In the spring of 2023, a few months after ChatGPT launched, Nvidia was making $7.19 billion per quarter. Today it's $81.6 billion. More than eleven times as much, in just three years.
No company has profited from the AI boom as directly as Nvidia. For this article, I worked through the last three years of quarterly reports, the current annual report, and the most important market data, and compiled the most interesting numbers, data, and facts. From the revenue explosion and the market cap records to the risks that get far less attention.
- Nvidia generated $81.6 billion in revenue in the February-April 2026 quarter (up 85% year-over-year). In the fiscal year ending January 2026, it recorded $215.9 billion in revenue and $120.1 billion in net income
- With a record high of $5.5 trillion in market cap (May 2026), Nvidia is the most valuable company in the world. It was the first company ever to crack the $4 trillion mark, in July 2025
- 92% of revenue now comes from the data center business, and just two customers account for 36% of total revenue. The flip side of AI dominance
1. Company Overview & Milestones
Nvidia was founded on April 5, 1993, by Jensen Huang, Chris Malachowsky, and Curtis Priem. Legend has it the idea was born in a Denny's diner in California. More than 30 years later, Huang still runs the company, making him the longest-serving CEO among the big tech corporations.
1.1. The Key Figures
Three numbers show the scale Nvidia now operates at:
The most important facts at a glance:
1.2. Milestones From 1993 to 2026
From graphics card maker for gamers to the backbone of the AI industry. The most important stages:
In my view, the most important milestone isn't the trillion-dollar valuation. It's CUDA. Since 2006, Nvidia has been building the software platform that lets GPUs handle general-purpose computing. By the time the deep learning boom arrived, the entire ecosystem was already locked into Nvidia. That moat is still the main reason competitors have such a hard time catching up.
2. Revenue Growth
Nvidia's revenue curve since 2023 may be the steepest growth story any company of this size has ever pulled off.
2.1. Quarterly Revenue Since 2023
The numbers speak for themselves. Within three years, quarterly revenue has grown more than elevenfold:
What strikes me most is the jump from April 2023 ($7.19 billion) to July 2023 ($13.51 billion). Nearly a doubling in a single quarter. That was the moment the ChatGPT effect first fully showed up in Nvidia's books.
And growth is accelerating again:
From February through April 2026, Nvidia generated $81.6 billion, up 85% year-over-year and 20% quarter-over-quarter. That single quarter brought in more revenue than the entire fiscal year from February 2023 through January 2024 ($60.9 billion).
2.2. Fiscal Year 2026: $215.9 Billion
In fiscal year 2026 (February 2025 through January 2026), Nvidia posted $215.9 billion in revenue, up 65% from the prior year. Of that, $120.1 billion remained as net income (GAAP).
Let that number sink in for a moment:
Over $120 billion in profit in a single year. More than the total revenue of most Fortune 500 companies. And yet the stock market valuation still looks ambitious, more on that later.
3. Revenue by Segment: From Gaming to AI Powerhouse
If you still think of Nvidia as a graphics card company for gamers, you're dramatically underestimating how much it has changed. The segment numbers make that unmistakable.
3.1. The Data Center Dominates Everything
In the November 2025 through January 2026 quarter, the last one reported under the old segment structure, revenue broke down like this:
$62.3 billion of the $68.1 billion total came from the data center business, meaning AI chips and the infrastructure around them. That's 91.5%. Gaming, once the core business, contributed just $3.7 billion, with the rest spread across workstations, automotive, and OEM business.
3.2. The New Segment Structure Since 2026
Starting with the February through April 2026 quarter, Nvidia reports only two segments: data center and edge computing. Gaming, workstations, and automotive all disappear into the edge catch-all. A company can hardly make its core business any clearer.
The details are worth a closer look:
Of the $75.2 billion in data center revenue, $60.4 billion came from compute (the actual AI chips) and $14.8 billion from networking. Nvidia's networking division alone, largely the legacy of the 2020 Mellanox acquisition, is bigger than AMD's entire data center business.
3.3. The Transformation in One Picture
How quickly Nvidia's business model has flipped shows up best in a direct comparison of revenue shares:
In the spring of 2023, the data center accounted for 59.5% of revenue. Less than three years later, it's 91.5%. Today, Nvidia is effectively a pure AI infrastructure company with a gaming department attached.
