NVIDIA reported revenue of $96.221 billion, up 106% year-over-year and 18% sequentially. That is about 5.7% above the company's own guidance midpoint of $91.0 billion and about 4.3% above the FactSet consensus of $92.27 billion. Data Center revenue was $89.023 billion, up 117% year-over-year. Non-GAAP diluted EPS of $2.22 came in about 6.2% above the $2.09 FactSet consensus, and GAAP diluted EPS was $2.46. For the current quarter, NVIDIA guided to revenue of $108.0 billion, plus or minus 2 percent, with GAAP and non-GAAP gross margin of 74.0 percent, plus or minus 50 basis points.
Separately, on the earnings call, CFO Colette Kress gave a preliminary expectation of roughly 70 percent revenue growth in fiscal 2028 and described it as a supply-constrained outlook. Third-party coverage put analyst expectations closer to 44 to 45 percent. She also walked investors through a lower near-term margin path, with a gross margin range of 71 to 72 percent expected in Q4 FY2027 before settling in a 72 to 73 percent range during fiscal 2028, citing memory costs. Those forward figures are management expectations rather than reported results. Demand was not the constraint this quarter. The cost of meeting it is the open question.
The bracket on the guided bar marks the company guidance range of $105.84B to $110.16B around the $108.0B midpoint. That midpoint is about 12.2% above reported Q2 FY2027 revenue.
The Q3 FY2027 band shows the guided range of 73.5% to 74.5%, with the tick at the 74.0% midpoint. The Q4 FY2027 and fiscal 2028 bands are preliminary expectations management described on the call, not company guidance and not reported results.
The beat was wide on both lines. Revenue of $96.221 billion cleared NVIDIA's own $91.0 billion guidance midpoint by about 5.7% and the $92.27 billion FactSet consensus by about 4.3%. Non-GAAP diluted EPS of $2.22 beat the $2.09 FactSet consensus by about 6.2%. Sequentially, revenue rose 18% from $81.615 billion in Q1 FY2027, and it more than doubled from $46.743 billion in the year-ago quarter.
The fiscal 2028 growth expectation was the most consequential number in the release window, and it did not come from the release. NVIDIA formally guides one quarter ahead, and its formal guidance here covers Q3 FY2027 only. On the earnings call, management gave a preliminary expectation of roughly 70 percent revenue growth in fiscal 2028 and described it as a supply-constrained outlook, meaning it reflects what the company expects to be able to supply. Third-party coverage of the call put contemporaneous analyst expectations near 44 to 45 percent. Management also said it expects supply to remain a bottleneck through at least the end of fiscal 2028.
Management reset the margin path in the same session. Reported GAAP and non-GAAP gross margin were both 75.0 percent, up from 74.9 percent GAAP in Q1 FY2027 and 72.4 percent in Q2 FY2026. Q3 FY2027 guidance is 74.0 percent on both measures, plus or minus 50 basis points. On the call, management said it expects gross margin to reach a range of 71 to 72 percent in Q4 FY2027 and then settle in a 72 to 73 percent range during fiscal 2028, pointing to memory costs that have risen beyond what the company had previously expected. Management tied the expected recovery partly to price increases planned to begin in Q1 FY2028.
Only the Q3 FY2027 revenue and gross margin figures are formal company guidance. The fiscal 2028 growth rate, the Q4 FY2027 margin trough, the fiscal 2028 margin range, and the planned price increases are preliminary expectations management described during the earnings call. They are not reported results and they are not commitments.
The tension in this quarter is unusual. Demand was not the constraint, and profitability is not collapsing. What changed is the cost of the inputs NVIDIA needs, memory in particular, which management said has risen faster than the company had planned for.
Taking management's own numbers at face value, a path from a reported 75.0 percent gross margin to an expected 71 to 72 percent in Q4 FY2027 is roughly 3 to 4 percentage points of compression. Against a Q3 guidance midpoint of $108.0 billion, one percentage point of gross margin is a little over $1 billion in a single quarter. It is worth keeping the level in view as well. Even at the expected trough, a 71 percent gross margin would remain far above what hardware businesses normally earn.
The line that changed most is future supply and capacity commitments, which rose from $119 billion to $279 billion, primarily because of memory procurement. NVIDIA disclosed the expected timing as approximately $92 billion during the remainder of fiscal 2027, $87 billion in fiscal 2028, $88 billion in fiscal 2029, $6 billion in fiscal 2030, $5 billion in fiscal 2031, and $1 billion thereafter. These are future commitments to secure supply and capacity, and the disclosure does not characterize every dollar as an unconditional obligation payable regardless of demand.
Read together, the two items point the same way. Securing memory ahead of the Vera Rubin ramp reduces the risk of running short of supply, and it raises the share of near-term cost that is locked in while memory prices are elevated. Whether that trade works out depends on where memory prices go from here, which nobody at the company claimed to know.
