NVIDIA (NVDA)

Investment Review After Q1 FY2027 Earnings (2026.05.20)

TL;DR

NVIDIA is the world’s leading AI accelerated-computing platform — Blackwell + Rubin GPUs, Spectrum-X networking, and the CUDA software moat that powers frontier AI training and inference. Just reported: Q1 FY2027 revenue $82B (+85% YoY, +20% QoQ) record; Non-GAAP GM 75%; FCF $49B record — strong beat with 14th straight quarter of sequential growth. Q2 FY27 guide $91B ±2% (+12% QoQ), GM 75% — accelerating, with Data Center driving sequential growth and Vera CPU opening a new $200B TAM. Stock at $223 with 34x P/E(ttm) — re-rated from 34x at Q1 close ($174); $80B new buyback + dividend +25x to $0.25, $1T Blackwell+Rubin visibility through 2027.

Summary

Finding	Assessment
5-Year Stock Return	Stock currently $223, up over 1,500% from $14 (5 years ago, split-adjusted). TTM OP ~33x increase ($5B→$162B). P/E(ttm) compressed from 80x peak to 34x as earnings dramatically outpaced stock.
Business Model	DOMINANT. $268B FY26 revenue (+73% YoY); Q1 FY27 $82B (+85% YoY). Data Center $75B Q1. FCF $49B record. CUDA ecosystem (1.5M models) is the de facto AI platform; Vera CPU now adds $200B TAM.
Competitive Moat	WIDE CUDA ecosystem lock-in (25+ yrs, 1.5M models). Full-stack GPU+CPU+networking+systems+software co-design. Annual cadence at $20B+ R&D. Blackwell Ultra swept MLPerf; Vera Rubin promises 35x inference throughput.
Past 5-YR Drivers	AI training/inference explosion (13x since ChatGPT). Platform co-design generational leaps. TAM expansion into networking ($31B), sovereign AI (~40 countries / $50T GDP), physical AI (>$9B TTM), and now Vera CPU ($200B TAM).
Recent Stock Move	Stock at $223 vs $174.40 Q1 FY27 close = +28% post last quarter end. TTM OP +20% in 2 quarters ($135B→$162B). P/E(ttm) at 34x. R40 at +151%.
Revenue Trend	ACCELERATING Rev YoY: +85% (Q1’27) → +95% (Q2’27E).
Margin Trend	FLAT OM%: 66% (Q1’27) → 66% (Q2’27E). GM 75% Non-GAAP, mid-70s full-year FY27.
TTM OP Trajectory	DECELERATING TTM OP QoQ: +20% (Q1’27) → +19% (Q2’27E). TTM OP $162,285M.
Key Opportunities	$3–4T AI infra annual spend by end of decade. Hyperscaler CapEx >$1T 2027. Vera CPU $200B TAM (~$20B revenue this year). ACIE +31% QoQ, AI cloud >3x YoY, sovereign +80%. $1T Blackwell+Rubin visibility 2025-2027.
Key Risks	AI CapEx cycle deceleration risk. Hyperscale ~50% of DC revenue. China: H200 licenses approved but zero revenue; NO China DC compute in outlook. 34x P/E priced for sustained AI capex. Supply commitments $145B; manufacturing complexity at scale.
Catalysts	Q2 FY27 guide $91B ±2% / GM 75% / OpEx upper-40s growth. Vera Rubin production H2 FY27 (35x inference vs Blackwell). $80B + $39B buyback authorization; dividend +25x to $0.25; ~50% of FCF returned. Anthropic + Microsoft Fairwater + AWS Rubin ramps.
P/E(ttm) Valuation	NEUTRAL at 34x (5-year range: 29x–80x). Near the low end of range — market pricing in growth deceleration from extraordinary base. With +85% Rev YoY and Q2 guide accelerating, significant upside if growth sustains; 34x is fair if growth normalizes to 20–30%.

