NVIDIA Corporation ($NVDA)
AI Infrastructure | Data Center · GPU · Networking | FY2026 Revenue: $215.9B
Sources: SEC EDGAR 10-K (FY2026, filed Feb 2026; FY2025), Q4 FY2026 Earnings Release & Call (Feb 25, 2026), Q1 FY2027 Earnings Release (May 20, 2026). CIK: 0001045810
Company Overview & Thesis
NVIDIA Corporation (NASDAQ: $NVDA) is a semiconductor and accelerated computing company headquartered in Santa Clara, California. Founded in 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem, NVIDIA designs GPUs, system-on-chip units, and full-stack AI infrastructure platforms. It is fabless — all wafer fabrication is contracted to TSMC. As of Q1 FY2027, NVIDIA is the world's most valuable publicly traded semiconductor company, with a market capitalization of approximately $3.3–3.6T.
NVIDIA operates in two reporting segments: Compute & Networking (Data Center GPUs, networking/interconnect, DGX systems, DRIVE, robotics) and Graphics(GeForce gaming GPUs, Quadro/RTX Pro workstation GPUs, Omniverse, licensing). For external reporting, NVIDIA breaks out five market platforms: Data Center, Gaming & AI PC, Professional Visualization, Automotive & Robotics, and OEM & Other.
The central thesis: NVIDIA is the dominant platform for AI infrastructure — the only vendor that ships a co-designed, full-stack AI factory (GPU compute + NVLink + networking + software stack) at hyperscale. The CUDA software moat (millions of developers, decades of optimized libraries) creates a high switching cost. The Blackwell architecture has extended NVIDIA's performance-per-watt leadership over AMD (MI300X) and custom silicon (Google TPU, Amazon Trainium) in AI training and inference workloads. FY2026 Data Center revenue of $193.7B (+68% YoY) — scaling nearly 13x since the ChatGPT inflection in FY2023 — is the defining financial metric of the current AI infrastructure buildout.
Revenue & Financial Metrics
| Metric | FY2026 | FY2025 | Change |
|---|---|---|---|
| Total Revenue | $215,938M | $130,497M | +65% YoY |
| Cost of Revenue | $62,475M | $32,639M | +91% |
| Gross Profit | $153,463M | $97,858M | +57% |
| Gross Margin (GAAP) | 71.1% | 75.0% | -390 bps |
| Gross Margin (Non-GAAP) | 71.3% | 75.5% | -420 bps |
| Research & Development | $18,497M | $12,914M | +43% |
| Sales, General & Administrative | $4,579M | $3,491M | +31% |
| Total Operating Expenses | $23,076M | $16,405M | +41% |
| GAAP Operating Income | $130,387M | $81,453M | +60% |
| GAAP Operating Margin | 60.4% | 62.4% | -200 bps |
| Non-GAAP Operating Income | $137,300M | $86,789M | +58% |
| Non-GAAP Operating Margin | 63.6% | 66.5% | -290 bps |
| Interest Income | $2,300M | $1,786M | +29% |
| GAAP Net Income | $120,067M | $72,880M | +65% |
| GAAP Net Margin | 55.6% | 55.8% | -20 bps |
| Non-GAAP Net Income | $116,997M | $74,265M | +58% |
| GAAP EPS (Diluted) | $4.90 | $2.94 | +67% |
| Non-GAAP EPS (Diluted) | $4.77 | $2.99 | +60% |
| Free Cash Flow | $97,000M | ~$60,900M | +59% |
| Metric | Jan 2026 | Jan 2025 |
|---|---|---|
