For Windows users, PC buyers, IT departments and policymakers, the results matter beyond Wall Street. Nvidia’s data-center spending cycle reaches into workstation availability, enterprise AI deployment, cloud pricing and the physical infrastructure communities must accommodate. The company’s report is strong evidence of demand for accelerated computing. It is not, by itself, proof that every AI deployment will produce profitable customer outcomes or that the industry can expand without supply, trade and local-infrastructure constraints.
A quarter defined by Data Center revenue
Nvidia reported GAAP net income of $59.688 billion and GAAP diluted earnings per share of $2.46. Its non-GAAP diluted EPS was $2.22. Operating expenses rose 55% year over year to $8.408 billion, demonstrating that Nvidia is spending substantially more to support its growth, even as revenue grew faster.
The dominant story is in Data Center. Revenue there rose 117% year over year to $89.023 billion. Nvidia broke that figure into two disclosed markets:
- Hyperscale: $48.710 billion
- AI Clouds, Industrial and Enterprise: $40.313 billion
That breakdown is useful because it prevents an overly narrow reading of the AI buildout. The company is not describing demand solely from a small group of giant consumer-internet platforms. Its reporting also places material volume in AI cloud, industrial and enterprise-oriented deployments. Still, these categories are market disclosures, not proof that every organization inside them is earning an attractive return on its AI investment.
Nvidia management has argued that AI demand is being driven by workloads whose output is commercially useful. That is an important part of its strategic case, but it remains a management assertion. The reported numbers establish that customers are purchasing infrastructure at an enormous scale; they do not independently settle whether AI tokens, models or agents are broadly profitable across customer types.
Revenue momentum remains remarkable, but the growth rate would slow
Nvidia forecast third-quarter fiscal 2027 revenue of $108.0 billion, plus or minus 2%. At the midpoint, that would be roughly 89% above the $57.006 billion reported in the comparable quarter a year earlier.
An approximately 89% year-over-year increase would be exceptional for a company of Nvidia’s size. But it would also be a deceleration from the 106% year-over-year revenue growth reported in the second quarter. The distinction matters.
There are two very different claims one could make about this outlook:
- Nvidia expects another quarter of exceptionally rapid expansion.
- Nvidia’s year-over-year growth rate is accelerating.
Only the first is supported by the figures. A lower percentage growth rate does not mean demand has disappeared, particularly when the revenue base has become so large. It does mean readers should avoid equating a higher revenue target with an accelerating growth rate.
The outlook also has a significant geographic assumption: Nvidia said it is not assuming any Data Center compute revenue from China in its third-quarter forecast. That makes the guide less dependent on a market affected by trade and export restrictions, but it does not eliminate China-related risk.
In the second quarter, shipments of Data Center Hopper products to China represented less than 1% of Data Center revenue. Nvidia also disclosed that licensed H200 shipments faced a 25% U.S. import tariff that it had been unable to pass on to customers. Low reported Hopper revenue from China limits the current contribution of that business, while the tariff disclosure shows how policy can still affect product economics and pricing flexibility.
Earnings quality: operating demand is real, but net income has another component
The revenue and Data Center figures leave little doubt that Nvidia’s operating business is generating immense sales. However, treating the full $59.688 billion GAAP net-income figure as a pure measure of AI-chip sales profitability would be incomplete.
Nvidia reported $7.773 billion of other income, primarily gains from equity securities, during the quarter. That is a material contributor to the bottom-line total. The company’s core operations were still highly profitable, but the distinction is essential for analysts and investors trying to understand repeatability.
Revenue from products and services generally reflects customer purchasing activity. Gains on equity securities can move with market valuations and may not recur in the same form or magnitude. A strong quarter can contain both powerful operating execution and a favorable investment-income contribution; acknowledging the latter does not diminish the former.
For technology buyers, the practical lesson is similar. A supplier’s headline profit is not the same thing as the cost-effectiveness of a deployment. Enterprise customers assessing GPUs, networking and AI software should focus on total cost of ownership, utilization, energy requirements, application performance and their own measurable business outcomes rather than assuming market enthusiasm guarantees a positive project return.
Supply is both Nvidia’s advantage and its constraint
Management’s longer-range comments were as consequential as the quarterly results. Nvidia preliminarily expects fiscal 2028 revenue to grow by approximately 70% year over year. It also characterized that outlook as supply-constrained and warned that supply could remain a bottleneck through the end of fiscal 2028.
This framing changes how the forecast should be interpreted. It is not simply a demand forecast. It is an estimate shaped by what Nvidia believes it can deliver through its manufacturing and supply-chain ecosystem.
Chief Executive Jensen Huang described demand as above supply and said the company’s supply chain had been challenged. That statement is evidence of management’s internal view, not an independently measured total of unfilled orders. Readers should therefore resist translating it directly into a precise amount of “hidden” revenue waiting to be recognized.
Nevertheless, supply limits can have concrete effects:
- Cloud providers and enterprises may need to wait longer for infrastructure.
- Hardware availability can influence which AI projects reach production first.
- Buyers may diversify designs, optimize existing accelerators more aggressively or rent capacity from cloud providers instead of purchasing systems.
