Nvidia posts $96.2B quarter, guides $108B next

Nvidia reported fiscal second-quarter revenue of $96.2 billion, up 106% from a year ago and 18% from the prior quarter, and guided the current quarter to $108.0 billion, according to its official results release for the period ended July 26, 2026. Read the guide for what it is: management outlook, not a booked backlog.

The quarter itself is the easy part. Data Center revenue was $89.0 billion, up 117% year over year and about 92% of the total. GAAP and non-GAAP gross margin both landed at 75.0%. GAAP diluted EPS was $2.46, non-GAAP $2.22. Nvidia returned roughly $26.0 billion to shareholders through buybacks and dividends in the quarter, with about $99.0 billion left on its repurchase authorization, and declared a $0.25 dividend payable October 1, 2026.

The number the headline leaves out: zero China

The $108 billion guide carries an assumption most one-line takes drop. In its own words, "NVIDIA is not assuming any Data Center compute revenue from China in its outlook." So the sequential step up from $96.2B to $108B is built on selling nothing in one of the world's largest AI markets. If Washington and Beijing settle the H20-class export question, that is upside the guide does not contain; if the standoff hardens, the number already absorbs it. Either way, China is a swing factor sitting outside the printed figure, not inside it.

Where the dispute actually is

The bull case is on the page. CEO Jensen Huang told investors he expects roughly 70% revenue growth in fiscal 2028, well above Street models, and called Vera Rubin, now in full production, the platform "built to power exactly this moment."

The caution is on the same page. Nvidia guided Q3 gross margin down to 74.0% (+/- 50 bps) from 75.0%, and the pressure is memory cost, not pricing weakness. Demand is not the question; the question is whether input costs keep compressing margin faster than volume can offset it. A company growing revenue 18% a quarter while guiding margin lower is telling you the supply chain, specifically high-bandwidth memory, is now the binding constraint, not orders.

The takeaway

If you are modeling Nvidia from here, do not start with the $108 billion. Start with three inputs the release hands you: the guide assumes zero China Data Center compute, so any policy thaw is unmodeled upside; gross margin is guided to 74.0%, so watch memory costs as the margin lever, not ASPs; and roughly $99 billion of buyback authorization remains, which sets a floor under EPS regardless of how the top line moves. Track whether next quarter's margin holds that 74% line. If it slips further on memory, the growth story stays intact but the profit-per-dollar story changes.

For related context on how that Data Center demand is being financed and placed, see our coverage of Nvidia's $105B OpenAI Ohio backstop and its $6B Poolside investment tied to a $1B license.