| Cash & Cash Equivalents | $10,605M | $8,563M |
| Marketable Debt Securities | $39,065M | $26,237M |
| Marketable Equity Securities | $12,886M | — |
| Total Current Assets | $125,605M | $73,300M |
| Total Assets | $206,803M | $111,600M |
| Short-Term Debt | $999M | $1,250M |
| Long-Term Debt | $7,469M | $8,462M |
| Total Shareholders' Equity | $157,293M | $42,978M |
| Free Cash Flow (FY) | $97,000M | ~$60,900M |
| Capital Returned to Shareholders | $41,100M | $15,400M |
| Share Repurchase Authorization (remaining) | $58,500M | N/A |
| Period | Total Rev. | Data Center | Gaming | ProViz | Auto | Gross Margin |
|---|---|---|---|---|---|---|
| Q4 FY2025 | $39,331M (+78% YoY) | $35,600M (+93%) | $2,500M (-11%) | $511M (+10%) | $570M (+103%) | 73.0% |
| Q1 FY2026 | $44,062M (+69% YoY) | ~$39,900M | ~$2,500M | ~$550M | ~$567M | 60.5% |
| Q2 FY2026 | $46,743M (derived) | ~$43,300M | ~$3,400M | ~$570M | ~$550M | ~72% |
| Q3 FY2026 | $57,006M (+94% YoY) | ~$51,100M | ~$4,300M | ~$770M | ~$590M | 73.4% |
| Q4 FY2026 | $68,127M (+73% YoY) | $62,300M (+75%) | $3,700M (+47%) | $1,300M (+159%) | $604M (+6%) | 75.0% |
| Q1 FY2027 | $81,615M (+85% YoY) | $75,200M (+92%) | ~$5,300M | ~$1,100M | ~$600M | 74.9% |
Q1–Q3 FY2026 segment figures partially estimated from available segment-level data. Q4 FY2025, Q4 FY2026, Q1 FY2027 exact from earnings releases. Q2 FY2026 revenue derived: FY2026 total minus Q1+Q3+Q4.
Key Business Segments
| Platform | FY2026 | FY2025 | YoY | % of Total |
|---|---|---|---|---|
| Data Center (total) | $193,700M | $115,200M | +68% YoY | 89.7% |
| — Compute | ~$162,700M (est.) | ~$99,500M (est.) | ~+64% | ~75% |
| — Networking | $31,000M+ (>10x FY21) | N/A | >10x FY21 | ~14%+ |
| Gaming & AI PC | $16,000M | $11,400M | +41% YoY | 7.4% |
| Professional Visualization | $3,200M | $1,900M | +70% YoY | 1.5% |
| Automotive & Robotics | $2,300M | $1,700M | +39% YoY | 1.1% |
| OEM & Other | ~$738M | ~$297M | ~+149% | 0.3% |
NVIDIA's primary business: GPU compute (H100, H200, Grace Blackwell NVL72/GB300), networking (InfiniBand, NVLink, Spectrum-X Ethernet, BlueField DPUs, ConnectX NICs), and DGX/HGX server systems. Customers span public cloud providers (AWS, Azure, GCP, OCI), hyperscalers (Meta, Microsoft, xAI, OpenAI), AI model makers, enterprises, and sovereign nations. Q4 FY2026 DC revenue: $62.3B (+75% YoY). Q1 FY2027: $75.2B (+92% YoY). Compute ($60.4B in Q1 FY2027, +77% YoY) and Networking ($14.8B in Q1 FY2027, +199% YoY) are both growing rapidly.
GeForce RTX 50 Series (Blackwell architecture) gaming GPUs for desktops and laptops. DLSS 4 with Multi Frame Generation is the key differentiator — delivering up to 8x performance multiplier via AI-based rendering. GeForce RTX is also the local inference platform for AI PC workflows (LLMs, image generation, local agents). FY2026 revenue: $16.0B (+41% YoY). Q4 FY2026: $3.7B (+47% YoY). Segment remains supply-constrained in early FY2027 as Blackwell consumer GPUs compete for wafer capacity with Data Center.
