2026-02-26 - NVDA - Nvidia Corp Deep Dive

Deep Dive Ticker Tape

Article published Feb 26, 2026. Prices below use latest available snapshots.

NVDA $225.01 +10.9% 30d

Conviction: High Status: Holding


Quick Snapshot

Signal Reading
Overall 🟡 AI infrastructure monopoly — post-earnings -5.1% pullback today despite record Q4. RSI 63.3, consolidating.
Moat Wide
Key insight Record Q4 FY2026: $68.1B revenue, $35B FCF in a single quarter. Post-earnings sell-the-news -5.1% today. Price $185.57, RSI 63.3, SMA20 $186.78. Near SMA — healthy consolidation.

Action Matrix

Action Level Why
Current 🔒 Hold Core position +178%, thesis accelerating with each quarter
Entry Zone $175 - $190 Recent consolidation range, RSI 45-50 pullback levels
Stop-Loss $155 (-21%) Below major support, would signal structural breakdown
Target $250 (+27%) FY2027 revenue $300B+ at 25x forward P/E

Price Data

Stock Price 1D 7D 30D 3M 52wkHi RSI Status Action
NVDA $185.57 -5.1% +0.3% -0.5% -5.1% 63 🟡 Consolidating 🔒 Hold

OHLC data as of Feb 26. Post-earnings pullback: -5.1% today after Q4 beat. RSI 63.3 — healthy, not overbought.

Short Interest

Date Short Interest Days to Cover Avg Daily Volume Trend
2026-02-13 254.3M shares 1.4 181.8M Declining (shorts covering)
2026-01-30 257.1M shares 1.6 164.1M
2026-01-15 261.8M shares 1.6 161.0M

Short interest declining from 262M to 254M over past month. Days to cover < 2 = low squeeze risk but also low bearish conviction.

Legend

  • RSI = Oversold (<30) - potential buying opportunity
  • RSI = Overbought (>70) - FOMO warning, may pullback
  • 52wkHi = % from 52-week high (how far below peak)

Company Overview

One-Liner

NVIDIA designs GPUs and data center-scale AI infrastructure that powers the majority of the world's AI training and inference workloads.

Business Model

Question Answer
What they sell GPUs, CPUs, DPUs, networking (NVLink/InfiniBand), full-rack AI systems (Blackwell, Grace), CUDA software platform
Who pays Hyperscalers (AWS, Azure, GCP), AI model makers (OpenAI, Anthropic, Meta), enterprises, sovereign AI initiatives
Revenue model Hardware sales (chips + systems), software licensing (NVIDIA AI Enterprise, vGPU), development services
How sticky Extremely sticky — CUDA ecosystem has 7.5M+ developers, massive switching costs. Entire AI software stack built on NVIDIA's platform.

Key Segments

Segment Revenue % Growth Notes
Data Center ~91% +73% YoY (Q4) AI training + inference, Blackwell ramping, $62.3B in Q4
Gaming ~6% +30% YoY GeForce RTX 50 series launched, neural graphics
Professional Visualization ~1% +56% YoY RTX PRO GPUs for AI workstation use cases
Automotive ~1% +32% YoY DRIVE Hyperion platform, autonomous vehicles

Geographic Mix

Region Revenue % Notes
US ~50% Primary hyperscaler customers
International ~50% Effectively foreclosed from China data center market due to export controls

Business Description (from 10-K, filed 2026-02-25)

NVIDIA describes itself as a "data center scale AI infrastructure company reshaping all industries." The technology stack includes the foundational CUDA development platform, hundreds of domain-specific software libraries, and vertical optimizations across healthcare, telecom, automotive, and manufacturing. The Blackwell architecture features extreme co-design where chips, networking, systems, software, and algorithms are holistically architected for maximum performance at scale. Hundreds of thousands of GPUs can be interconnected to function as a single giant computer.


Competitive Analysis

Industry Position

Question Answer
Market share ~80%+ of AI accelerator market
Market size (TAM) $500B+ AI infrastructure (growing rapidly)
Growth rate 30-40% CAGR through 2030
Key competitors AMD (MI300/Helios), Broadcom (custom ASICs), Intel (Gaudi), Google (TPUs), Amazon (Trainium)
Position Dominant leader

Competitive Moat

Moat Type Present? Evidence
Network effects 🟢 7.5M+ CUDA developers, massive software ecosystem creates virtuous cycle
Switching costs 🟢 Entire AI stack (PyTorch, TensorFlow, JAX) optimized for CUDA. Rewriting for alternatives costs millions in engineering time.
Cost advantages 🟢 Best performance per watt — Blackwell delivers 4x inference throughput vs Hopper at same power
Intangible assets 🟢 10,000+ patents, CUDA platform (18 years of optimization), brand as the "AI standard"
Efficient scale 🟢 Fabless model with TSMC's best process nodes, $76.7B cumulative R&D investment

