Nvidia just posted the kind of numbers that make you stop and reread the sentence. In its fiscal first quarter of 2027, the company pulled in $81.6 billion in total revenue, an 85% jump from the same period a year ago and converted $58.3 billion of that into pure profit. Not operating income. Not EBITDA. Net income. Cash profit.
To put that in perspective: Nvidia’s single quarter profit is larger than the annual revenue of most Fortune 500 companies. And the company is telling Wall Street the next quarter will be even bigger.
The $81.6 Billion Quarter, Broken Down
The headline numbers are striking enough on their own, but the growth rates hiding underneath them tell a more complete story about what’s actually happening across the global AI industry.
| Metric | Q1 Amount | Year over Year Growth |
|---|---|---|
| Total Revenue | $81.6 Billion | +85% |
| Data Center Revenue | $75.2 Billion | +92% |
| Networking Revenue | $14.8 Billion | +199% |
| Net Income (Profit) | $58.3 Billion | +139% |
The standout line isn’t revenue, it’s networking revenue nearly tripling year over year. That $14.8 billion figure represents the physical “plumbing”, technologies like InfiniBand and NVLink interconnects required to link tens of thousands of AI chips together so they can function as one unified system. The fact that networking grew at roughly twice the pace of compute chips signals something important: AI infrastructure is scaling faster than the chips alone can explain.
Nvidia’s gross margin held firm at 74.9% meaning for nearly every dollar coming in, roughly 75 cents flows down to profit. At this scale, that’s a degree of pricing power rarely seen in the hardware industry.
From $68 Billion to $81 Billion in Just Three Months
One of the most striking aspects of this report isn’t the year over year comparison, it’s how much Nvidia grew in a single quarter.
Just three months ago, in Q4, Nvidia reported $68.1 billion in revenue. This quarter: $81.6 billion. That’s $13.5 billion in new top line revenue added in 90 days more than many large companies make in a full year.
The profit acceleration was even more dramatic. Net income jumped from $42.9 billion to $58.3 billion, a $15.4 billion increase in a single quarter. Profit grew faster than revenue because as Nvidia scales production on its Blackwell architecture, manufacturing efficiency improves. With demand still outpacing supply, the company doesn’t need to discount anything to move product, keeping those margins elevated.
The Data Center segment which accounts for 92% of all revenue grew from $62.3 billion to $75.2 billion quarter over quarter, a fresh $12.9 billion increase driven by cloud giants racing to lock in chip access.
What “Data Center Revenue” Actually Means
This is the part that confuses most people. When Nvidia reports $75.2 billion in Data Center revenue, they are not collecting rent on server space. Nvidia owns zero commercial data centers.
Instead, Nvidia operates as what CEO Jensen Huang calls the “arms dealer” of the AI revolution building the brains and the plumbing, then selling those components to the companies that own the physical real estate. That $75.2 billion breaks into two product lines:
- AI Compute Chips ($60.4 billion): The flagship hardware, Blackwell and Hopper GPUs, alongside the newly deployed Vera CPUs, that power large scale AI model training and inference.
- Data Center Networking ($14.8 billion): The interconnect chips and cables that link those GPUs together at the scale required for modern AI workloads.
Nvidia also restructured how it categorizes the rest of its business. What was previously reported as separate Gaming, Automotive, and Robotics segments is now consolidated into a single Edge Computing segment worth $6.4 billion, or roughly 8% of total revenue. It covers consumer graphics cards like the GeForce RTX 50-series, autonomous vehicle systems, and factory robotics.
The reshuffling makes the company’s identity impossible to misread: Nvidia is now, overwhelmingly, an enterprise AI infrastructure company that also happens to sell gaming GPUs on the side.
Who Actually Buys the Hardware
Nvidia sells to three distinct types of customers, and the fastest growing group isn’t the one most people think of.
The hyperscalers — Microsoft Azure, Google Cloud, Amazon Web Services, and Meta remain the largest single bloc, purchasing hundreds of thousands of chips to fill their warehouse scale data centers. They buy the hardware, rack it in their own facilities, and charge other companies to use it. This group still accounts for a slight majority of Data Center revenue.