Q1 FY2027 Key Data (Feb-Apr 2026), reported 2026.05.20

Revenue $82B (+85% YoY, +20% QoQ) record; GAAP GM 74.9%, Non-GAAP 75% flat; FCF $49B record (vs $35B Q4); 14th straight Q of sequential growth.
Segments (new segmentation): Data Center $75B (+92% YoY, +21% QoQ) — Hyperscale $38B (~50% DC, +12% QoQ), ACIE $37B (+31% QoQ, AI cloud >3x YoY). Edge Computing $6.4B (+29% YoY).
Q2 FY27 guide $91B ±2% (+12% QoQ at mid); GM 75% ±50bps. $80B new buyback + dividend +25x to $0.25/share.

Key Takeaways from Earnings Call

Record quarter with all-time-high revenue and FCF; 14th straight Q of sequential growth despite manufacturing complexity at scale. CEO frames AI as “now a necessity” — transitioning from one-shot inference → reasoning → agentic. Forecasts $3-4T AI infrastructure spend annually by end of decade; hyperscaler CapEx forecast >$1T in 2027. H100 cloud rental price +20% YTD; A100 +15% — pricing power intact.
NEW SEGMENTATION rolled out: Data Center split into Hyperscale (5-6 companies, public cloud + largest consumer internet) and ACIE (AI Clouds, Industrial, Enterprise, Sovereign — 250K+ companies). ACIE +31% QoQ; sovereign +80% YoY; AI cloud revenue more than tripled YoY. Edge Computing surfaced as a separate platform: $6.4B (+29% YoY), with Physical AI revenue >$9B trailing 12 months across PCs, gaming, robotics, automotive.
Massive capital return acceleration: $20B returned in Q1 (record); dividend +25x to $0.25/share; $80B new buyback authorization + $39B remaining = $119B total. Plan to return ~50% of FCF to shareholders this year. Anthropic strategic partnership deepened (AWS, Azure, CoreWeave, SpaceXAI capacity expansion); Microsoft Fairwater data center live with 100,000+ Blackwell GPUs.
Forward platform: Vera CPU opens $200B NEW TAM — NVDA’s first CPU effort, custom Arm cores co-designed with Rubin GPUs/NVLink (+1.5x perf/core, +2x perf/watt, +4x density per rack vs x86). ~$20B CPU revenue visibility this year. Vera Rubin production starts H2 FY27 — 7 chips across 5 racks delivering up to 35x inference throughput and 10x AI-factory revenue vs Blackwell. $1T Blackwell+Rubin revenue visibility 2025-2027.

Key Segments and Revenue Mix (FY2026)

Segment	FY26 Rev	Mix	Products	End Markets	Key Competitors
Data Center	$194B	72%	Blackwell/Hopper GPUs, NVLink, Spectrum-X, DGX, Vera CPU (FY27)	AI training/inference, hyperscalers, sovereign AI, enterprise	AMD (MI300/MI400), Intel Gaudi, custom ASICs (Google TPU, AWS Trainium)
Gaming	$13B	5%	GeForce RTX GPUs, DLSS, G-SYNC	PC gaming, creators, AI PC	AMD Radeon, Intel Arc
Professional Visualization	$3B	1%	RTX PRO workstations, Omniverse	Design, engineering, digital twins, AI development	AMD Pro, Intel
Automotive & Robotics	$2B	1%	DRIVE, Isaac, Cosmos, GR00T	Self-driving, robotics, industrial AI	Qualcomm, Mobileye, Tesla FSD
Networking	$31B	12%	NVLink, Spectrum-X, InfiniBand, ConnectX, BlueField	AI factory interconnect, data center networking	Broadcom, Arista
Other/OEM	$25B	9%	Various	Various	Various
Total FY2026	$268B	100%	+73% YoY. Data Center +68%. Networking +3.5x since Mellanox. FCF $97B.