RTX PRO workstation GPUs (Blackwell, up to 72GB VRAM), Omniverse digital twin platform, and DGX Spark personal AI supercomputers. Targeted at AI developers, designers, engineers, and content creators. Q4 FY2026 revenue: $1.3B (+159% YoY, +74% QoQ) — the first quarter crossing $1B — driven by Blackwell workstation demand. Full-year FY2026: $3.2B (+70% YoY). Enterprise adoption of Omniverse and RTX PRO for agentic workflows is the growth driver.
NVIDIA DRIVE platform (DriveOS, DRIVE AGX Orin, DRIVE Hyperion) for Level 2+ through Level 4 autonomous vehicles. Automotive FY2026: $2.3B (+39% YoY). Partners include Mercedes-Benz, Toyota, BYD, Hyundai/Kia, Uber, Waymo, and Zoox. NVIDIA Cosmos (world foundation models for physical AI) and Isaac GR00T (humanoid robot policies) are expanding the addressable market to industrial robotics and manufacturing. Physical AI contributed >$6B in total NVIDIA revenue in FY2026.
| Metric | Value | Period |
|---|---|---|
| Data Center Revenue — FY2026 | $193,700M | FY2026 (12-mos ended Jan 25, 2026) |
| Data Center Revenue — FY2025 | $115,200M | FY2025 (12-mos ended Jan 26, 2025) |
| Data Center YoY Growth | +68% | FY2026 vs FY2025 |
| Data Center Revenue — Q4 FY2026 | $62,300M | Q4 FY2026 (ended Jan 25, 2026) |
| Data Center Revenue — Q1 FY2027 | $75,200M | Q1 FY2027 (ended Apr 26, 2026) |
| DC Compute Revenue Q1 FY2027 | $60,400M (+77% YoY) | Q1 FY2027 |
| DC Networking Revenue Q1 FY2027 | $14,800M (+199% YoY) | Q1 FY2027 |
| Networking Revenue FY2026 (full year) | $31,000M+ | FY2026 |
| Sovereign AI Revenue FY2026 | >$30,000M (>3x YoY) | FY2026 |
| Physical AI Revenue FY2026 | >$6,000M | FY2026 |
| Blackwell % of DC Revenue (Q4 FY26) | ~67% (GB-based) | Q4 FY2026 |
| Blackwell Deployed Capacity | ~9 gigawatts | As of Feb 2026 |
| Top-5 Cloud Providers % of DC Revenue | >50% | FY2026 |
| Analyst 2026 CapEx Forecast (top-5 CSPs) | ~$700B+ | Calendar 2026 est. |
SEC Filing Key Findings
FY2026 total revenue reached $215,938M — a 65% YoY increase from $130,497M in FY2025. This represents a ~13x increase in Data Center revenue since the emergence of ChatGPT in fiscal year 2023 (FY2023 total revenue: ~$26.9B). Data Center alone at $193.7B exceeds FY2025 total company revenue of $130.5B.
FY2026 10-K, filed Feb 2026 (CIK: 0001045810); Q4 FY2026 Earnings Release, Feb 25, 2026
Grace Blackwell (GB200/GB300 NVL72) systems accounted for roughly two-thirds of Data Center revenue in Q4 FY2026. As of February 2026, approximately 9 gigawatts of Blackwell infrastructure had been deployed by cloud service providers, hyperscalers, AI model makers and enterprises. The Blackwell ramp drove Q4 FY2026 gross margin recovery to 75.0%, up from 60.5% in Q1 FY2026 when early Blackwell production costs weighed on margins.
Q4 FY2026 Earnings Call, Feb 25, 2026; FY2026 10-K (CIK: 0001045810)
NVIDIA generated $97 billion in free cash flow in FY2026 and returned $41.1 billion (43% of FCF) to shareholders via share repurchases and dividends. As of January 25, 2026, $58.5 billion remained under the share repurchase authorization. On May 18, 2026, the Board authorized an additional $80.0 billion repurchase authorization. The quarterly dividend was raised from $0.01 to $0.25 per share in May 2026.