But the growth is happening elsewhere.
Sovereign governments are now buying Nvidia hardware to build domestic AI clouds keeping sensitive national data localized rather than routing it through American cloud providers. Countries including India, Japan, Denmark, and Saudi Arabia have all made significant commitments to AI infrastructure that runs on Nvidia silicon.
Enterprise companies — names like Salesforce, ServiceNow, and Accenture are buying Blackwell systems to power AI agents that automate complex, multi-step business workflows. Jensen Huang describes this shift as the transition from basic AI models to “agentic AI” and notes that demand has gone “parabolic” as enterprises move from experimenting to deploying at scale.
For companies that want to own their own Nvidia AI supercomputers without building a warehouse, Nvidia partners with colocation providers like Equinix and Digital Realty. These firms build the ultra-dense, liquid-cooled facilities; Nvidia certifies them as “DGX-Ready,” confirming their power grids and cooling loops can handle the thermal demands of high-performance chip racks.
Nvidia also offers DGX Cloud — a rental model where developers pay for AI computing power directly from Nvidia. Even here, Nvidia doesn’t own the buildings. They place Nvidia-owned server racks inside facilities operated by Oracle, Google, and Microsoft, hosting the cloud service from within partner infrastructure.
The China Cliff and Air Force One
Perhaps the most geopolitically charged detail in the entire earnings report is buried in a single line: Nvidia’s high end Data Center shipments to China dropped to exactly zero last quarter.
One quarter prior, China represented $4.6 billion in data center compute revenue. This quarter, tightened U.S. export controls rendered even Nvidia’s custom “downgraded” chips illegal to ship into Beijing severing the pipeline entirely and creating an instant, $4.6 billion revenue cliff.
This context explains why Jensen Huang was a prominent member of President Donald Trump’s corporate delegation to Beijing for a high stakes 36-hour summit with President Xi Jinping alongside Tim Cook and Elon Musk.
Huang’s position at the negotiating table wasn’t ceremonial. He has been actively lobbying the Trump administration to approve export licenses for the H200 processor, or to establish a legally stable workaround, arguing that China represents a $50 billion market opportunity that Nvidia is currently forfeiting to domestic Chinese competitors like Huawei.
Trump has publicly backed the tech delegation, stating he is asking President Xi to “open up” China so American business leaders can operate there but the policy remains stuck in a tug of war between trade interests and national security hawks on both sides.
Huang isn’t waiting for a diplomatic breakthrough to hedge his bets. Coinciding with the Beijing trip, Nvidia made a multi-billion dollar prepayment to fund the construction of new Corning fiber-optic factories inside the United States. The move is effectively a supply chain insurance policy locking in a domestic fiber supply capable of supporting a tenfold expansion of U.S. AI infrastructure, regardless of how China policy evolves.
The math here is worth noting: Nvidia projected $91 billion in revenue for the upcoming quarter while already accounting for $0 in China data center revenue. If diplomatic efforts eventually restore any level of China access, that would be purely incremental upside on top of an already record-breaking trajectory.
$91 Billion Next Quarter and an $80 Billion Gift to Shareholders
Nvidia’s guidance for Q2 FY27 came in at $91.0 billion shattering Wall Street’s consensus estimate of $86.8 billion by more than $4 billion. The engine behind it is the continued full-throttle production ramp of the Blackwell architecture, which is replacing the older Hopper line across the board.
At the same time, Nvidia announced it would return significant capital to the investors who have funded this run. The company approved an $80 billion share buyback program and raised its quarterly dividend by 25 fold to $0.25 per share, a signal of confidence in the durability of its cash generation.
The broader picture Nvidia is painting for the market: the AI infrastructure buildout is not a short-term phenomenon. Tech giants, sovereign governments, and enterprise companies are committing capital at a scale and duration that suggests this cycle has years, not months, left to run. And as long as that remains true, Nvidia as the primary supplier of the hardware making it all possible sits at the center of the most significant infrastructure investment wave since the construction of the internet.
