Nvidia's financials have undergone a transformation with few precedents in semiconductor history. The numbers tell a story the narrative alone cannot capture.
Nvidia grew revenue from $26.9 billion in FY2022 to $215.9 billion in FY2026, a 703% increase in four years. Free cash flow grew from $8.1 billion to $96.7 billion over the same period. These are not projections. They are reported annual figures.
The growth trajectory has few precedents in the history of large-cap equities. What matters now is not the historical narrative but the analytical questions it raises: what drove the numbers, how durable is the demand, what do the margins reveal about competitive position, and whether the $4.1 trillion market capitalization is pricing an accurate or optimistic version of the next five years.
For most of its history, Nvidia was primarily a gaming company. Its GeForce GPU line dominated the consumer graphics card market and generated steady, cyclical revenue. The data center business existed but was secondary: generating roughly $10-11 billion annually and viewed as a diversification play.
FY2023 was the last pre-inflection year. Revenue came in at $27.0 billion, essentially flat with FY2022's $26.9 billion. The gaming market had softened following pandemic-era demand, and the data center business had not yet accelerated. Operating income fell from $10.0 billion to $4.2 billion as gaming revenue collapsed faster than data center could offset.
The ChatGPT launch in late 2022 catalyzed a demand surge for AI training compute that Nvidia was uniquely positioned to capture. The A100 and H100 GPU families became the default infrastructure for large language model training. No competitor was close on software compatibility, ecosystem, or raw training throughput.
What followed was a revenue ramp that most analysts, including optimistic ones, failed to forecast correctly even six months ahead.
TickerXray Report
Get the complete Nvidia report with all 12 quantitative models, AI-generated investment thesis, and real-time data.
These are Nvidia's reported annual figures. Nvidia's fiscal year ends January 31.
FY2022 (ended Jan 2022): Revenue $26.9B, Gross Profit $17.5B, Operating Income $10.0B, Net Income $9.8B FY2023 (ended Jan 2023): Revenue $27.0B, Gross Profit $15.4B, Operating Income $4.2B, Net Income $4.4B FY2024 (ended Jan 2024): Revenue $60.9B, Gross Profit $44.3B, Operating Income $33.0B, Net Income $29.8B FY2025 (ended Jan 2025): Revenue $130.5B, Gross Profit $97.9B, Operating Income $81.5B, Net Income $72.9B FY2026 (ended Jan 2026): Revenue $215.9B, Gross Profit $153.5B, Operating Income $130.4B, Net Income $120.1B
Revenue doubled from FY2024 to FY2025, then grew another 65% from FY2025 to FY2026. Operating income grew from $4.2 billion to $130.4 billion in three years, a 3,000% increase. No large-cap company in recent history has posted comparable operating leverage.
The FY2023 trough is important context. Nvidia's profitability roughly halved in that year as gaming demand collapsed. The subsequent recovery was not a continuation of prior trends. It was a structural shift to an entirely different revenue base.
The data center segment became the dominant revenue driver beginning in FY2024. Hyperscalers, including Microsoft Azure, Google Cloud, Amazon Web Services, and Meta, collectively deployed tens of billions of dollars in Nvidia GPU clusters. Each major AI model training run required thousands of H100 and then H200 GPUs. Nvidia was the only supplier at scale.
The CUDA software ecosystem is the moat that explains why competitors could not immediately respond. Nvidia has spent over a decade building developer tools, libraries, and frameworks that AI researchers and engineers know how to use. Switching to AMD or Intel compute requires rewriting tooling that took years to build. The software lock-in is as significant as the hardware advantage.
Nvidia's customers are not just buying GPUs. They are buying into an ecosystem: CUDA, cuDNN, TensorRT, NeMo, Triton, and dozens of other tools that collectively make Nvidia hardware more productive. A competing chip that is 15% faster but lacks this ecosystem does not necessarily win enterprise contracts.
The Blackwell architecture, which began shipping in FY2026, extended the performance lead. Nvidia has demonstrated a consistent ability to ship each GPU generation ahead of meaningful competitive alternatives. The H100 benefited from an 18-24 month window without serious competition. Blackwell appears to be following a similar pattern.
Gross margin is the clearest indicator of pricing power and competitive position. For Nvidia, the data center transition produced a dramatic margin expansion.
