No company in semiconductor history has lost its largest international market overnight and responded with an 85% revenue increase. Nvidia just did exactly that.
On May 20, Nvidia reported $81.6 billion in first-quarter revenue for fiscal year 2027, up 85% from the year-ago quarter. Data Center revenue hit $75.2 billion, up 92% year over year. Non-GAAP earnings per share landed at $1.87, beating the $1.76 Wall Street consensus. Gross margins held at approximately 75%.
Every number beat expectations. Revenue topped the $78.9 billion Street consensus by nearly $3 billion. The forward guidance made the beat look routine: Nvidia expects $91 billion in Q2 revenue (plus or minus 2%), clearing the $86.84 billion analyst estimate by more than $4 billion.
At that quarterly run rate, Nvidia is on pace for roughly $350 billion in annual revenue. Intel’s trailing twelve-month revenue sits at approximately $52 billion. Nvidia now generates more revenue in one quarter than Intel generates in eighteen months.
From 95% Market Share to Zero
The most telling number in the earnings call was not revenue. It was zero.
Nvidia assumed zero Data Center compute revenue from China in its Q2 guidance. CEO Jensen Huang confirmed on the call what the numbers had been signaling for three quarters: “We went from 95% market share to 0%, and I can’t imagine any policymaker thinking that’s a good idea… it has already largely backfired.”
The exit was progressive. Washington banned the A100, then the H100, then the H20 (a chip Nvidia designed specifically to comply with earlier restrictions). In September 2025, China’s Cyberspace Administration ordered domestic companies, including ByteDance and Alibaba, to stop purchasing Nvidia hardware entirely. Huawei’s Ascend 910C, roughly comparable to Nvidia’s 2020-era A100, became the default AI chip inside China, with 600,000 units planned for production in 2026.
Nvidia estimated the lost China market at roughly $50 billion annually. The company had been earning approximately $19.7 billion per year from China and Hong Kong before the restrictions took full effect. That revenue is now gone with no structural path to recovery.
And Nvidia grew 85% anyway.
The entire AI demand story outside China was strong enough to not only replace the lost revenue but accelerate past it. Hyperscalers in the U.S., Europe, Japan, and allied markets are spending at a pace that makes China’s absence a rounding error on the income statement. That is perhaps the most important signal in the entire earnings report: the non-China AI buildout is large enough to sustain record growth on its own.
$80 Billion in Buybacks and a 25x Dividend
Nvidia’s board approved an additional $80 billion in share repurchase authorization, joining $38.5 billion remaining from prior programs. Combined, that is roughly $118.5 billion in buyback capacity. The company also raised its quarterly dividend from $0.01 to $0.25 per share, a 25-fold increase. In Q1 alone, Nvidia returned approximately $20 billion to shareholders through buybacks and dividends.
Apple’s $110 billion buyback announced in May 2024 had been the largest single authorization in tech history. Nvidia’s cumulative authorization now exceeds Apple’s single-year figure, making it the second company ever to operate capital returns at this scale.
The numbers signal two things. First, Nvidia is generating cash faster than it can reinvest in R&D and capacity, even while ramping next-generation chip architectures on an accelerated timeline. Second, management believes the current stock price (around $220 per share, roughly $5.4 trillion in market capitalization) undervalues the business enough to make buybacks a better use of capital than holding reserves.
For investors tracking the AI infrastructure spending cycle, the $80 billion figure is arguably more important than the earnings beat itself. Revenue beats are backward-looking confirmations. A buyback of this size is a forward statement: Nvidia’s own board, with full visibility into the customer pipeline, sees enough demand durability to commit more than $100 billion to equity retirement.
$1 Trillion in Orders Through 2027
Huang revealed that Nvidia has secured $1 trillion in combined orders for its Blackwell and upcoming Vera Rubin architectures, spanning deliveries through the end of 2027. The Vera Rubin platform, which hit full production six months ahead of schedule, promises a 10x reduction in inference token cost compared to Blackwell. AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure are among the first to deploy Vera Rubin instances.
Huang also introduced Vera, Nvidia’s new central processor targeting what the company sizes as a $200 billion addressable market separate from GPUs. Nvidia projects $20 billion in Vera CPU revenue by fiscal year-end, representing a direct expansion into territory where Intel and AMD have historically dominated.
Supply constraints remain the central tension. Huang acknowledged that demand will outstrip supply throughout the entire Vera Rubin product lifecycle. Every major cloud customer is capacity-constrained today. The buildout driving Nvidia’s revenue is happening slower than buyers want, which means the demand curve still has room to steepen rather than flatten.
Why the Stock Barely Moved
Despite beating on every metric and raising guidance $4 billion above consensus, Nvidia shares dropped 1.6% in after-hours trading. They recovered most of the decline the following day but remained below their pre-earnings high of $227.
One analyst summarized the dynamic: “Nvidia delivered another beat, but at this point that’s essentially priced in.”
At $5.4 trillion in market capitalization, Nvidia is the world’s most valuable company. The stock trades at roughly 25 times forward revenue. The market has already embedded years of AI infrastructure spending into the price. Beating estimates by $3 billion on a $78 billion base barely registers when investors need proof that AI capex continues at this pace through 2028 and beyond.
The paradox of Nvidia’s current moment: the business has never been stronger and the stock’s response has never been more muted. Revenue is accelerating. Margins are stable. The product roadmap is stacked through 2027. But the market’s central question has shifted from “Is AI real?” to “How long does the buildout last?”
Three Signals for the Broader AI Industry
Nvidia’s Q1 results confirm three dynamics that extend well beyond the ticker.
AI infrastructure spending is not slowing down. The $91 billion Q2 guidance means the capex plans that hyperscalers announced at events like Google I/O (where Google confirmed $190 billion in annual AI spending) are converting into purchase orders. The checks are clearing, not just the slide decks.
The compute supply chain has permanently split in two. Nvidia is running a $350 billion annual revenue business with zero contribution from the world’s second-largest economy. Huawei is building a parallel chip ecosystem. The bifurcation is structural. Every enterprise buyer making infrastructure decisions today should plan around two separate AI hardware ecosystems going forward, because reunification is not on any realistic timeline.
Inference economics are about to shift by an order of magnitude. Nvidia’s expansion into CPUs (Vera) and its 10x inference cost reduction (Rubin) position the company to capture value across the entire AI compute stack, not just training. When inference costs drop tenfold, AI applications that are currently too expensive to deploy at scale become viable overnight. That expansion of what is economically feasible is what keeps Nvidia’s growth curve alive after the initial training clusters are built.
Running enterprise AI infrastructure, I watch procurement cycles closely. None of the infrastructure waves over the last 20+ years in IT operations (virtualization, cloud migration, containerization) produced a single vendor growing revenue 85% at the $80 billion quarterly scale. The closest analog is Cisco during the late-1990s internet buildout, and Cisco never hit these absolute dollar figures. Nvidia’s position is historically unprecedented. These earnings confirm the demand curve has not peaked.