NVIDIA Q3 FY2026 Earnings: Inside the $57B Quarter Powering the AI Revolution
On November 20, 2025, NVIDIA Corporation reported third quarter fiscal 2026 results that obliterated every remaining doubt about the sustainability of AI compute demand: $57.0 billion in revenue, up 63% year-over-year and 22% sequentially, driven by a Data Center segment that alone generated $51.2 billion -- more than the entire company's revenue just two quarters earlier. This is no longer a cyclical semiconductor business riding a temporary wave. This is the toll booth operator for the most important technology buildout since the internet, and the traffic is accelerating.
The Headline Numbers
The Q3 print was emphatic. Revenue of $57.0 billion beat consensus estimates. Net income of $31.9 billion -- more than many Fortune 500 companies earn in an entire year -- rose 65% year-over-year. Operating income of $36.0 billion delivered a 63.2% operating margin, expanding nearly 100 basis points year-over-year despite massive R&D investment in next-generation architectures.
| Metric | Q3 FY2026 | Q3 FY2025 | YoY Change |
|---|---|---|---|
| Total Revenue | $57,006M | $35,082M | +62.5% |
| Cost of Revenue | $15,157M | $8,926M | +69.8% |
| Gross Profit | $41,849M | $26,156M | +60.0% |
| Gross Margin | 73.4% | 74.5% | -1.1pp |
| Operating Income | $36,010M | $21,869M | +64.7% |
| Operating Margin | 63.2% | 62.3% | +0.9pp |
| Net Income | $31,910M | $19,309M | +65.3% |
The slight gross margin compression from 74.5% to 73.4% reflects the early-stage Blackwell ramp, where new product introduction costs temporarily weigh on mix before yields and volumes mature. Management guided Q4 gross margin to improve to the mid-70s, consistent with the typical trajectory of new GPU architecture launches.
Revenue Deep Dive: Data Center Dominance Intensifies
NVIDIA reports five revenue segments, and the story is unambiguous: Data Center is the franchise, and everything else is a sidecar. We tracked every dollar through double-entry accounting in Beancount. Here is the quarterly trajectory for FY2026:
| Segment | Q1 FY2026 | Q2 FY2026 | Q3 FY2026 | Q/Q Growth | YoY Growth |
|---|---|---|---|---|---|
| Data Center | $39,112M | $41,096M | $51,215M | +24.6% | +66.5% |
| Gaming | $3,763M | $4,287M | $4,265M | -0.5% | +30.1% |
| Professional Visualization | $509M | $601M | $760M | +26.5% | +56.4% |
| Automotive | $567M | $586M | $592M | +1.0% | +31.8% |
| OEM & Other | $111M | $173M | $174M | +0.6% | +79.4% |
| Total | $44,062M | $46,743M | $57,006M | +22.0% | +62.5% |
Several things jump out.
Data Center ($51.2B, +66.5% YoY): This single segment now represents 89.9% of total revenue, up from 87.7% in Q3 FY2025. The sequential acceleration from $41.1B to $51.2B -- a $10.1 billion jump in a single quarter -- signals that the Blackwell architecture ramp is hitting its stride. Hyperscalers (Microsoft, Google, Amazon, Meta) are all pulling forward orders as they race to build AI training and inference infrastructure. The demand is not speculative; these are production deployments generating real revenue for cloud providers.
Jensen Huang's commentary on the call was characteristically direct: demand for Blackwell is "staggering" and "different from anything we have ever seen." NVIDIA shipped its first Blackwell systems in the quarter, and the architecture is already generating multi-billion dollar revenue.
Gaming ($4.3B, +30.1% YoY): A solid result that reflects the GeForce RTX 50-series positioning and continued strength in notebook GPU sales. The flat sequential result (-0.5%) is typical for the July-October quarter before holiday demand. Gaming remains the reliable base business -- generating more revenue per quarter than many semiconductor companies produce in a year -- but it is increasingly overshadowed by Data Center's trajectory.
Professional Visualization ($760M, +56.4% YoY): The strongest growth rate outside Data Center, driven by enterprise adoption of NVIDIA Omniverse and RTX-powered workstations for AI-augmented design workflows. This segment is small in absolute terms but strategically important as it establishes NVIDIA's ecosystem in enterprise creative and simulation workloads.
Automotive ($592M, +31.8% YoY): DRIVE Orin and successor platforms continue ramping across Chinese and Western OEMs. The $14 billion automotive design win pipeline provides multi-year revenue visibility as autonomous driving features expand from premium to mass-market vehicles.