Business Model

SUPPLIERS

TSMC (leading-edge fabrication)
SK Hynix / Samsung / Micron (HBM)
CoWoS advanced packaging
PCB & substrate suppliers

Silicon / Memory →

NVIDIA

GPU + CPU design (Blackwell → Vera Rubin)
CUDA software ecosystem (1.5M+ AI models)
NVLink / Spectrum-X networking
DGX / HGX systems
$20B+ R&D budget

GPUs / Systems → $ Revenue

CUSTOMERS

Hyperscalers (Microsoft, Google, Meta, Amazon, xAI)
AI clouds + model makers (OpenAI, Anthropic)
Sovereign nations (~40 countries)
Enterprises
Edge / Gaming / Robotics

Moat Assessment: WIDE

CUDA ecosystem lock-in: 1.5M+ AI models on Hugging Face, all running on CUDA. 25+ years of software ecosystem investment. Developers, frameworks, and toolchains all optimized for NVIDIA. Switching to AMD/Intel/custom ASICs requires rewriting entire software stacks — a multi-year, multi-billion dollar effort no customer is willing to undertake at scale.
Full-stack co-design advantage: Only company designing GPUs + CPUs (Vera) + networking (NVLink, Spectrum-X) + systems (DGX) + software (CUDA, Omniverse) as integrated platform. Data centers running NVIDIA generate the highest revenue per watt. MLPerf: Blackwell Ultra swept every benchmark. Vera Rubin promises 35x inference throughput vs Blackwell.
Annual innovation cadence at unprecedented scale: $20B+ R&D budget. Hopper → Blackwell → Vera Rubin in 3 years. Rubin trains MoE models with 1/4 the GPUs and reduces inference cost dramatically vs Blackwell. Competitors are 1–2 generations behind on both silicon and software. Customers are generating profitable revenue beyond the depreciable life of their GPUs.

Stock vs. Earnings — 5-Year Performance

Stock currently $223, up over 1,500% from $14 (5 years ago, split-adjusted), driven by the AI revolution making NVIDIA the world’s most important semiconductor company.
TTM OP exploded from ~$5B to $162,285M — a ~33x increase driven by data center GPU demand for AI training and inference; full-year FY26 OP base now $135B.
P/E(ttm) ranged 29x–80x over 5 years. Currently 34x — well below the 5-year average as earnings growth has caught up with stock appreciation. R40 at +151% is extraordinary.

Past 5-Year Key Drivers

AI training/inference demand explosion: Data center revenue grew 13x since ChatGPT emergence (FY2023). Every major AI model trained on NVIDIA GPUs. Inference deployments now growing alongside training as agentic AI reaches useful intelligence (OpenAI GPT-5.5, Anthropic, Cursor adoption scaling). Frontier AI compute share is rising.
Platform co-design creating generational leaps: Extreme co-design across chips, systems, networking, and software delivers performance/watt improvements that exceed Moore’s Law. CUDA ecosystem (1.5M models) creates massive switching costs. Annual architecture cadence (Hopper → Blackwell → Vera Rubin) compounds the advantage; GB300 delivers 2.7x throughput + 60% cost-per-token reduction vs 6 months ago.
TAM expansion beyond training into networking, sovereign AI, physical AI, and now CPUs: Networking grew to $31B (+3.5x since Mellanox). Sovereign AI deployments now across ~40 countries representing $50T GDP. Automotive/robotics (Cosmos, GR00T, DRIVE) expanding into physical AI (>$9B TTM). Vera CPU opens $200B NEW TAM (FY27).

Quarterly Key Metrics

	Q1'25	Q2'25	Q3'25	Q4'25	Q1'26	Q2'26	Q3'26	Q4'26	Q1'27	Q2'27(E)
Quarter End	Apr 2024	Jul 2024	Oct 2024	Jan 2025	Apr 2025	Jul 2025	Oct 2025	Jan 2026	Apr 2026	Jul 2026
Stock Price	$90	$124	$121	$134	$108	$158	$187	$186	$174	$223
Stock QoQ	+88%	+38%	-2%	+11%	-19%	+46%	+18%	-0%	-6%	+28%
Rev YoY	+262%	+122%	+94%	+78%	+69%	+56%	+62%	+73%	+85%	+95%
OM%	65%	62%	62%	61%	59%	61%	63%	65%	66%	66%
R40 (Rev YoY + OM%)	+327%	+184%	+156%	+139%	+128%	+116%	+126%	+138%	+151%	+161%
TTM OP ($M)	$47,740	$59,582	$71,034	$81,454	$90,503	$100,301	$114,442	$134,707	$162,285	$193,887
TTM OP QoQ	+45%	+25%	+19%	+15%	+11%	+11%	+14%	+18%	+20%	+19%
FCF/OP ttm	82%	79%	80%	75%	80%	72%	68%	72%	73%	77%
Debt/EBITDA	0.2x	0.1x	0.1x	0.1x	0.1x	0.1x	0.1x	0.1x	0.1x	0.1x
P/E(ttm)	53x	59x	48x	46x	35x	48x	50x	45x	34x	34x

