I still remember the first time I watched a single company reshape an entire industry overnight. That feeling came rushing back when the latest numbers landed. NVIDIA just reported fiscal second-quarter revenue of $96.2 billion. Let that sink in for a moment. The figure more than doubled from a year earlier and cleared what most analysts had quietly hoped for. In my view, this is no ordinary earnings print. It is proof that artificial intelligence infrastructure has moved from experimental budgets into the core of global computing spend.
Why These Numbers Matter More Than The Headlines Suggest
Revenue climbed 18 percent from the prior quarter and 106 percent year over year. Visible Alpha consensus had sat near $92.2 billion, so the beat was clean. Yet the real story lives deeper inside the segments. Data Center revenue alone reached $89.0 billion. That is an 18 percent sequential jump and a 117 percent surge from the same period last year. Analysts had looked for roughly $85.7 billion. The gap between expectation and reality keeps widening.
Non-GAAP diluted earnings hit $2.22 per share against a $2.09 consensus. GAAP earnings came in even stronger at $2.46 per share. Net income under GAAP more than doubled to $59.7 billion. Gross margins held at 75 percent on both GAAP and non-GAAP bases. Those are the kind of figures that make portfolio managers sit up straighter.
I’ve found that markets often focus on the top-line number and miss the quieter signals. This quarter the quieter signals are loud. NVIDIA returned about $26 billion to shareholders through buybacks and dividends. Roughly $99 billion remains authorized for future repurchases. That combination of growth and capital return is rare at this scale.
Data Center Dominance Continues To Accelerate
The Data Center business now accounts for the overwhelming majority of total revenue. Demand for accelerated computing has not cooled. Hyperscalers, cloud providers, and an expanding set of enterprise buyers continue to place large orders. In my experience, once a technology crosses from early adoption into must-have infrastructure, the spending curve rarely flattens quickly.
Customer forecasts point to growth doubling next year. Management itself is more measured. CFO Colette Kress noted that the company expects approximately 70 percent growth for fiscal 2028 because available supply may not satisfy every forecast. That gap between indicated demand and deliverable capacity is both a challenge and a vote of confidence.
Customer forecasts point to NVIDIA’s growth doubling next year, but the company expects approximately 70 percent growth because available supply may not satisfy all forecast demand.
The 70 percent figure is management’s internal planning assumption rather than a guarantee. Still, even that more conservative outlook would represent extraordinary expansion for a company already operating at this revenue level.
The China Factor And The $108 Billion Outlook
Looking ahead, NVIDIA guided fiscal third-quarter revenue to $108 billion, plus or minus 2 percent. That sits above the roughly $104.2 billion consensus that had been circulating. Gross margins are expected to settle at 74 percent, plus or minus 50 basis points. The guidance explicitly excludes any Data Center compute sales into China. Ongoing restrictions and policy uncertainty have removed that revenue stream from the model.
Shares moved around after the release as investors digested the slightly lower margin outlook. Later in extended trading the stock climbed about 4.1 percent following the earnings call. Markets tend to punish uncertainty, yet they also reward companies that clear the bar while acknowledging real constraints.
Perhaps the most interesting aspect is how cleanly the company has isolated the China impact. By stating the assumption outright, management removed one major source of guessing from the model. Investors can now focus on execution in the rest of the world.
Supply Commitments Jump To $279 Billion
One of the more striking details in the filing is the sharp rise in supply and capacity commitments. The total moved from $119 billion in the prior quarter to $279 billion as of July 26. The increase stems mainly from memory components and manufacturing capacity needed for current and future products.
Of that $279 billion, $92 billion is due during the remainder of fiscal 2027. Another $87 billion falls in fiscal 2028 and $88 billion in fiscal 2029. Some of these agreements can still be canceled, rescheduled, or adjusted before firm orders are placed. Even so, the sheer size signals confidence that demand will remain elevated for years.
Total future commitments across supply, cloud services, leases, equity investments, and capital expenditure reached $366 billion. Separate from the core supply figure, the company also holds $29 billion in cloud service agreements and $25 billion in data center leases that have not yet commenced. These are long-duration obligations that lock in capacity well into the future.
I’ve watched semiconductor cycles for years. Rarely does a company expand its purchase commitments this aggressively unless it sees sustained multi-year demand. Memory has become a particular bottleneck. Securing large volumes early is a competitive necessity rather than an optional hedge.
Vera Rubin Moves Into Full Production
Product execution continues in parallel with the financial results. The Vera Rubin platform has entered full production. Systems are already running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius. That list of early customers spans pure-play AI cloud specialists and the largest hyperscalers.
New architectures usually take time to ramp. Seeing multiple major platforms adopt the platform so quickly suggests both technical readiness and strong customer pull. In my view, the ability to move a new platform from design into revenue-generating deployments remains one of NVIDIA’s quiet strengths.
Partnerships Aimed At Mobilizing Capital At Scale
Beyond silicon, the company announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR. The stated goal is to mobilize more than $500 billion for AI infrastructure. These arrangements remain subject to definitive agreements, so the figure is an ambition rather than secured capital. Still, the lineup of financial institutions involved is notable.
Building the physical layer of AI requires enormous amounts of capital for power, cooling, land, and specialized facilities. Chip makers alone cannot fund every data center. Bringing large asset managers into the conversation creates a broader financing ecosystem. Whether the full $500 billion materializes will depend on many variables, yet the intent is clear.
Crypto Miners Join The AI Infrastructure Buildout
An interesting side current is the behavior of public Bitcoin mining companies. Nine of them spent $5.11 billion on capital assets during the first half of 2026 while recording $341.2 million in AI and high-performance computing revenue. Many of these firms already control power contracts and data center real estate. Converting that footprint toward GPU cloud services is a natural evolution.
