NVIDIA Corporation

Total revenue was $96 billion, more than doubling year over year, and management said Q2 also set records for operating income and EPS, though specific EPS and operating income figures were not stated on the call. Data center revenue was $89 billion, up 18% sequentially; hyperscale revenue was $49 billion, up 13% sequentially; and ACIE revenue was $40 billion, up 25% sequentially and 138% year over year. GAAP and non-GAAP gross margin were both 75%; GAAP and non-GAAP operating expenses rose 10% and 11% sequentially; inventory increased to $32 billion; DSOs rose to 60 days; and NVIDIA returned a record $26 billion to shareholders, including $20 billion of buybacks and $6 billion of dividends. For Q3, revenue is guided to $108 billion, plus or minus 2%, with GAAP and non-GAAP gross margin guided to 74% plus or minus 50 basis points and operating expenses of about $9.2 billion and $9.0 billion, respectively. Fiscal 2028 revenue is expected to grow approximately 70% year over year, with supply expected to remain a bottleneck through at least the end of fiscal 2028; margin guidance was reset to 71% to 72% in Q4 before improving to 72% to 73% in fiscal 2028.

Jensen Huang framed the quarter as evidence that AI demand is broadening beyond hyperscalers into sovereigns, enterprises, NeoClouds, and frontier labs, with agentic AI increasing compute intensity. He emphasized NVIDIA’s advantage as a full-stack AI factory platform that can run every model across the AI lifecycle, and said the company is seeing acceleration in demand even at very large scale. His tone was highly confident and expansive, but he repeatedly stressed that growth is limited by supply, not demand.

Colette Kress said Q2 was an “outstanding quarter” with record revenue, operating income, and EPS, and she quantified revenue at $96 billion with 75% gross margin. She pointed to strong data center growth, AWS partnership expansion, and ACIE momentum, while noting inventory rose to $32 billion ahead of Vera Rubin and DSOs increased to 60 days because of extended payment terms for certain investment-grade customers. She also said NVIDIA returned $26 billion to shareholders in Q2, was at 60% of free cash flow returned year to date versus its 50%+ target, and now expects to increase excess free cash flow returns net of strategic uses. On margins, she warned that memory scarcity is worsening, Q3 gross margin is expected at 74% plus or minus 50 basis points, margins may bottom in the 71% to 72% range in Q4, and then recover to 72% to 73% in fiscal 2028 as price increases take effect.

Analysts focused on whether NVIDIA’s 70% fiscal 2028 growth guide implies demand far above supply, what the main bottlenecks are, and whether margin pressure from memory can be offset. Jensen said unconstrained demand would be “significant” or “a lot higher,” but the company can only confidently guide to what its supply chain supports; he said the entire supply chain is challenged, with land, power, shells, cooling, labor, memory, and manufacturing all requiring coordination. Questions also probed open-source versus closed models, competition from custom chips at frontier labs, and whether recursive self-improvement/AGI changes demand; management argued both open and closed models are growing on NVIDIA, custom chips do not replace the full-stack platform, and more agentic workflows only increase compute needs. Analysts also asked about the scale of ecosystem investments and financing commitments; Kress said the largest supply commitments are concentrated in the first three years and are essential to ramping Vera Rubin.