- Nvidia’s future revenue may be limited by what it can manufacture and assemble, not only by customer willingness to spend.
Supply constraints are not automatically positive. They can support demand visibility and pricing discipline, but they also leave room for competitors, alternative architectures and customer-developed silicon. The dossier does not establish how much share Nvidia could lose or retain under those conditions, so a definitive competitive conclusion would be premature.
AWS expands the infrastructure commitment, with important limits
Nvidia and AWS announced plans to deploy 2 million additional Nvidia GPUs across AWS global infrastructure in 2027 and 2028. The companies also described deeper work in AI factories, CPUs, networking, open models, data processing and robotics.
This is a major forward-looking infrastructure commitment, but its boundaries matter. The announcement pairs the GPU plan with work on physical AI, including warehouse robots. It does not say that the 2 million GPUs are dedicated to warehouse robotics. Those GPUs are planned across AWS’s global infrastructure and could support a much wider set of AI uses.
For Windows-focused organizations, the AWS commitment reinforces an important trend: many AI capabilities will be consumed as cloud services rather than deployed entirely inside a corporate server room. That can reduce the need for an organization to buy and maintain its own large GPU clusters. It can also increase dependence on cloud capacity, cloud-region availability and recurring usage costs.
Nvidia also said it preliminarily expects CPU revenue to more than double in fiscal 2028, as it seeks to become a leading server-CPU supplier. That is a forecast rather than a reported result. Still, it signals that Nvidia’s ambition extends beyond standalone accelerators toward more of the server platform: CPUs, GPUs, networking and systems designed to operate together.
The strategic implication is that enterprise infrastructure choices may become more integrated. Such integration can simplify deployment and improve performance for some workloads, while potentially making it harder to substitute individual components later. IT leaders should assess portability and interoperability alongside raw benchmark performance.
Windows PCs and workstations: a split demand picture
Nvidia reported $7.198 billion in Edge Computing revenue, up 27% year over year and 13% sequentially. The company attributed the increase to strong Blackwell workstation sales, partly offset by slower consumer-PC sales amid elevated memory and system prices.
This is the most direct connection in the report to the Windows hardware ecosystem. Workstation demand appears to be benefiting from professional and AI-oriented use cases, while the consumer PC market faces cost pressure. Elevated memory and system prices can matter to buyers even if GPU performance keeps improving: the practical purchase decision is about the price of a complete system, not one component.
The filing does not isolate Windows-specific unit sales, pricing or market share. It would therefore be too strong to claim that Nvidia’s results prove a broad Windows PC boom. The evidence supports a narrower conclusion: professional Blackwell workstation demand was a stated source of Edge Computing growth, and consumer PC demand was described as slower in an environment of higher memory and system prices.
For prospective buyers, that suggests a familiar split. Users who need local AI inference, 3D creation, engineering software or accelerated professional workloads may find workstation-class hardware increasingly relevant. Mainstream consumers should weigh whether local AI features justify paying for higher-priced systems when cloud-based services may cover many everyday tasks.
Data-center growth has a public-infrastructure limit
The economic narrative surrounding AI often focuses on chips, models and cloud capacity. Local communities increasingly focus on electricity, water, land use, construction disruption and jobs. Independent reporting documents growing opposition to data-center projects across political lines, with concerns about power demand, water supplies, local effects and potential AI-related job displacement. Some labor groups, meanwhile, see opportunities in construction and ongoing operations.
This does not establish that local opposition will reduce Nvidia’s revenue. The connection between a specific permitting dispute and Nvidia’s future sales is too indirect to state as fact. But it does identify a real constraint on the broader infrastructure buildout that Nvidia and its customers depend on.
For public officials, the challenge is not well captured by a simple choice between welcoming or rejecting data centers. Communities may need clearer standards for grid upgrades, water use, project transparency, tax arrangements, backup generation, local employment commitments and accountability for promised benefits. For residents, the key question is whether a project’s infrastructure costs and benefits are shared fairly.
For technology users, these debates can eventually affect cloud-region expansion, capacity availability and the pace at which AI services appear in particular markets. AI infrastructure is physical infrastructure as well as software.
What to watch next
Nvidia’s quarter provides firm evidence of an extraordinary current demand cycle, particularly in Data Center. The third-quarter outlook, AWS deployment plan and fiscal 2028 management forecast indicate that the company expects the buildout to continue. But each is subject to a different kind of uncertainty.
The next quarter will test whether Nvidia reaches its $108.0 billion revenue outlook without assumed Data Center compute revenue from China. The following fiscal year will test whether supply can expand enough to support management’s approximate 70% growth expectation. And the wider industry will have to demonstrate whether massive infrastructure spending translates into durable, customer-level economic returns.
The most defensible reading is neither that AI demand is a passing illusion nor that the path ahead is frictionless. Nvidia has reported a business expanding at a historic pace. It has also acknowledged supply pressure, China-related trade complications, a consumer-PC softness offset and the limits of what a revenue forecast can prove about end-user value. Those details will matter as much as the headline number for anyone planning a Windows workstation purchase, an enterprise AI rollout or an infrastructure policy response.