FY2022 gross margin was 65.0%. FY2023 fell to 56.9% as gaming volumes dropped and data center had not yet scaled enough to offset. FY2024 recovered to 72.7% as data center at premium pricing became the majority of revenue. FY2025 reached 75.0%. FY2026 came in at 71.1%, a modest step back as Blackwell ramp costs were absorbed.
A 71-75% gross margin on $215.9 billion in revenue generates $153.5 billion in gross profit. That is more gross profit than almost any company in the world generates in total revenue. The margin profile reflects a company that sets prices rather than accepts them.
The slight compression from 75% to 71% in FY2026 is worth monitoring. It may reflect early Blackwell production costs normalizing as yield improves. It may also reflect the first signs of pricing pressure as alternative compute becomes more competitive. The direction over the next two fiscal years will reveal whether 75% was peak margin or a sustainable floor.
These margins are calculated from Nvidia's reported annual income statements.
FY2022: Gross Margin 65.0%, Operating Margin 37.3% FY2023: Gross Margin 56.9%, Operating Margin 15.6% FY2024: Gross Margin 72.7%, Operating Margin 54.1% FY2025: Gross Margin 75.0%, Operating Margin 62.4% FY2026: Gross Margin 71.1%, Operating Margin 60.4%
Operating margin of 60.4% on $215.9 billion in revenue has essentially no precedent among large-cap companies. For comparison, Apple's operating margin is approximately 30%. Google runs near 32%. Microsoft operates at 46%, itself considered exceptional.
Nvidia's operating margin compression from 62.4% to 60.4% in FY2026 is modest. But R&D spending also grew from $12.9 billion to $18.5 billion, a 43% increase. If Nvidia needs to invest heavily to maintain its technology lead, operating margins may face more pressure ahead. The current levels are extraordinary and the natural question is whether they are sustainable.
AMD is Nvidia's most credible GPU competitor. The MI300X and subsequent MI series have achieved meaningful adoption, particularly for AI inference workloads. Some hyperscalers have publicly committed to diversifying their compute supply away from exclusive Nvidia dependency. AMD is winning share, but from a much lower base.
Custom silicon is the other major competitive threat. Google's TPUs, Amazon's Trainium and Inferentia chips, and Meta's custom AI chips are designed to run specific workloads at lower cost than Nvidia GPUs. If hyperscalers succeed in shifting 30-40% of their AI compute to custom silicon, Nvidia's revenue growth rate slows materially.
Nvidia's response has been to accelerate its product roadmap. The shift from H100 to H200 to Blackwell happened faster than historical GPU cycles. The next generation, Rubin, is already in planning. Each generation widens the performance gap before competitors can close the prior one.
The CUDA ecosystem remains the deepest structural moat. Billions of lines of code run on CUDA. Hundreds of thousands of trained engineers know it. A faster chip that requires retraining the developer ecosystem faces a switching cost that money alone cannot solve quickly.
Nvidia generated $96.7 billion in free cash flow in FY2026, a figure larger than most companies' total revenue. Capital expenditure was only $6.0 billion, remarkably low for the scale of business. Nvidia is asset-light by design: TSMC manufactures the chips, and Nvidia designs the architecture.
The buyback program has been substantial. Nvidia repurchased $40.1 billion of its own shares in FY2026 and $33.7 billion in FY2025. Combined, that is $73.8 billion in buybacks over two fiscal years. Share count has declined from 24.99 billion in FY2022 to 24.43 billion in FY2026, a modest 2.2% reduction given the dollar amounts spent.
The relatively small share count reduction despite massive buyback spending reflects the stock price appreciation. Spending $73.8 billion at average prices of roughly $80-130 per share (pre-split adjusted) bought far fewer shares per dollar than buybacks at 2022 prices would have. SBC of $6.4 billion in FY2026 offset a portion of the buyback benefit.
The balance sheet shows $10.6 billion in cash against $7.5 billion in long-term debt at the end of FY2026. Total assets grew from $44 billion in FY2023 to $206.8 billion in FY2026. The asset growth reflects higher receivables and inventory rather than physical plant, consistent with the fabless model.
The law of large numbers is Nvidia's biggest challenge. Growing $215.9 billion in revenue by 65% requires generating roughly $141 billion in new annual revenue. That is roughly five times Amazon Web Services' entire annual revenue. The growth rate will decelerate because the base demands it.