Recent Stock Performance & Drivers

Stock currently $223 vs $174.40 Q1 FY27 close = +28% post last quarter end. Powered by record $82B Q1 print + $91B Q2 guide + Vera CPU launch + $80B buyback announcement.
TTM OP surged from $134.7B to $162.3B (+20%) in 2 quarters — driven by Blackwell ramp and Data Center revenue acceleration to +85% YoY.
P/E(ttm) at 34x. Re-rated higher post-print as stock outran earnings; R40 +151% is extraordinary; FCF/OP ttm 73% (vs 72% last quarter); Debt/EBITDA 0.07x.

Opportunities

$3–4T AI infra spend annually by end of decade. Hyperscaler CapEx forecast >$1T in 2027. $1T Blackwell+Rubin revenue visibility 2025-2027 confirmed. Frontier AI customers (Anthropic, OpenAI, Gemini, xAI, Meta) all scaling on Blackwell.
Vera CPU opens $200B NEW TAM — NVDA’s first CPU effort, custom Arm with +1.5x perf/core vs x86. Visibility to ~$20B CPU revenue this year. Co-designed with Rubin GPUs + NVLink, +4x density per rack vs x86.
ACIE growth: $37B (+31% QoQ); AI cloud revenue >3x YoY; sovereign +80% YoY. AI infrastructure deployed across ~40 countries representing $50T GDP. Partner data centers >10MW nearly doubled in 1 year to >80 sites.

Challenges

China headwind continues. U.S. government approved H200 licenses for shipment to China but zero revenue generated yet; uncertain whether imports will be allowed. NO China data center compute revenue in outlook — consistent with last quarter.
Concentration risk + extreme valuation. 34x P/E(ttm) prices in sustained AI capex; Hyperscale ~50% of Data Center revenue. Any AI capex moderation would hit growth rate hard from a $268B+ base.
Supply chain & manufacturing complexity. $145B total supply commitments (inventory + purchase commitments + prepaids). Management acknowledges “not immune to supply challenges”; manufacturing complexity at scale remains the operational risk.

Next 12 Months

Q2 FY27 guidance: $91B ±2% revenue (+12% QoQ), Non-GAAP GM 75%, full-year FY27 GM mid-70s; OpEx upper-40s growth (R&D + AI productivity tools); FY27 tax rate 16-18%.
Vera Rubin production starts H2 FY27 (Q3). 7 purpose-built chips across 5 accelerated racks delivering 35x inference throughput and 10x AI-factory revenue vs Blackwell. Successful ramp validates the annual cadence.
Capital return execution. $80B new buyback + $39B remaining = $119B authorization; dividend +25x to $0.25/share; plan ~50% of FCF returned this year.

Next 12-24 Months

Vera CPU $200B TAM ramp + Anthropic capacity expansion. CEO noted “our coverage of Anthropic has been largely zero until just recently” — deepened partnership across AWS, Azure, CoreWeave, xAI infrastructure.
Hyperscaler CapEx >$1T in 2027 + AI factory build-out across ~40 countries. Microsoft Fairwater live with 100,000+ Blackwell GPUs ahead of schedule. AWS adding 1M+ Blackwell/Rubin GPUs starting this year. Google A5X supports up to 960,000 Rubin GPUs.
Physical AI proliferation. Uber Robotaxi partnership scaling to nearly 30 cities, 4 continents by 2028; robotics, automotive, AI RAN, Blackwell workstations growing. Physical AI revenue >$9B TTM and expanding.