Moat Assessment

Moat Width: Wide Moat Trend: Stable (widening in systems, narrowing at chip level from custom ASIC competition)

Summary: NVIDIA's moat is the deepest in semiconductors. CUDA's 18-year head start means virtually all AI software is optimized for NVIDIA hardware. The shift to data center-scale systems (Blackwell NVL72, Grace Blackwell) where networking, chips, and software are co-designed makes the full-stack advantage even harder to replicate. AMD's Helios and hyperscaler custom chips are credible threats at the chip level, but NVIDIA's system-level integration and software ecosystem remain unmatched.


Management Assessment

Leadership

Role Name Since Background Notes
CEO Jensen Huang 1993 (co-founder) AMD microprocessor designer, LSI Logic director Visionary, #1 HBR best-performing CEO (2017). Net worth $164B.
CFO Colette M. Kress 2013 Cisco SVP/CFO, Microsoft Server & Tools CFO, Texas Instruments Financial discipline through 65% revenue growth.
EVP Worldwide Field Ops Ajay K. Puri 2005 Sun Microsystems (22 years), Harvard MBA
EVP Operations Debora Shoquist 2007 JDS Uniphase EVP Ops, Quantum, HP
EVP General Counsel Timothy S. Teter 2017 Cooley LLP (20+ years patent litigation), Stanford Law

Founder Involvement

Question Answer
Founder-led? Yes — Jensen Huang has been CEO since founding in 1993
Founder ownership Significant (direct + indirect holdings)
Skin in the game Maximum — identity and company are synonymous

Capital Allocation

Metric Track Record
M&A discipline Good — Mellanox ($7B, 2020) was transformative. ARM deal failed but showed ambition.
Buyback timing Good — $41B returned to shareholders in FY2026
R&D investment High — $76.7B cumulative, $13B in FY2026 alone. 31,000 of 42,000 employees in R&D.
Debt management Conservative — $8.5B long-term debt vs $118.9B equity

Red Flags

  • Excessive exec compensation — Large but aligned with shareholder value creation
  • High turnover in key roles — No (3.7% turnover rate, extremely low)
  • Related party transactions — None noted
  • Aggressive accounting — Clean, straightforward hardware revenue recognition

Financials

Key Metrics

Metric Q3 FY2026 Q2 FY2026 Q1 FY2026 Trend
Revenue $57.0B $46.7B $44.1B 📈 Accelerating
Revenue Growth QoQ +22.0% +6.1% - 📈
Gross Margin 73.4% 72.4% 60.5% 📈 Expanding
Operating Margin 63.2% 60.8% 49.1% 📈 Expanding
Net Margin 56.0% 56.5% 42.6% 📈
EPS (diluted) $1.30 $1.08 $0.76 📈

Full Year FY2026: Revenue $215.9B (+65.5% YoY). GAAP Gross Margin 75.0% in Q4. FCF $97B for the year.

Q4 FY2026 (just reported Feb 25): Record $68.1B revenue, +73% YoY. $35B FCF in a single quarter.

Quality Checks

Check Status Notes
FCF positive? 🟢 $97B annual FCF, $35B in Q4 alone. Best-in-class.
Profitable? 🟢 56% net margins (Q3), 75% gross margins (Q4)
Debt manageable? 🟢 $8.5B long-term debt vs $118.9B equity. D/E = 0.07.
Cash runway Infinite Generating more cash than they can deploy. $41B returned to shareholders in FY2026.

Revenue Quality

Factor Assessment
Recurring % Low (hardware-driven, but replacement cycles + capacity expansion = recurring demand)
Customer concentration Moderate — hyperscalers (40-50% of revenue) are concentrated but spending is non-discretionary
Contract length Multi-year cloud commitments, sovereign AI deals
NRR/NDR N/A (not SaaS), but customers consistently buy more each generation

Leading Indicators

Indicator Value Signal
Revenue Acceleration +15.88% 🟢 Strong
YoY Growth +62.5% 🟢🟢 Very Strong
Earnings Beat Rate 100% 🟢🟢 Perfect
Insider Trading -$3.0B net (0 buys, 100 sells) 🔴 Bearish
Composite 0.7 (Bullish) 🟢

Balance Sheet (Q3 FY2026)