Q4 FY2026 Earnings Release Feb 25, 2026; Q1 FY2027 Earnings Release, May 20, 2026
Sovereign AI revenue (governments and national entities building AI infrastructure) exceeded $30 billion in FY2026, more than tripling year-over-year. Key sovereign customers included Canada, France, the Netherlands, Singapore and the UK. NVIDIA views sovereign AI as an enduring market growing at least in line with total AI infrastructure spending.
Q4 FY2026 Earnings Call, Feb 25, 2026
U.S. export regulations restrict sales of advanced AI compute chips (H100, H200, Blackwell) to China. No Data Center compute revenue from China is assumed in guidance. NVIDIA noted a small number of H200 products were approved for China-based customers but no revenue had been generated. NVIDIA's filing period also noted loss of Chinese market share to domestic competitors (bolstered by recent IPOs) as a long-term structural risk.
FY2026 10-K Risk Factors; Q4 FY2026 Earnings Call
FY2026 GAAP gross margin compressed to 71.1% from 75.0% in FY2025, primarily due to the ramp of Blackwell systems — which have higher initial production costs — in the first half of FY2026. By Q4 FY2026, gross margins recovered to 75.0% as Blackwell yields improved. Non-GAAP gross margin for Q1 FY2027 guidance is 75.0%, consistent with the Q4 FY2026 exit rate.
FY2026 10-K; Q4 FY2026 Earnings Call; Q1 FY2027 Guidance
NVIDIA's networking business (InfiniBand, Spectrum-X Ethernet, NVLink switches, BlueField DPUs, ConnectX NICs) exceeded $31 billion in FY2026 revenue — a greater than 10x increase versus FY2021 (when Mellanox was acquired for ~$6.9B). Q1 FY2027 networking revenue reached $14.8B, up 199% YoY, driven by NVLink scale-up fabric and Spectrum-X adoption in AI factories.
Q4 FY2026 Earnings Call; Q1 FY2027 Earnings Release
The top five cloud providers and hyperscalers collectively account for slightly over 50% of Data Center revenue. NVIDIA's SEC filings disclose customer concentration risk — in FY2026, the XBRL data shows two Compute & Networking segment customers each exceeding 10% of segment revenue. Accounts receivable concentration from three customers was also disclosed as material in the FY2026 10-K.
FY2026 10-K (CIK: 0001045810); XBRL data
Earnings Call Intelligence
Sources: Q4 FY2026 Earnings Call, February 25, 2026 (Jensen Huang, CEO; Colette Kress, CFO). Q1 FY2027 Earnings Release, May 20, 2026.
Computing demand is growing exponentially — the agentic AI inflection point has arrived. Grace Blackwell with NVLink is the king of inference today — delivering an order-of-magnitude lower cost per token — and Vera Rubin will extend that leadership even further. Enterprise adoption of agents is skyrocketing. Our customers are racing to invest in AI compute — the factories powering the AI industrial revolution and their future growth.
Q4 FY2026 Earnings Release Statement, February 25, 2026 — Announcing record full-year and Q4 FY2026 results.
The buildout of AI factories — the largest infrastructure expansion in human history — is accelerating at extraordinary speed. Agentic AI has arrived, doing productive work, generating real value and scaling rapidly across companies and industries. NVIDIA is uniquely positioned at the center of this transformation as the only platform that runs in every cloud, powers every frontier and open source model, and scales everywhere AI is produced — from hyperscale data centers to the edge.
Q1 FY2027 Earnings Release Statement, May 20, 2026 — Announcing record Q1 FY2027 results of $81.6B revenue.
We generated free cash flow of $35 billion in Q4 and $97 billion in fiscal year 2026. For the year, we returned $41 billion, or 43% of free cash flow to our shareholders in the form of share repurchases and dividends. On a full year basis, data center generated revenue of $194 billion, up 68% year-over-year. We have now scaled our data center business by nearly 13x since the emergence of ChatGPT in fiscal 2023.