Consensus analyst estimates for Nvidia's current fiscal year project EPS of $4.70, implying revenue in the range of $210-235 billion. That reflects continued data center demand but at a slower growth rate. The forward P/E of 21.5x on those estimates is reasonable if the estimates prove accurate.
The longer-term question is whether AI capex investment by hyperscalers and enterprises continues at current levels or normalizes. Microsoft, Google, Meta, and Amazon collectively plan to spend over $300 billion on AI infrastructure in calendar 2025. Much of that flows to Nvidia. If any of those companies slows their deployment or finds alternatives, the impact on Nvidia's revenue is immediate.
The PEG ratio of 0.71 suggests the market views current earnings as understating the future trajectory. A sub-1 PEG on a $4.1 trillion company implies the growth rate is expected to outpace the current P/E. That is a bold forward assumption, but it is not without data support given the FY2026 results.
Export controls are Nvidia's most immediate risk. U.S. government restrictions on exporting advanced AI chips to China have already reduced Nvidia's addressable market. China represented a significant revenue opportunity that is now partially or fully blocked depending on chip specification. Further tightening of export rules could expand the restriction.
Hyperscaler custom silicon is the medium-term competitive threat. If Google, Amazon, and Meta each succeed in running 30-40% of their AI workloads on proprietary chips, Nvidia loses hundreds of billions in potential revenue over a five-year horizon. The question is execution and timeline. Custom silicon programs have historically taken longer and underdelivered on performance relative to Nvidia.
A broader AI investment slowdown is the macro risk. If enterprise AI deployments generate disappointing ROI, corporations may reduce AI compute spending. The current capex boom in hyperscaler infrastructure is predicated on AI revenue materializing at scale. If that revenue is slower than expected, the capex impulse could reverse.
Finally, TSMC concentration risk is material. Nvidia's entire product line is manufactured by TSMC. Any disruption to TSMC's advanced process nodes, whether from geopolitical tension, natural disaster, or yield problems, flows directly into Nvidia's ability to ship product.
The five-year data is unambiguous. Nvidia executed a complete business transformation from gaming GPU supplier to the dominant infrastructure provider for the AI supercycle. Revenue grew 703%. Free cash flow grew 1,091%. Operating margin reached 60%. The numbers are genuine and remarkable.
At $4.1 trillion in market cap and a 34x trailing P/E, investors are now paying for the next five years rather than the last five. The forward P/E of 21.5x is reasonable if consensus estimates prove accurate. But consensus has consistently underestimated Nvidia for three years running, and the next few quarters will test whether that pattern continues or normalizes.
The bear case is not that Nvidia is a bad business. It is that at $4.1 trillion, the market has priced in a version of the future where the AI infrastructure investment cycle continues uninterrupted, competition remains limited, and margins hold near 60-65%. If any of those assumptions proves optimistic, the downside is meaningful despite the exceptional underlying business.
Full forensic analysis of Nvidia
+ 6 more models included
150,000+ stocks covered
Global coverage across 60+ exchanges. Every report includes all 12 quantitative models and AI analysis.
View plansEvery report runs 12 quantitative models and generates an AI investment thesis. From Piotroski scores to manipulation detection -- get the full picture in seconds.
12 forensic models
Piotroski, Altman, Beneish, DuPont & more
AI investment thesis
Synthesized outlook on every stock
Manipulation detection
Spot red flags before they hit the news
150,000+ tickers
Global coverage across 60+ exchanges
Expected return
Forward return projections for every stock
Real-time data
Live prices, insider trades, news sentiment
Free accounts get 1 report per month. Pro gets unlimited.
Our earlier 'Five Trillion' narrative argued Nvidia was being underpriced for the AI capex cycle. Since publication, free cash flow has reached $96.7 billion, the multiple has held, and Wall Street's biggest AI bull just got louder. We update the thesis.
A $5.2 trillion market cap with revenue still compounding above 70% year over year is a profile the public market has never seen at this scale. The signals are telling a specific story.
NVDA's market cap has tripled in two years on FCF that grew from $3.8B to $96.7B. Most of the bear arguments are about the multiple, not the cash. Here's what the consensus is pricing wrong.