Item Value
Total Assets $161.1B
Total Equity $118.9B
Long-term Debt $8.5B
Current Assets $116.5B
Current Liabilities $26.1B
Current Ratio 4.47
Debt/Equity 0.07
Inventory $19.8B

Cash Flow (Q3 FY2026)

Item Value
Operating Cash Flow $23.8B
Investing Cash Flow -$9.0B
Financing Cash Flow -$14.9B
Est. FCF (OCF + ICF) $14.7B
FCF Margin ~25.8%

Valuation

Current Multiples

Metric Current 5Y Avg Industry Avg
P/E (trailing) 39.7 63.9 25
P/E (forward) 23.7 - 20
P/S ~22 - 5-8
EV/EBITDA 35-38 - 15-20
EV/FCF 48.6 - -

Historical Context

Current trailing P/E of 39.7 is significantly below the 5-year average of 63.9, reflecting the fact that earnings have grown much faster than the stock price. Forward P/E of ~24 is approaching reasonable territory for a company growing revenue 65%+ YoY.

Fair Value Estimate

Method Fair Value Upside/Downside
Forward P/E (25x FY2027E $8.50 EPS) $213 +8%
Revenue multiple (20x FY2027E $300B) $245 +24%
DCF ($97B FCF growing 25%/yr, 10% discount) $250+ +27%

Fair Value: $230 (blended estimate, based on sustained 30%+ growth)


Bull Case

Why This Could Work (5 Reasons)

  1. AI infrastructure spending is accelerating, not decelerating

    • Evidence: Q4 revenue $68.1B was +73% YoY, Q1 guidance crushed estimates
    • Implication: Hyperscalers committing to hundreds of billions in AI capex through 2027+
  2. Blackwell ramping + Rubin on deck

    • Evidence: Blackwell Ultra launched FY2026, Rubin production starting H2 FY2027
    • Implication: Each generation delivers 10x+ inference improvement, ensuring customers keep upgrading
  3. Best-in-class financial profile — $97B FCF, 75% gross margins

    • Evidence: FY2026 FCF of $97B, $41B returned to shareholders
    • Implication: Can fund massive R&D while returning capital. Self-funding growth machine.
  4. CUDA ecosystem moat deepening

    • Evidence: 7.5M+ developers, 6,000+ applications, 78% of TOP500 supercomputers
    • Implication: Switching costs increase with each application optimized for NVIDIA
  5. Sovereign AI and enterprise adoption expanding TAM

    • Evidence: Countries building national AI infrastructure on NVIDIA platforms
    • Implication: New multi-billion-dollar demand vector beyond hyperscalers

Upside Scenario

If This Happens Stock Could
FY2027 revenue exceeds $300B, margins hold $250-275 (+27-40%)
Inference demand explodes with agentic AI $300+ (+52%)

Bear Case

What Could Go Wrong (5 Risks)

  1. Custom ASIC competition from hyperscalers

    • How it plays out: Google TPUs, Amazon Trainium, Meta's AMD Helios deal reduce NVIDIA dependency for inference workloads
    • Probability: Medium — happening gradually, but NVIDIA's full-stack advantage limits displacement
  2. Export controls permanently foreclose China market

    • How it plays out: NVIDIA cannot sell competitive data center products in China, competitors (Huawei) fill the gap
    • Probability: High — already effectively foreclosed, per 10-K filing
    • Impact: China was historically 20-25% of revenue. Lost opportunity benefits competitors.
  3. AI capex cycle slowdown / correction

    • How it plays out: Hyperscalers pull back on spending if AI ROI doesn't materialize fast enough
    • Probability: Low-Medium — all signals point to accelerating spend through 2027, but cycles always end eventually
  4. Gross margin compression from competition

    • How it plays out: AMD MI300 offers 20-30% lower cost, Broadcom custom ASICs pressure pricing
    • Probability: Medium — AMD held ~7% share in Q3 2025, growing slowly
  5. Heavy insider selling (-$3.0B net)

    • How it plays out: Insiders selling into strength is normal for a $5T company, but sustained selling can signal caution
    • Probability: Low risk — likely routine portfolio diversification at these levels

Thesis Killers

  • Gross margins decline below 65% for 2+ consecutive quarters (would signal pricing power erosion)
  • Major hyperscaler publicly shifts >50% of AI compute to custom silicon
  • Revenue growth decelerates to <20% YoY (signals demand saturation)

Downside Scenario

If This Happens Stock Could
AI capex cycle peaks, growth slows to 20% $150 (-24%) — multiple compression
Custom ASICs take 30%+ of inference market $130 (-34%) — structural share loss

Key Risk Factors (from 10-K, filed 2026-02-25)

From the most recent 10-K filing: NVIDIA is "effectively foreclosed from competing in China's data center computing/compute market" and notes that "lost opportunity and the benefit to our competitors will have a material and adverse impact." The Q1 FY2026 H20 writedown was $4.5B. The USG replacement for the AI Diffusion IFR remains uncertain in scope and timing.