Q4 FY2026 Earnings Call Prepared Remarks, February 25, 2026 — Presenting FY2026 full-year results and capital returns.
Networking, a cornerstone of our data center scale infrastructure offering, was a standout this quarter, generating $11 billion in revenue, up more than 3.5x year-over-year. Demand for our scale up and scale out technologies reached record levels, both growing double digit sequentially, driven by strong adoption of NVLink, Spectrum-X Ethernet and InfiniBand. For the full year, our networking business exceeded $31 billion in revenue, up more than 10x compared to fiscal 2021, the year we acquired Mellanox.
Q4 FY2026 Earnings Call Prepared Remarks, February 25, 2026 — Discussing Q4 FY2026 networking performance.
In fiscal year 2026, our sovereign AI business more than tripled year-over-year, exceeding $30 billion, driven primarily by customers based in Canada, France, the Netherlands, Singapore and the UK. Every country will build and operate some parts of its AI infrastructure, just like with electricity and Internet today.
Q4 FY2026 Earnings Call Prepared Remarks, February 25, 2026 — Discussing the sovereign AI opportunity.
NVIDIA produces the lowest cost per token and data centers running on NVIDIA generate the highest revenues. GB300 NVL72 achieving up to 50x performance per watt and 35x lower cost per token, compared with Hopper. Continuous optimization of CUDA software helped deliver up to five times better performance on GB200 NVL72 just within four months.
Q4 FY2026 Earnings Call Prepared Remarks, February 25, 2026 — Citing SemiAnalysis InferenceX benchmark results.
| Metric | Value | Period | Source |
|---|---|---|---|
| Q2 FY2027 Revenue Guidance | $91.0B (±2%) | Q2 FY2027 (ends ~Jul 2026) | Q1 FY2027 Earnings Release, May 20, 2026 |
| Q2 FY2027 GAAP Gross Margin | 74.9% (±50 bps) | Q2 FY2027 | Q1 FY2027 Earnings Release |
| Q2 FY2027 Non-GAAP Gross Margin | 75.0% (±50 bps) | Q2 FY2027 | Q1 FY2027 Earnings Release |
| Q2 FY2027 GAAP OpEx | ~$8.5B | Q2 FY2027 | Q1 FY2027 Earnings Release |
| Q2 FY2027 Non-GAAP OpEx | ~$8.3B | Q2 FY2027 | Q1 FY2027 Earnings Release |
| FY2027 GAAP Tax Rate | 16.0%–18.0% | FY2027 full year | Q1 FY2027 Earnings Release |
| Sequential Revenue Growth Visibility | Throughout calendar 2026+ | Calendar 2026–2027 | Q4 FY2026 Earnings Call, Feb 25, 2026 |
| FY2027 Non-GAAP OpEx Growth | Low 40s % YoY | FY2027 full year | Q4 FY2026 Earnings Call |
| Vera Rubin Production Shipments | H2 Calendar 2026 | Expected start | Q4 FY2026 Earnings Call |
| China DC Compute Revenue Assumption | $0 (no H200/H20 approval assumed) | FY2027 ongoing | Q1 FY2027 Earnings Release |
Risk Factors
U.S. export regulations prohibit NVIDIA from selling H100, H200, and Blackwell GPU systems to Chinese customers without licenses. NVIDIA assumes zero Data Center compute revenue from China in its ongoing guidance. Chinese domestic competitors (including entities with recent IPOs) are gaining technical capability and domestic market share, with potential to disrupt NVIDIA's global AI compute dominance over the long term.
NVIDIA is a fabless semiconductor company — all leading-edge wafer production is contracted to TSMC (N4, N3 process nodes). Any disruption to TSMC, including geopolitical conflict involving Taiwan, natural disaster, or production issues, would severely limit NVIDIA's ability to ship product. NVIDIA's 10-K identifies sole-source supply arrangements as a material risk.
