- Nvidia’s Q4 data center revenue of $62.3 billion was 75 percent higher than the same quarter in 2024. The Q1 guidance of $78 billion implies the acceleration is continuing, not plateauing.
- Blackwell, Nvidia’s latest GPU architecture, generated $11 billion in its first full quarter of production — what management described as “the fastest product ramp ever” in the company’s history.
- At $78 billion in guidance, analysts expect earnings per share of $1.76 — a 117 percent year-on-year increase — making the question not whether Nvidia beats but by how much, and what forward guidance signals about demand into Q2.
Nvidia will report its fiscal first-quarter results on May 20, with management having guided to $78 billion in revenue — plus or minus 2 percent — for the quarter ending April 2026. The guidance represents a significant sequential increase from the $68.1 billion in fourth-quarter revenue, all of it driven by an AI compute demand cycle that has now stretched well past initial enthusiasm into what several analysts describe as a multi-year infrastructure buildout by hyperscale cloud providers.
The Q4 results, reported in late January, showed data center revenue of $62.3 billion — up 75 percent year-on-year — from customers including Microsoft Azure, Amazon Web Services, Google Cloud, and Meta’s AI infrastructure division. The Blackwell platform, which generated $11 billion in Q4 alone, is described by Jensen Huang as “the fastest ramp ever” for a new GPU architecture, and the guidance implies Blackwell’s contribution is continuing to expand in Q1.
The analyst expectations game
With consensus estimates for Q1 at $78.5 billion in revenue and $1.76 earnings per share — representing 78 and 117 percent year-on-year growth, respectively — the question framing the May 20 report is not whether Nvidia beats guidance but by how much, and what the Q2 guidance signals about durability. In recent quarters, Nvidia has guided conservatively and beaten by a margin that caused significant upward price moves. If Q1 guidance was also conservative, a beat of several billion dollars would not surprise the market.
AMD’s MI300X and forthcoming MI325X chips have made progress in serving some hyperscale workloads, particularly inferencing, where Nvidia’s pricing premium is harder to justify than in training. Intel has announced next-generation products but faces significant execution risk. For now, Nvidia controls an estimated 80 percent of the AI accelerator market and the Blackwell architecture’s performance advantages are sufficient to maintain that share against known alternatives.
Geopolitics on the balance sheet
One risk factor that Nvidia’s Q1 report will need to address is the impact of US export controls on its China revenue. The restriction on H100 and H200 chip exports to China — now under a case-by-case review framework rather than a blanket ban following the January 2026 rule change — has reduced but not eliminated Chinese customers from Nvidia’s order book. Jensen Huang attended the Trump-Xi summit in Beijing on May 14, signalling the company’s appetite for Chinese commercial access and the administration’s awareness that chip export policy is now a bilateral diplomatic issue rather than a purely commercial one.