NVIDIA Corporation — Through the Warren Buffett Lens
Thesis
NVIDIA designs the essential computing engines—GPUs and full-stack systems—that power the artificial intelligence revolution. Its durable edge is not just silicon but a deeply entrenched software ecosystem (CUDA) that creates high switching costs for developers and data center architects. The business generates extraordinary owner earnings, with net income of $29.76B on revenue of $60.92B in fiscal 2026, and returns on tangible capital that are rarely seen. The current price of $195.53 implies a market capitalization near $4.8 trillion. The one number that matters most is the sustainability of data center revenue growth, which more than doubled last year, as the world builds out AI infrastructure. The question is whether this growth is a one-time build-out or the beginning of a decades-long compounding stream. Supporting evidence: • Fiscal 2026 revenue reached $60.92B, more than doubling from $26.97B in fiscal 2025, driven by data center platform adoption (Source 1, Source 2). • Gross margin expanded to 75.0% in fiscal 2025 and remained near 75% in subsequent quarters, demonstrating pricing power and scale advantages (Source 1, Source 3). • The company returned $41.1B to shareholders in fiscal 2026 through buybacks and dividends, while still investing heavily in R&D, signaling both capital discipline and growth reinvestment (Source 7). Key uncertainties: • Can data center revenue growth be sustained at current rates, or will hyperscaler capex normalize after the initial AI build-out? Resolution: monitor quarterly data center revenue growth rates over the next 4-6 quarters. • Does the CUDA ecosystem lock-in truly constitute a durable moat, or could alternative software frameworks (e.g., open-source ROCm) erode switching costs over time? Resolution: track developer surveys and adoption metrics for competing platforms. Management & culture: Jensen Huang co-founded NVIDIA in 1993 and has served as CEO since inception, providing unusual continuity and technical vision. His long tenure and deep engineering background suggest a founder-led culture focused on innovation cycles rather than quarterly earnings. CFO Colette Kress has been with the company since 2013, adding financial stability. The management commentary emphasizes full-stack solutions and ecosystem lock-in, which aligns with a long-term orientation. However, the rapid growth and promotional language around AI ('industrial revolution') warrant caution. Capital allocation is mixed: massive R&D spending is essential for maintaining the technology lead, but the aggressive buyback pace at historically high valuations may not reflect the conservatism this lens prefers. There is no evidence of self-correction or admission of mistakes in the provided sources, which is a minor concern. Overall, management appears competent and aligned, but the lens would prefer a more conservative approach to buybacks and less promotional rhetoric.
Key Value Drivers
- Data center revenue growth: hyperscaler and enterprise AI infrastructure spending
- Gross margin sustainability: pricing power and product mix (Blackwell, Rubin platforms)
- CUDA ecosystem stickiness: developer lock-in and software moat
- Capital allocation: buyback pace and R&D reinvestment returns
- Geopolitical risk management: export controls and supply chain diversification
Key Risks
- Hyperscaler capex digestion: AI infrastructure spending could plateau or decline after the initial build-out, causing a sharp drop in data center revenue. (Trigger: Data center revenue growth falls below 10% year-over-year for two consecutive quarters.)
- CUDA ecosystem erosion: open-source or competing software frameworks reduce developer lock-in, commoditizing NVIDIA's hardware pricing. (Trigger: Third-party surveys show CUDA developer share declining by more than 5 percentage points in a year.)
- Geopolitical export restrictions: further tightening of US-China technology controls could block a significant portion of NVIDIA's addressable market. (Trigger: New USG regulations explicitly ban sales of data center GPUs to all Chinese entities, and NVIDIA discloses a material revenue impact.)
- Customer concentration: a few large hyperscalers account for a disproportionate share of revenue; loss of any one could materially harm results. (Trigger: Any single direct customer exceeds 25% of total revenue for two consecutive quarters and then reduces orders by more than 30%.)
- Gross margin compression: competition or product transition costs could erode the exceptional 75% gross margin. (Trigger: Gross margin falls below 65% for two consecutive quarters.)
Key Metrics to Monitor
- Owner earnings (NI + D&A − maint capex) — not disclosed in fetched sources
- ROE (5yr average) — not disclosed in fetched sources
- ROIC — not disclosed in fetched sources
- Gross margin trend (5yr): 75.0% in FY2025, up from 60.5% in FY2024 (Source 1)
- Retained earnings / market cap — not disclosed in fetched sources
- Debt / equity — not disclosed in fetched sources
- Moat description (type + durability): ecosystem switching costs (CUDA); durability uncertain given rapid tech evolution
- Capital allocation grade (buybacks / debt / M&A): B (aggressive buybacks at high valuations, limited debt, no major M&A)
Want to see how different lenses come to different conclusions?
Run your own thesis →