The top five cloud hyperscalers (AWS, Azure, GCP, OCI, CoreWeave) account for over 50% of Data Center revenue. A reduction in CapEx spending by even one major hyperscaler — whether due to a change in AI strategy, recession, or alternative compute architecture adoption — would materially impact NVIDIA's near-term revenue trajectory.
Each new GPU architecture (Hopper → Blackwell → Rubin) involves elevated early-production costs that temporarily compress gross margins. Blackwell's ramp caused FY2026 gross margin to fall to 71.1% from FY2025's 75.0%. As Vera Rubin enters production (H2 calendar 2026), investors should expect a similar transitional gross margin dip before yields normalize.
Hyperscalers are developing proprietary AI accelerators: Google's TPU, Amazon's Trainium/Inferentia, Microsoft's Maia, and Meta's MTIA. If these ASICs achieve competitive performance-per-watt and economics, hyperscalers could reduce external GPU purchases. NVIDIA's CUDA software moat and full-stack co-design advantages (NVLink, networking, software) are the primary barriers to ASIC displacement.
NVIDIA's dominant position in AI training GPUs (~70–90% market share by revenue) has attracted scrutiny from regulators in multiple jurisdictions, including the U.S. DOJ, EU, France, and China. Investigations focus on CUDA lock-in, DGX/HGX bundling practices, and potential anticompetitive effects on alternative AI chip suppliers. The FY2026 10-K lists regulatory risk as a material risk factor.
NVIDIA's stock trades at a significant premium to revenue and earnings, reflecting high expectations for continued hypergrowth. Any sequential deceleration in Data Center revenue growth, guidance shortfall, or macro shock triggering hyperscaler CapEx cuts could result in significant multiple compression. The stock has historically been volatile around earnings.
NVIDIA is investing heavily in physical AI (robotics, autonomous vehicles, industrial AI) which contributed >$6 billion in FY2026. This market has a longer development cycle than hyperscale AI. Delays in autonomous vehicle certification (DRIVE Hyperion), industrial robot deployment, or changes in regulatory environments could cause physical AI revenue to miss expectations.
Competitive Positioning
NVIDIA dominates the AI training GPU market with an estimated 70–90% revenue share. The competitive moat rests on three pillars: (1) the CUDA software ecosystem — 20+ years of developer tooling, libraries (cuDNN, cuBLAS, NCCL, TensorRT), and framework integrations (PyTorch, JAX, TensorFlow) that are extremely difficult to replicate; (2) co-designed full-stack systems — GPU compute, NVLink switch fabric, InfiniBand/Spectrum-X networking, and software (NEMO, Triton, NIM microservices) extreme co-designed and optimized together; and (3) supply chain depth — years of advanced TSMC wafer reservations and partnerships with CoWoS HBM packaging suppliers.
AMD's MI300X is the closest alternative in the market, with competitive performance in some LLM inference benchmarks, but lags in multi-GPU scale-up (no equivalent to NVLink 72) and software maturity (ROCm vs CUDA). Intel's Gaudi platform has gained limited enterprise traction. Custom ASICs (Google TPU v5e, Amazon Trainium 2, Microsoft Maia 100, Meta MTIA) reduce hyperscaler external GPU dependency at the margin, but remain complementary to NVIDIA rather than displacement-level alternatives for frontier model training.