The Margin Machine
NVIDIA's margin profile tells the story of a company with unassailable pricing power in the most critical technology market in the world:
| Metric | FY2024 | FY2025 | FY2026 YTD (Q1-Q3) |
|---|---|---|---|
| Revenue | $60,922M | $130,497M | $147,811M |
| Cost of Revenue | $16,621M | $32,638M | $45,441M |
| Gross Margin | 72.7% | 75.0% | 69.3% |
| R&D | $8,674M | $12,914M | $12,985M |
| SG&A | $2,655M | $3,491M | $3,297M |
| Operating Income | $32,972M | $81,454M | $86,088M |
| Operating Margin | 54.1% | 62.4% | 58.2% |
The FY2026 YTD gross margin of 69.3% looks compressed versus FY2025's 75.0%, but this is almost entirely attributable to the $4.5 billion H20 inventory charge in Q1 FY2026. Strip out that one-time charge -- which was related to U.S. export restrictions on China-destined chips -- and the normalized YTD gross margin is approximately 72.3%, tracking closely with the prior year.
The operating margin trajectory is remarkable. NVIDIA went from 54.1% operating margins in FY2024 to 62.4% in FY2025, and Q3 FY2026 alone posted 63.2%. This is the highest sustained operating margin in the semiconductor industry by a wide margin. For context, the next closest peers (Broadcom, Qualcomm) operate in the 30-40% range. NVIDIA's pricing power -- driven by the fact that its GPUs have no functional equivalent for large-scale AI training -- creates a margin structure more typical of software than hardware.
R&D spending of $13.0 billion through Q3 FY2026 is already outpacing FY2025's full-year $12.9 billion, reflecting aggressive investment in next-generation Rubin architecture, CUDA software stack enhancements, and the NVIDIA NIM inference platform. Yet R&D as a percentage of revenue has actually declined from 14.2% in FY2024 to 8.8% in Q3 FY2026 -- a dramatic illustration of the operating leverage inherent in NVIDIA's platform model.
The $4.5 Billion H20 Charge: A One-Time Hit With Strategic Implications
Q1 FY2026 included a $4.5 billion charge within cost of revenue related to inventory write-downs and purchase obligation losses for H20 GPUs, the China-specific product designed to comply with U.S. export controls. In October 2025, the U.S. government imposed new licensing requirements that effectively prohibited shipment of H20 chips without prior authorization.
The charge was significant -- it pushed Q1 gross margin down to 60.5% and depressed Q1 net income to $18.8 billion versus $22.1 billion in the prior quarter. But the market correctly treated it as non-recurring. By Q2, gross margin recovered to 72.4%, and Q3's 73.4% confirmed the trajectory back toward the mid-70s.
The strategic implication is more nuanced. NVIDIA's China business, which at its peak contributed an estimated 20-25% of Data Center revenue, is now structurally impaired. But the impact has been more than offset by surging demand from U.S. hyperscalers and sovereign AI programs globally. The loss of the China market, paradoxically, may have strengthened NVIDIA's bargaining position with its largest customers, who now face even tighter supply constraints.
Three-Year Trajectory: From $61B to $148B in Nine Months
The full scope of NVIDIA's transformation becomes apparent only when viewed across multiple years. Using the complete ledger data:
| Metric | FY2024 (Full Year) | FY2025 (Full Year) | FY2026 (Q1-Q3 Only) |
|---|---|---|---|
| Revenue | $60,922M | $130,497M | $147,811M |
| Data Center | $47,525M | $115,186M | $131,423M |
| Net Income | $29,759M | $72,880M | $77,107M |
| Gross Margin | 72.7% | 75.0% | 69.3%* |
| Operating Margin | 54.1% | 62.4% | 58.2%* |
* FY2026 YTD includes $4.5B Q1 H20 one-time charge
Revenue has more than doubled from FY2024 to FY2025, and FY2026 has already surpassed FY2025's full-year total through just three quarters. Data Center revenue through Q3 FY2026 ($131.4B) has exceeded the full FY2025 Data Center result ($115.2B) with an entire quarter remaining. If Q4 FY2026 matches the Q3 run rate, NVIDIA will close the fiscal year with approximately $200 billion in revenue -- a 53% increase over FY2025 and more than triple FY2024.
Net income of $77.1 billion through nine months already exceeds FY2025's full-year $72.9 billion. Even backing out the Q1 H20 charge, the earnings power of this business at current scale is staggering.
