The world's largest company beat the Street for the 19th time in 20 quarters, printing a Q2 guide of US$91 billion, and announced a 25-fold dividend increase - with China Data Centre revenue excluded from the forward outlook entirely.
NVIDIA (NASDAQ: NVDA) reported revenue of $81.6 billion for the three months ended 26 April 2026 - up 85% year-on-year and 20% from the prior quarter - clearing consensus estimates of $79.2 billion.
Non-GAAP diluted earnings per share came in at $1.87 against a Street consensus of $1.77, extending the chipmaker's unbroken run of exceeding Wall Street forecasts to four straight quarters, with the topline topping estimates by just over 3%.
Data centre pulls further ahead
The segment posted $75.2 billion for the period - up 92% year-on-year and 21% sequentially - while the networking sub-line reached $14.8 billion, up 199% from a year ago.
Data Centre compute revenue came in at a record $60.4 billion, up 77% year-on-year and 18% sequentially, driven by Blackwell architecture absorption across hyperscale and enterprise deployments.
"We should be growing faster than hyperscale capex," NVIDIA founder and CEO Jensen Huang told analysts on the earnings call, putting combined hyperscaler spend at roughly $1 trillion this year.
And he's not far off. So far, declared hyperscaler capital commitments for 2026 stood at:
- Meta: full-year capex guidance of $125 billion to $145 billion
- Microsoft: calendar-year capex of $190 billion
- Alphabet: full-year capex guidance of $180 billion to $190 billion
The four largest cloud operators combined are on a roughly $700 billion spend trajectory for the year, directed primarily at accelerated compute and purpose-built AI data centres.
NVIDIA guided Q2 revenue of $91 billion (±2%), with GAAP and non-GAAP gross margins expected at 74.9% and 75%, respectively.
Shareholders pocket windfall
The chipmaker returned approximately $20 billion to holders during the quarter through buybacks and cash dividends, with $38.5 billion remaining under the prior repurchase authorisation at quarter's end.
On 18 May, the Board approved an additional $80 billion repurchase authorisation with no expiry date, alongside a dividend increase from $0.01 per share to $0.25 per share, payable 26 June 2026 to shareholders of record on 4 June.
Free cash flow for the quarter was $48.5 billion, up from $34.9 billion in Q4 FY26 and $26.1 billion in the prior corresponding period.
China stays off the ledger
NVIDIA's Q2 guidance explicitly excludes any Data Centre compute revenue from China.
The policy timeline is material context here: the Trump administration banned H20 chip exports to China in April 2025, reversed that decision in July 2025, then cleared H200 sales to China subject to a 15% export levy on revenue generated from those shipments.
During Q1 FY2026, NVIDIA absorbed a $4.5 billion inventory charge tied to stranded H20 stock from the original ban, and the company has previously noted that the China AI accelerator market could grow to nearly $50 billion - a figure that places the China exclusion in context against the broader revenue trajectory.
Bernstein senior analyst Stacy Rasgon has noted that restricting NVIDIA's access to China directs AI training demand toward Huawei, Cambricon and domestic alternatives, given that no Chinese-built accelerator currently matches Blackwell-generation performance benchmarks.
What to watch:
- Q2 revenue guidance of $91.0 billion sets the new consensus floor; the August print will be measured against that figure as the primary demand signal for the cycle
- GB300 Ultra's transition from sampling to volume production is the next hardware catalyst, with any scheduling change carrying direct read-through to TSMC and Broadcom estimates
- Export licensing policy for China Data Centre compute remains the key revenue variable for the second half of FY27
- The new Hyperscale versus ACIE reporting framework will provide granular visibility into enterprise AI adoption, which sits alongside the hyperscale segment in the Q1 revenue base