One example that stands out is a $3.4 billion contract covering managed GPU cloud services over five years. These deals show that the AI demand wave is reaching players who previously focused almost exclusively on cryptocurrency mining. The infrastructure crossover is real and accelerating.
Perhaps this is the part that feels most under-discussed. Capital that was once tied to hash-rate competition is now flowing into accelerated computing. That shift expands the total addressable market for GPU suppliers and creates new channels for capacity deployment.
Margins, Memory Costs, And The Next Test
Gross margins held at 75 percent this quarter but are guided slightly lower next quarter. Memory pricing and manufacturing costs are rising. Protecting profitability while scaling output is the central operational challenge. Higher input costs are not temporary noise. They reflect genuine tightness across the supply chain.
The company must convert its expanded supply commitments into delivered systems without letting margins erode too far. Investors will watch the third-quarter delivery numbers closely. Hitting the $108 billion target without any China compute contribution would reinforce the strength of demand elsewhere.
In my experience, the companies that navigate this phase best are those that treat capacity as a strategic asset rather than a pure cost center. Locking in long-term supply at known prices reduces future volatility even if it pressures near-term margins.
What The Broader Market Is Signaling
The reaction in after-hours trading was constructive. A 4.1 percent rise after an initial wobble suggests that many investors ultimately focused on the demand picture rather than the modest margin step-down. Growth of this magnitude rarely comes without some cost pressure.
I’ve noticed that the market is increasingly willing to accept slightly lower margins if the absolute dollar growth remains exceptional. At this scale, even a 70 percent annual growth rate would add tens of billions of dollars in incremental revenue. That arithmetic is hard to ignore.
At the same time, the exclusion of China revenue removes a source of policy risk from the near-term model. Clarity has value. Companies that can quantify the impact of restrictions and still post record results tend to earn greater confidence over time.
Looking Further Ahead
Fiscal 2028 planning assumes roughly 70 percent growth. Customer forecasts sit higher. The difference between the two numbers is essentially a statement about supply constraints. If manufacturing capacity expands faster than currently modeled, upside remains possible. If memory and advanced packaging remain tight, the 70 percent figure may prove realistic.
Either outcome leaves NVIDIA operating at a scale few technology companies have ever reached. The combination of product leadership, customer lock-in, and aggressive capacity commitments creates a powerful feedback loop. Demand funds more supply, which in turn enables more demand.
The Vera Rubin ramp and the large infrastructure financing discussions add further layers. Hardware cycles used to move more slowly. Today the time from announcement to production volume has compressed dramatically. That acceleration benefits the company that can execute across design, manufacturing partnerships, and software ecosystems simultaneously.
Key Takeaways From The Quarter
- Total revenue of $96.2 billion beat consensus and more than doubled year over year
- Data Center revenue of $89.0 billion rose 117 percent from the prior year
- Third-quarter guidance of $108 billion excludes China compute sales
- Supply and capacity commitments expanded to $279 billion
- Vera Rubin platform has entered full production across multiple major clouds
- Approximately $26 billion returned to shareholders during the quarter
- Management expects around 70 percent growth in fiscal 2028 due to supply constraints
These points form the core of the story. Everything else is context and color.
A Personal Reflection On The Pace Of Change
I’ve covered technology cycles long enough to recognize when something feels different. The current AI infrastructure buildout has that quality. Spending is no longer concentrated among a handful of early experimenters. It has become a broad industrial priority. Power contracts, land acquisitions, specialized cooling, and long-term chip supply are all being negotiated at scale.
NVIDIA sits at the center of that activity. The latest results simply quantify what many operators already feel on the ground. Demand remains stronger than the industry can fully satisfy. That imbalance is uncomfortable for customers who want immediate capacity, yet it is highly supportive for the supplier that can deliver reliable performance.
The next few quarters will test whether the company can keep converting commitments into revenue while defending margins against rising memory and manufacturing costs. Execution will matter more than narrative. So far the track record remains strong.
One final observation. The involvement of traditional asset managers in funding AI infrastructure suggests that the capital markets themselves are adapting. Silicon alone is no longer sufficient. The physical plant required to run large models has become a distinct asset class. That evolution may prove as consequential as any single product launch.
The numbers are extraordinary. The outlook is ambitious yet grounded. The supply commitments are large. The product pipeline continues to advance. For anyone following the intersection of computing and capital allocation, this quarter offers a clear snapshot of where the industry stands today and where the pressure points will sit tomorrow.
Growth of this magnitude never arrives without friction. Memory costs, packaging capacity, power availability, and geopolitical constraints all remain live issues. NVIDIA has chosen to address them openly while still posting results that reset expectations higher. That combination of transparency and performance is rare. It is also the reason these results will be studied for some time.
In the end, the story is straightforward. Artificial intelligence continues to drive unprecedented demand for specialized computing. One company currently supplies the majority of that demand. The latest quarter simply makes the scale of that reality impossible to ignore.
Whether the next leg of growth lands closer to customer forecasts or to management’s more cautious 70 percent planning number, the trajectory remains steep. Capacity is the binding constraint. Everything else is secondary. That is the clearest message from $96.2 billion in a single quarter.
Investors, operators, and policymakers will all draw their own conclusions. For now the data is unambiguous. Demand for AI infrastructure is still expanding faster than most models predicted even a year ago. The companies positioned to meet that demand are rewriting the size of their own opportunity sets in real time.
And that, more than any single headline number, is what makes this quarter stand out.