Core Metrics Scorecard

	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	Q2'26	Q3'26	Q4'26	Q1'27	Q2'27(E)
Quarter End	Oct21	Jan22	Apr22	Jul22	Oct22	Jan23	Apr23	Jul23	Oct23	Jan24	Apr24	Jul24	Oct24	Jan25	Apr25	Jul25	Oct25	Jan26	Apr26	Jul26
R40 > 40%	✅	✅	✅					✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
R40 accelerate		✅	✅			✅	✅	✅	✅	✅	✅						✅	✅	✅	✅
TTM OP QoQ > 10%	✅	✅	✅					✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
TTM OP QoQ accelerate			✅					✅									✅	✅	✅
TTM FCF/OP > 60%	✅	✅	✅				✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
Debt/EBITDA < 3.5	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
Total Passes	4	5	6	1	1	2	3	6	5	5	5	4	4	4	4	4	6	6	6	5
Stock QoQ	+4%	+42%	-7%	-44%	-20%	+18%	+94%	+52%	+6%	+8%	+88%	+38%	-2%	+11%	-19%	+46%	+18%	-0%	-6%	+28%
P/E(ttm)	65x	78x	64x	42x	41x	61x	143x	103x	59x	41x	53x	59x	48x	46x	35x	48x	50x	45x	34x	34x

Scorecard at 6 of 6 in Q1’27 — 4th straight quarter of full passes, tied for highest reading in recent history. R40, R40 accel, TTM OP QoQ, OP QoQ accel, FCF/OP, Debt/EBITDA ALL pass. Q2’27E slips to 5/6 as TTM OP QoQ acceleration breaks (+20% → +19%).
Historical pattern: R40 and TTM OP QoQ drive the scorecard. R40 >40% passes in 17 of 20 quarters; TTM OP QoQ >10% in 16 of 20. FCF/OP and Debt/EBITDA always pass (floor of 2). Q2’23–Q4’23 downturn (1–2 passes) was R40+OP QoQ failing simultaneously.
Interpretation given stock move + valuation: Stock +28% post Q1 close on 6/6 scorecard + new segmentation + Vera CPU $200B TAM + $80B buyback. P/E(ttm) 34x is LOW end of 5-year range (29x–80x); $91B Q2 delivery + Vera Rubin H2 ramp = upside path.

Chart: R40 vs Stock Price (Since 2020)

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

Q4 FY2026 Key Data

Revenue +73% YoY, OM 65%, TTM OP $134.7B (+38% over 4 quarters); R40 +138%. FCF/OP 72%, Debt/EBITDA 0.1x. P/E compressed from 45x to 29x as earnings outpaced stock.
$3-4T data center CapEx opportunity by 2030. Top-5 hyperscaler CapEx approaching $700B in CY2026 (up $120B YTD); sovereign AI tripled to $30B+; Blackwell + Rubin revenue opportunity exceeds $500B previously shared.
Vera Rubin platform launching H2 CY2026. Trains MoE models with 1/4 GPUs; cuts inference cost 10x vs Blackwell. First samples shipped. Networking $31B (+3.5x since Mellanox).

Key Takeaways from Earnings Call

AI infrastructure spending is accelerating, not decelerating. Top-5 hyperscaler CapEx approaching $700B in CY2026; sovereign AI tripled to $30B+; every country building AI infrastructure. Blackwell + Rubin revenue opportunity exceeds the $500B previously shared. Inference inflection: agentic AI reached "useful intelligence" — token demand going exponential.
Vera Rubin + Networking + Physical AI extend the moat. Vera Rubin (H2 CY2026) trains MoE models with 1/4 GPUs and cuts inference cost 10x vs Blackwell; every cloud model builder expected to deploy. GB300 NVL72 achieves 50x perf/watt for inference vs Hopper. Networking now $31B (+3.5x since Mellanox). Physical AI (robotics, automotive) emerging as the next growth vector.
China + supply constraints + concentration are the bear case. No data center compute revenue from China assumed in guidance; Chinese competitors making progress with IPO bolster — long-term risk of bifurcated AI ecosystem. CoWoS packaging and HBM memory remain bottlenecks. Top-5 hyperscalers = ~50% of data center revenue; any AI capex deceleration would hit growth from a $268B FY2026 base.

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

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