What the Other Income Line Tells You
One line item that warrants attention is "Other Income, net," which includes gains from NVIDIA's strategic equity investments. In Q2 FY2026, this line contributed $2.2 billion, and in Q3, another $1.4 billion. These gains -- primarily from investments in AI ecosystem companies -- are real but not necessarily recurring. Combined with $1.7 billion in interest income through three quarters (earned on the company's growing cash and investment balances), the non-operating income line has contributed approximately $3.5 billion to nine-month net income.
Investors should note that Q1's Other Income line was actually a $180 million loss, making the non-operating income highly variable quarter to quarter. The core operating business, however, has been remarkably consistent in its upward trajectory.
Tracking a $148B Revenue Machine in Plain Text
To truly internalize the financial architecture of NVIDIA's transformation, we built the entire FY2024-FY2026 income statement in Beancount, the open-source double-entry accounting system, using plain-text files that anyone can audit and verify.
Here is what the Q3 FY2026 revenue recognition looks like in the ledger:
2025-10-26 * "Q3 FY2026 Revenue" "Data Center"
Assets:Current:Accounts-Receivable 51,215,000,000.00 USD
Income:Data-Center -51,215,000,000.00 USD
2025-10-26 * "Q3 FY2026 Revenue" "Gaming"
Assets:Current:Accounts-Receivable 4,265,000,000.00 USD
Income:Gaming -4,265,000,000.00 USD
2025-10-26 * "Q3 FY2026 Revenue" "Automotive"
Assets:Current:Accounts-Receivable 592,000,000.00 USD
Income:Automotive -592,000,000.00 USD
Every revenue segment, every cost line, every tax provision -- modeled as double-entry transactions that balance to the penny. Year-end closing entries for FY2024 and FY2025 zero all income and expense accounts into retained earnings, providing a complete audit trail from quarterly press releases to annual financial statements. Explore the full NVIDIA ledger interactively below:
The entire three-year history -- FY2024 through FY2026 Q3 -- fits in roughly 800 lines of plain text. When the numbers in this analysis cite "$51.2 billion in Data Center revenue" or "$72.9 billion in FY2025 net income," those figures are pulled directly from the ledger's transactions and closing entries, cross-referenced against SEC filings and earnings press releases.
The Verdict: Bull vs. Bear
The Bull Case:
- Data Center revenue of $51.2B in a single quarter -- growing 66% YoY and 25% sequentially -- proves AI infrastructure demand is accelerating, not plateauing
- The Blackwell architecture ramp is ahead of schedule, with management guiding gross margins back to mid-70s as yields improve and volumes scale
- FY2026 revenue through Q3 ($148B) already exceeds FY2025's full-year total ($130.5B), with Q4 likely to push the annual figure toward $200B
- Operating margins above 63% in Q3 demonstrate pricing power that has no parallel in the semiconductor industry
- The $14B automotive design win pipeline and expanding enterprise AI inference market provide long-duration growth visibility beyond the current training buildout cycle
The Bear Case:
- Gross margin compression from 75.0% (FY2025) to 73.4% (Q3 FY2026) during the Blackwell ramp raises questions about long-term margin sustainability as product transitions become more frequent
- The $4.5B H20 charge is a reminder that geopolitical risk -- particularly U.S.-China export controls -- can create material one-time hits with little warning
- Customer concentration risk is extreme: the top four hyperscalers (Microsoft, Google, Amazon, Meta) likely represent 40-50%+ of Data Center revenue
- AI infrastructure spending by hyperscalers is funded by their own massive capex budgets ($175B+ each for 2026), and any pullback in those budgets would directly impact NVIDIA
- Competition from AMD's MI300 series, Intel's Gaudi accelerators, and custom ASICs (Google TPUs, Amazon Trainium) will intensify, though NVIDIA's CUDA ecosystem moat remains formidable
Our Take: NVIDIA is the defining technology franchise of this decade. A company that generates $31.9 billion in net income in a single quarter while growing revenue 63% year-over-year is not overearning -- it is capitalizing on a structural shift in how the world's largest companies allocate compute spending. The near-term Blackwell margin compression is transitory. The H20 charge was a one-time event. The customer concentration is a feature of serving the companies building the infrastructure layer of the AI era. At its current trajectory, NVIDIA will close FY2026 with approximately $200 billion in revenue and $100 billion in net income -- numbers that, even at a premium multiple, suggest the stock's risk-reward remains compelling for investors willing to look past quarterly noise and focus on the decade-long AI infrastructure buildout ahead.