| Dimension | NVIDIA | AMD | Custom ASICs | Intel |
|---|---|---|---|---|
| GPU Compute Leadership | ✓ Blackwell/Rubin | MI300X (competitive) | Domain-specific | Gaudi (limited) |
| Multi-GPU Scale-Up Fabric | ✓ NVLink 72 (72-GPU) | Infinity Fabric (limited) | Custom interconnect | No equiv. |
| AI Networking | ✓ IB + Spectrum-X ($31B+) | No equiv. | Custom (some) | Omni-Path (niche) |
| Software Ecosystem | ✓ CUDA (20+ yrs) | ROCm (maturing) | Proprietary only | oneAPI (limited) |
| DC Revenue (FY2026) | $193.7B | ~$5–8B est. | Internal only | <$1B |
| Full-Stack AI Factory | ✓ DGX/HGX + NIM | Partial | Partial | No |
Verikal Assessment
NVIDIA delivered $215.9 billion in FY2026 revenue (+65% YoY), $120.1 billion in GAAP net income (+65% YoY), and $97 billion in free cash flow. These are the largest absolute revenue, profit, and FCF figures ever reported by any semiconductor company in a single fiscal year. Data Center revenue of $193.7 billion (+68% YoY) now represents 89.7% of total revenue — and at $193.7B, exceeds NVIDIA's entire FY2025 company-wide revenue of $130.5B. The Q1 FY2027 quarter (ended April 2026) set another record at $81.6 billion revenue (+85% YoY), with Q2 FY2027 guided to $91 billion.
The Blackwell architecture ramp is the most significant product cycle in NVIDIA's history. Grace Blackwell NVL72 systems represented approximately two-thirds of Data Center revenue in Q4 FY2026. Approximately 9 gigawatts of Blackwell infrastructure had been deployed as of February 2026. The Blackwell Ultra platform delivers up to 50x better performance and 35x lower cost per token versus Hopper, per SemiAnalysis InferenceX benchmarks. This performance-per-watt advantage is the key driver of hyperscaler infrastructure purchasing decisions, as every data center is power-constrained. Vera Rubin production shipments are on track for H2 calendar 2026, with an announced 10x reduction in inference token cost versus Blackwell.
The networking segment is a structurally underappreciated component of NVIDIA's business. Networking revenue exceeded $31 billion in FY2026 — greater than 10x FY2021 revenue in the year NVIDIA acquired Mellanox for ~$6.9 billion. In Q4 FY2026 alone, networking generated $11 billion (+3.5x YoY). In Q1 FY2027, networking reached $14.8 billion (+199% YoY). NVLink scale-up fabric, Spectrum-X Ethernet, and InfiniBand together form the interconnect layer that binds tens of thousands of GPUs into unified AI factories — a capability that is extremely difficult for competitors to replicate given the co-design requirements across compute and networking silicon.
The China export control regime and the competitive response from Chinese domestic AI chip companies represent the most visible medium-term structural risk. NVIDIA assumes zero Data Center compute revenue from China in its guidance. Sovereign AI ($30B+ in FY2026, >3x YoY) and the enterprise/national-infrastructure wave outside China are partially offsetting dynamics, but China's share of the global AI hardware TAM remains significant. The U.S. government's posture on export controls — whether tightened further or partially relaxed for specific products — is a major variable in NVIDIA's medium-term revenue trajectory. NVIDIA is actively engaging with both U.S. and Chinese governments and advocating for market access.
Sources & Methodology
All financial data sourced from SEC EDGAR: NVIDIA Corporation (CIK: 0001045810) Form 10-K Annual Report for fiscal year ended January 25, 2026 (FY2026), filed February 2026 (Acc-no: 0001045810-26-000021); Form 10-K Annual Report for fiscal year ended January 26, 2025 (FY2025); Condensed Consolidated Financial Statements in Q4 FY2026 Earnings Press Release (February 25, 2026) and Q1 FY2027 Earnings Press Release (May 20, 2026). Management quotes sourced from Q4 FY2026 Earnings Call Transcript (February 25, 2026, Corrected Transcript). FY2026 Networking revenue and Sovereign AI figures sourced from management commentary in Q4 FY2026 earnings call. Q1–Q3 FY2026 segment-level figures partially estimated from disclosed data. Derived FY2025 segment figures computed from disclosed YoY growth rates. No forward projections or price targets are provided.