Market-Moving News

Historical (What Moved This Stock Before)

Date Event Impact Market Reaction Lesson
2024-06 10-for-1 stock split Neutral Positive sentiment, no fundamental change Splits drive retail interest
2025-01 DeepSeek AI efficiency breakthrough -17% drawdown Overreaction — recovered within weeks Efficient AI = more demand for compute, not less
2025-04 H20 export ban to China, $4.5B writedown -15% Partially recovered China loss was already priced in, market focused on rest of world
2025-10 Q3 FY2026: $57B revenue, beat estimates +5% Justified rally Data center demand robust
2026-02-25 Q4 FY2026: $68.1B record revenue, guidance beat +4% (AH) Positive but muted — high expectations Market demands perfection at this valuation

Recent News

Date Headline Impact Relevance
2026-02-25 Q4 FY2026: $68.1B revenue, $97B FY FCF, guidance beat 🟢 Record quarter, AI demand accelerating
2026-02-24 Meta signs AMD Helios deal alongside NVDA commitment 🟡 Diversification but NVDA remains primary
2026-02-25 10-K filed: Rubin platform shipping H2 FY2027 🟢 Next-gen on track, 10x cost/token reduction vs Blackwell
2026-01 CES 2026: Jensen confirms Vera Rubin in production 🟢 Roadmap clarity, annual cadence maintained

News Patterns

NVIDIA trades on data center revenue growth and guidance. Every earnings beat drives modest rallies (market expects perfection). Supply chain concerns (TSMC, export controls) cause sharp but temporary drawdowns. DeepSeek-style "AI efficiency" scares create buying opportunities.


Symbol Name Relationship Weight Notes
SMH VanEck Semiconductor ETF Top holding 18.7% Best semi-sector proxy
SOXX iShares Semiconductor ETF Top holding 9.9% Broader semi exposure
SPY SPDR S&P 500 Major component 6.6% $41.9B market value in fund
XLK Technology Select SPDR 3rd largest holding 4.7% Tech sector proxy
BOTZ Global X Robotics & AI 2nd largest holding 9.4% AI/robotics focused

Sector Context

Semis outperforming broader tech. NVDA ecosystem scan (Feb 16) showed NVDA RSI 47 (neutral), TSM RSI 66 (strong), ARM RSI 63 (strong). AMD lagging at RSI 31. The chip cycle remains intact — NVDA and its supply chain are healthier than the growth/SaaS selloff suggests.


Cross-References (Where This Appears in Our System)

Location File Context
Holdings holdings/stocks/2026-02-16-stocks.md NVDA $182.81, +178% gain, At SMA, Hold
Scans scans/2026-02-16-nvda-ecosystem.md NVDA ecosystem scan — semis resilient vs growth selloff
Scans scans/2026-02-23-insider-scan.md -$3.0B insider selling, heavy but routine
Scans scans/2026-02-16-ai-scan.md RSI 47, neutral, best relative strength in AI space
Candidates candidates/2026-02-05-portfolio.md Accumulate — DCA into quality on weakness
Watchlists mag7, ai-infra Core position across multiple lists

Ticker Why Related Suggested Placement Priority Notes
AMD GPU competitor, MI300/Helios candidates/nvda-ecosystem/ 🟢 RSI 31, oversold — value play vs NVDA
TSM Primary foundry (TSMC) candidates/nvda-ecosystem/ 🟢 RSI 66, strong — benefits from NVDA volume
ARM IP/design partner, edge AI candidates/nvda-ecosystem/ 🟢 RSI 63, strong — mobile + data center
AVGO Custom ASIC competitor (Broadcom) candidates/nvda-ecosystem/ 🟡 Primary ASIC threat
MU Memory supplier (HBM) candidates/nvda-ecosystem/ 🟡 Direct NVDA supply chain play
MRVL Custom networking/ASICs candidates/nvda-ecosystem/ 🟡 Networking for AI data centers

