NVIDIA reported fiscal second-quarter revenue of $96.2 billion, more than doubling from a year ago and exceeding the $91.9 billion consensus estimate by $4.3 billion. Adjusted earnings per share came in at $2.22, above the $2.08 forecast. Data center revenue rose 18% sequentially to $89 billion, accounting for 92.7% of total sales.
Gross margin held steady at 75%, while return on equity reached 114%. The company returned $26 billion to shareholders in the quarter, including $20 billion in buybacks and $6 billion in dividends. Operating expenses increased 10% on a GAAP basis and 11% on a non-GAAP basis from the prior quarter. Inventory stood at $32 billion, positioned for the Vera Rubin platform launch.
Data center growth was led by hyperscale revenue of $49 billion, up 13% sequentially, and the ACIE segment—covering sovereign AI, regional cloud providers, enterprises, and NeoClouds—at $40 billion, up 25% sequentially and 138% year-over-year. Automotive on-prem revenue reached $8 billion over the trailing 12 months, while financial services, manufacturing, and healthcare contributed $7 billion. Grace CPU revenue exceeded $5 billion in the same period.
Global venture capital funding in AI surpassed $400 billion in the first half of 2026, with roughly 70% directed toward compute infrastructure, exceeding the $265 billion raised in all of 2025.
Shares rose 3.98% in after-hours trading to $218, following a regular session close at $209.66. The stock has traded between $164.07 and $236.54 over the past 52 weeks.
For the fiscal third quarter, NVIDIA guided revenue to $108 billion, plus or minus 2%, with gross margin expected at 74%, plus or minus 50 basis points. Operating expenses are projected at $9.2 billion on a GAAP basis and $9.0 billion on a non-GAAP basis. The company also projected fiscal 2028 revenue growth of about 70%, with CPU revenue expected to more than double.
Vera Rubin production shipments began in August, with the platform expected to account for roughly 20% of data center revenue in the current quarter. NVIDIA estimates Vera Rubin delivers 30 times higher throughput per megawatt and 35 times lower token costs than Grace Blackwell Ultra, expanding the total addressable market to $40 billion per gigawatt, compared with $18 billion for Hopper and $25 billion for Blackwell. NeoCloud capacity is projected to reach 8 gigawatts by year-end, up from about 3 gigawatts at the end of 2025.
AWS will deploy an additional 2 million GPUs through the second quarter of fiscal 2029, alongside Vera CPUs, and will adopt NVIDIA’s physical AI stack for warehouse robotics. Oracle Cloud Infrastructure and SpaceX AI are lead partners for Vera CPU, while Samsung Electronics is using NVIDIA’s cuLitho for computational lithography. Bristol Myers Squibb is investing in a Vera Rubin AI factory, and quantitative trading firms Hudson River Trading and Jane Street are leveraging NVIDIA-powered AI factories.
Top hyperscalers’ capital expenditures are projected to reach nearly $800 billion in 2026 and $1.3 trillion in 2027, with cloud industry backlog exceeding $2 trillion.