4. Profitability: Margins Others Can Only Dream Of
Growth is one thing. What makes Nvidia truly exceptional is the combination of growth and profitability.
The GAAP gross margin most recently stood at 74.9% and has hovered between 73 and 78% for two years. Of every dollar in revenue, roughly 75 cents remain before research, sales, and administration are paid. For a hardware company, that's an absurdly good figure; classic chipmakers often sit at 40 to 60%.
The downward outlier in the spring of 2025 (60.5%) has a very specific cause:
US export controls on the China-focused H20 chip forced Nvidia to take a $4.5 billion write-down on unsellable inventory. Without that one-off effect, the margin would have topped 71% then as well. More on the China business in section 7.
Net income makes the scale even clearer:
In the February through April 2026 quarter, Nvidia earned $58.3 billion (GAAP), up 211% year-over-year.
5. Market Cap: The Most Valuable Company in the World
On the stock market, Nvidia has pulverized every historical benchmark over the past three years.
5.1. From $1 Trillion to $5.5 Trillion in Three Years
It took Nvidia a little over twelve months to get from the $1 trillion to the $3 trillion mark. From there to the record high of $5.5 trillion took less than two years.
5.2. The Most Valuable Companies Compared
After a first brief stint at the top in January 2025, which the DeepSeek sell-off ended within days, Nvidia has been the world's most valuable company without interruption since June 3, 2025. Here's what the leading group currently looks like:
Apple pulled clearly into second place ahead of Alphabet after a rally in July 2026. The gap between Nvidia and Microsoft, which practically had the top spot on subscription for years, is now a good $2 trillion. Nvidia alone is worth nearly as much on the stock market as Microsoft and Amazon combined.
6. Market Share: Dominance in Two Markets
Nvidia's market position plays out on two levels: gaming graphics cards and AI accelerators for data centers.
6.1. Gaming GPUs: 90% Market Share
In discrete graphics cards (standalone GPUs, not integrated into the processor), the market is practically decided:
Roughly 90% of all discrete graphics cards shipped in the first quarter of 2026 came from Nvidia (about 11.8 million units shipped in total). AMD holds around 8%, while Intel barely registers at about 1%. The values fluctuate somewhat from quarter to quarter, but the overall picture doesn't change.
6.2. AI Accelerators: Estimates Between 80 and 85%
For the far more important market of data center AI chips, there is no official statistic, only analyst estimates. The commonly cited figures put Nvidia at roughly 80 to 85% revenue share and AMD at about 5 to 7%; for Google's in-house TPUs and Amazon's Trainium, no reliable market share figures have been published.
These values come from market analyses without disclosed methodology, though, so I'd treat them as rough guideposts only. That said, the hard numbers from the financial reports confirm the overall picture, as the competitive comparison in section 10 shows.
7. China & Export Controls: The $4.5 Billion Problem
Few topics have shaken up Nvidia's numbers over the past year and a half as much as US export controls targeting China.
The chronology:
The outcome of this policy is remarkable:
China's share of Nvidia's revenue fell to around 9% in the fiscal year ending January 2026. The H20 licenses granted in August 2025 produced just about $50 million in revenue per CFO Colette Kress (November 2025), a rounding error in Nvidia's books. For all practical purposes, Nvidia has written off the Chinese AI chip market, and it's still growing 85% a year.
8. Products: Hopper, Blackwell, Rubin
Since 2022, Nvidia has shipped a new chip architecture every year. The roadmap:
The Rubin architecture, unveiled in March 2026, shows where things are headed: 336 billion transistors (1.6 times as many as Blackwell), 288 GB of HBM4 memory, and 22 TB/s of memory bandwidth, nearly triple Blackwell's. The first Vera Rubin rack is already running at Microsoft Azure.
Blackwell nevertheless remains the revenue driver in 2026. Analysts expect the architecture to account for roughly 70% of Nvidia's high-end GPU shipments. The reported backlog of 3.6 million B200/GB200 units was sold out through mid-2026 as of late 2025.
9. Customers & Dependencies
This is where it gets most interesting, in my view. Nvidia's greatest strength is also its greatest risk.
