TTM OP QoQ accelerate									✅		✅	✅	✅	✅			✅		✅	✅
TTM FCF/OP > 60%	✅	✅	✅	✅	✅	✅	✅	❌	❌	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
Debt/EBITDA < 3.5	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
Total passes	4	4	4	3	2	2	2	1	3	3	4	5	5	6	5	4	6	5	6	6
Stock QoQ	+10%	+40%	-24%	-30%	-17%	+1%	+53%	+16%	-9%	+34%	+32%	-14%	+4%	-26%	-15%	+38%	+14%	+32%	-5%	+75%
P/E ttm	32x	56x	47x	41x	47x	85x	431x	-3309x	961x	275x	275x	193x	147x	108x	69x	77x	76x	79x	66x	95x

Q1 2026 scored 6/6 passes — first perfect score since Q4’24; Q2 2026E projects 6/6 again. All four growth metrics plus FCF/OP>60% and low leverage are firing. The +75% post last quarter end move is the market reading the perfect scorecard signal.
FCF/OP and Debt/EBITDA almost never fail; R40 and TTM OP growth drive scorecard swings. FCF/OP failed only twice (Q2–Q3’23 trough); Debt/EBITDA never failed in 20 quarters. Swings between 1 and 6 passes are R40 + TTM OP QoQ driven.
Pattern: 6-pass quarters can precede returns OR multiple compression. Q4’24 hit 6 passes but stock fell -26% on P/E compression 165x→120x. Q1’23 scored 2 passes but stock rallied +53% on anticipation. P/E now 95x — future returns hinge on multiple stability.

Chart: R40 vs Stock Price (Since 2020)

Chart: Scorecard Total Passes vs Stock Price (Since 2020)

Source: SEC EDGAR annual and quarterly filings, company earnings conference calls and presentations.

Disclaimer: This report is for educational purposes only. NO investment advice.