Catalysts & Timing

Upcoming Catalysts

Date Event Impact Watch For
2026-02-25 Q4 FY2026 Earnings 🟢 Beat $68.1B revenue, guidance crushed
2026-05 (est.) Q1 FY2027 Earnings 🟢 Blackwell Ultra ramp, guidance trajectory
H2 FY2027 Rubin platform ships 🟢 Next-gen transition, 10x inference improvement
2026 USG export control replacement rule 🟡 Scope and restrictions unknown — risk factor
2026-03 GTC Conference (expected) 🟢 Product roadmap updates, customer wins

Key Dates

Event Frequency Next Date
Earnings Quarterly ~May 2026 (Q1 FY2027)
GTC Conference Annual March 2026 (expected)
Dividend Quarterly Ex-date 2026-03-11, $0.01/share

Entry Strategy

Position Sizing

Conviction Allocation
Speculative 0% (research only)
Low 1-2%
Medium 3-5%
High 5-10%

My conviction: High Target allocation: Already held, ~6% of portfolio

Entry Approach

Strategy Details
Entry zone $175 - $190 (recent consolidation support)
Starter position Already held — full position
Add on RSI < 35 pullback (like Feb 4-5 dip to $171-$174)
Full position at Current allocation appropriate

Technical Levels

Level Price Notes
Support 1 $185 SMA-20, recent consolidation floor
Support 2 $172 Feb 4-5 low, strong buying appeared
Resistance $207 All-time high (Oct 29, 2025)
52-week high $207.03
52-week low ~$90 April 2025 (DeepSeek/export control selloff)

Research Checklist

Fundamentals

  • Read most recent 10-K — Filed 2026-02-25, FY2026
  • Review Q4 FY2026 earnings — Record $68.1B revenue
  • Understand revenue drivers — Data Center 91%, Blackwell ramp
  • Review financial statements via Massive/Polygon data

Competitive

  • Map competitive landscape — AMD, Broadcom, hyperscaler ASICs
  • Assess moat durability — Wide, CUDA ecosystem deepening
  • Check customer concentration details

Management

  • Research CEO/CFO background — Founder-led since 1993
  • Check insider ownership — Heavy selling (-$3B) but routine
  • Review capital allocation — $41B returned, $13B R&D

Valuation

  • Review historical multiples — Trailing P/E 39.7 vs 5Y avg 63.9
  • Compare to peers — Premium justified by growth/margins
  • Identify entry/exit zones — $175-190 entry, $155 stop

Risk

  • Write bear case — ASIC competition, China ban, capex cycle risk
  • Identify thesis-killers — Margins <65%, revenue growth <20%
  • Set stop-loss level — $155 (-21%)

Sources

Type Link Notes
Investor Relations https://investor.nvidia.com
SEC Filings (10-K FY2026) Filed 2026-02-25 Via Massive/Polygon API
Q4 FY2026 Earnings NVIDIA Newsroom, CNBC Feb 25, 2026
Valuation Data StockAnalysis, MacroTrends, Yahoo Finance

Research Log

Date Update
2026-02-26 Created deep dive with full Massive financial data (income statements, balance sheets, cash flow, short interest, 10-K sections), Q4 FY2026 earnings, and cross-references
2026-02-26 Price refresh (same day). Post-earnings pullback: -5.1% to $185.57. RSI 63.3, SMA20 $186.78. Near SMA consolidation — healthy. 30D -0.5%, 7D +0.3%. Core position +178% hold maintained.

The Gold

Key Discoveries

Discovery Implication
NVIDIA generated $97B FCF in FY2026 — more than most companies' total revenue Can self-fund any R&D program, return massive capital, and acquire strategically
Trailing P/E (39.7) is 38% below 5-year average (63.9) Earnings growth has massively outpaced stock price — valuation is compressing naturally
Export controls effectively foreclose China data center market Lost 20-25% addressable market, but rest-of-world demand more than compensates for now
42,000 employees, 31,000 in R&D (74%), 3.7% turnover Elite engineering culture with exceptional retention — moat is the people
Short interest declining (262M to 254M over 30 days) Bears gradually capitulating
Q4 gross margin expanded to 75.0% despite Blackwell ramp No margin compression from product transitions — pricing power intact

Open Questions

  • How does the USG export control replacement rule (rescinding AI Diffusion IFR) affect NVIDIA's product scope?
  • What is the H200 China licensing program's actual revenue potential?
  • At what point does custom ASIC adoption reach a tipping point for inference workloads?
  • How will the Rubin/Vera Rubin transition affect gross margins in FY2027?
  • What is NVIDIA's actual customer concentration (% of revenue from top 5 customers)?

Mistakes (If Applicable)

Mistake Lesson
N/A — holding since ~$65, +178% Patience in high-conviction names pays off