9.1. Two Customers, 36% of Revenue
According to the annual report, two direct customers accounted for 36% of Nvidia's total revenue in the fiscal year ending January 2026. In the quarter ending October 2025, Nvidia even disclosed four direct customers, each with more than a 10% revenue share:
The four largest direct customers thus combined for 61% of revenue (as of fall 2025). Nvidia doesn't name names, but it's an open secret in the industry that the big hyperscalers and AI labs are behind them.
9.2. The Hyperscalers Are Investing Like Never Before
How long can this go on? The 2026 investment plans of the four largest cloud companies offer an answer:
Per their current guidance (as of April 2026), the four companies plan around $725 billion combined for 2026, 77% more than the roughly $410 billion in 2025. A substantial share of that flows directly or indirectly into Nvidia hardware. As long as this capex cycle keeps running, so does Nvidia's growth. For what this buildout means for power grids and emissions, see my AI energy statistics.
10. Competition: AMD Is Catching Up, but the Gap Remains Huge
AMD is the most important direct competitor in the data center business and is growing impressively fast itself. Just from a completely different starting point:
AMD's entire data center segment (server CPUs plus AI accelerators) generated $5.8 billion in the quarter ending March 2026, up 57%. Nvidia's data center division brought in $75.2 billion over nearly the same period. That's 13 times as much.
The more interesting competition may come from Nvidia's own customers:
Google (TPU) and Amazon (Trainium) are developing their own AI chips to reduce their dependency on Nvidia. Both use their chips primarily in-house and don't publish revenue figures. The effect remains limited for now; the ecosystem around CUDA and Nvidia's sheer delivery speed keep it in the lead.
11. Nvidia as an AI Investor
Nvidia doesn't just sell to the AI industry anymore, it finances it too. The three largest stakes:
The $30 billion for OpenAI (finalized in February 2026) is already the slimmed-down version. A letter of intent for up to $100 billion had been on the table since September 2025 before shrinking to the $30 billion in early 2026. The $10 billion for Anthropic was announced in November 2025, alongside a $5 billion investment from Microsoft.
Jensen Huang himself hinted in March 2026 that the OpenAI stake was "possibly the last" of this magnitude, and per media reports he extended the remark to the Anthropic stake as well. His reasoning makes sense. Once OpenAI and Anthropic go public, these early-stage opportunities close.
Critics, however, see a structural problem in these deals:
Nvidia invests billions in companies that spend much of that money on Nvidia hardware. This circular dynamic inflates the entire industry's revenue growth without fresh money flowing in from outside. For the full picture of the AI industry's investment web, see my article on AI statistics.
12. Team & Leadership
Jensen Huang has led Nvidia since its founding in 1993. No other big tech company is still run by a founder from the 90s.
At the end of its fiscal year in January 2026, Nvidia employed roughly 42,000 full-time staff across the US and 37 other countries, about 6,000 more than a year earlier. For a company this size, that's astonishingly few:
Per employee, Nvidia generated a good $5.1 million in revenue and around $2.9 million in net income in fiscal year 2026. Microsoft, Google, and Apple each employ several times as many people. Nvidia's business model (design the chips, outsource manufacturing to TSMC) scales almost without adding headcount.
13. Conclusion: Impressive, but Not Without Risks
The raw numbers are historically unprecedented. $81.6 billion in quarterly revenue, a 75% gross margin, a market cap of around $5 trillion, 90% market share in gaming GPUs, and an estimated 80%+ in AI accelerators. Nvidia is the most profitable growth company tech history has ever seen.
Still, there are three risks you should keep in mind when putting these numbers in context:
First, customer concentration. Two customers account for 36% of annual revenue, and four most recently made up 61% of a single quarter's revenue. If even one of the big hyperscalers cuts its AI budget, it hits immediately. Second, the circular investment structures, where Nvidia's own capital finances part of the demand for Nvidia chips. And third, geopolitics: the China business is already largely lost, and further export controls remain a sword of Damocles.
For the AI industry as a whole, Nvidia remains the most important barometer. As long as the quarterly numbers look like they did recently, the oft-proclaimed end of the AI boom is nowhere to be found in the data. For how the models themselves are evolving, check out my LLM statistics.
