NVIDIA Hits $96.2B Revenue as Vera Rubin Ships

NVIDIA reported revenue of $96.2 billion for the second quarter of fiscal 2027, ended July 26, 2026, up 18% sequentially and 106% year over year. GAAP and non-GAAP gross margins reached 75.0%, while net income hit $59.7 billion, yielding a GAAP diluted EPS of $2.46 (up 128%) and a non-GAAP diluted EPS of $2.22 (up 120%). The Data Center segment generated $89.0 billion, up 117% annually, and Edge Computing brought in $7.2 billion, up 27%. R&D spending rose to $7.1 billion from $4.3 billion. For the third quarter, NVIDIA expects $108.0 billion in revenue, plus or minus 2%, with gross margins of 74.0% (plus or minus 50 basis points), assuming zero China data center compute revenue. Operating expenses are projected at $9.2 billion GAAP and $9.0 billion non-GAAP.

The company generated $21.3 billion in free cash flow, returning $26.0 billion to shareholders via buybacks and dividends, with $99.0 billion remaining under its repurchase authorization. A dividend of $0.25 per share will be paid on October 1, 2026, to shareholders of record on September 10, 2026. NVIDIA's balance sheet grew to $320.3 billion in total assets, aided by a $24.9 billion debt issuance and $42.4 billion deployed into equity securities. To fund future growth, NVIDIA partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR on August 10, 2026, to mobilize over $500 billion in third-party capital.

NVIDIA also secured infrastructure, partnering with SB Energy on August 17, 2026, at the PORTS-Pike Technology Campus, and buying a minority stake in Cloverleaf Infrastructure on August 21, 2026. SpaceXAI will deploy Vera CPUs, Lancium will build gigawatt-scale factories, and Vera Rubin racks are now running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius. The edge business added the Jetson Orin Nano 2 and partnered with Microsoft on RTX Spark.

For developers, the launch of the Groq 3 LPX interactive inference accelerator, licensed from Groq, Inc. for $2.9 billion, is a major milestone. In Artificial Analysis benchmarks, it achieved 3,400 output tokens per second running Gemma 4 31B with a 100,000-token context. The hardware pairs the accelerator with BlueField-4 DPUs, Vera CPU racks, Vera BlueField-4 STX storage, and Spectrum-6 SPX Ethernet. Additionally, Blackwell led in MLPerf Training 6.0 and AgentPerf benchmarks.