Data Center revenue of $89.023 billion rose 18 percent sequentially and 117 percent year-over-year, and split into Hyperscale at $48.710 billion, up 13 percent sequentially and 102 percent year-over-year, and AI Clouds, Industrial, and Enterprise at $40.313 billion, up 25 percent sequentially and 138 percent year-over-year. One caveat applies to that comparison. NVIDIA reclassified one company from AI Clouds, Industrial, and Enterprise into Hyperscale and recast the prior period figures, so the two segment growth rates are measured against restated history. Edge Computing revenue was $7.198 billion, up 13 percent sequentially and 27 percent year-over-year.
NVIDIA commenced production shipments of Vera Rubin in August 2026, and management said it expects the platform to become the fastest product ramp in the company's history. That expectation is management commentary, not a reported outcome.
On the customer side, AWS and NVIDIA announced that AWS plans to deploy 2 million additional NVIDIA GPUs. The joint announcement describes deployment during 2027 and 2028, while management described the schedule on the call as beginning in the current quarter and continuing through Q2 FY2029. The two companies are also working to bring Vera CPU based infrastructure to AWS.
China stayed out of the forward numbers. NVIDIA assumed no Data Center compute revenue from China in its Q3 FY2027 outlook. In Q2, shipments of Data Center Hopper products to China represented less than 1 percent of Data Center revenue.
The faster growth rate in AI Clouds, Industrial, and Enterprise than in Hyperscale is consistent with demand reaching further than the largest cloud buyers, and one quarter of restated segment data is thin evidence for a durable shift. Treat it as something to confirm over the next few quarters rather than as an established trend.
NVIDIA returned approximately $26.0 billion to shareholders during the quarter through share repurchases and cash dividends, with approximately $99.0 billion remaining under its repurchase authorization. The next quarterly cash dividend of $0.25 per share is payable on October 1, 2026, to shareholders of record on September 10, 2026. Operating cash flow was $24.077 billion and free cash flow was $21.341 billion. Cash, cash equivalents, and marketable debt securities stood at $56.6 billion.
Two working capital items moved. Accounts receivable was $63.1 billion, and days sales outstanding, the average time it takes to collect payment from customers, rose from 45 days to 60 days after NVIDIA extended payment terms for large multi-quarter agreements with certain investment-grade customers. Inventory rose from $25.8 billion to $31.6 billion as the company prepared for Vera Rubin.
Total future commitments across supply and capacity, cloud agreements, leases, equity investments, and capital expenditures were $366 billion. Separately, NVIDIA disclosed $56 billion of additional commitments related to AI cloud agreements and data center leases intended for third parties. Maximum gross guarantee exposure was $108.5 billion, including $105 billion connected with the SB Energy arrangement. That guarantee figure is a conditional maximum that would apply only if the guaranteed obligations were triggered. It is neither an immediate cash payment nor a current expense.
Management said NVIDIA has invested nearly $50 billion in frontier AI laboratories. Management also described financing platforms involving Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR that are intended to mobilize more than $500 billion of third-party capital over time, and which remain subject to definitive agreements.
Longer collection terms, direct investments in AI laboratories, and financing platforms aimed at third-party capital all point in one direction. NVIDIA is taking on more of a role in funding the buildout that buys its products. This is an inference about the shape of the business rather than a disclosed risk, and the disclosures do not quantify how much of NVIDIA's revenue depends on customers it also finances.
Formal company guidance. Q3 FY2027 revenue of $108.0 billion, plus or minus 2 percent, implies a range of $105.84 billion to $110.16 billion. The midpoint is about 12.2 percent above the $96.221 billion just reported, a slower sequential pace than the 18 percent delivered in Q2 but a larger step in absolute dollars. GAAP and non-GAAP gross margin are guided to 74.0 percent, plus or minus 50 basis points. The outlook assumes no Data Center compute revenue from China.
Preliminary management expectations from the call. Management expects revenue growth of roughly 70 percent in fiscal 2028, described as a supply-constrained outlook, and expects supply to remain a bottleneck through at least the end of fiscal 2028. On margins, it expects a range of 71 to 72 percent in Q4 FY2027 and then a 72 to 73 percent range during fiscal 2028, with the recovery connected partly to price increases planned to begin in Q1 FY2028. None of these are formal guidance, and all of them can change.
Both columns below are TheValueTrader's interpretation of the disclosed figures and management commentary, not statements of fact about what will happen.
This quarter settled the demand question for now. Revenue beat both the company's guidance midpoint and consensus by a wide margin, the year-over-year growth rate stayed above 100 percent at a scale where that is rare, and management was confident enough to break its usual one-quarter convention and describe a full year ahead at roughly 70 percent growth.
It did not settle the margin question. Management has now told investors its memory cost assumptions were too low, and the trough it described is an expectation rather than a level the company has already reached. The return to 72 to 73 percent in fiscal 2028 depends on price increases that have been planned but not yet implemented. That is the part of the story with the least evidence behind it so far.
Can NVIDIA push through the price increases it says it needs to restore margins without slowing the demand that is currently outrunning its own supply?
Every figure above traces back to one of the sources below. Reported results, formal guidance, earnings call commentary, and third-party consensus are listed separately.