Nvidia reports fiscal first-quarter 2027 results today after the US market close, and the bar is steep: Wall Street is modeling roughly $78 billion in revenue — about 77–78% year-on-year growth — against the company's own ~$78.0B guidance midpoint. Non-GAAP EPS consensus is $1.77, and the data-center segment alone is pegged near $73 billion, with some desks modeling above $75B. The fiscal quarter closed April 26.
The number behind the number
For infra teams, the headline revenue matters less than the mix. Data center is essentially the entire story now, and the read-through is supply: how much Blackwell capacity Nvidia is actually shipping, not just booking. Blackwell drove close to 70% of data-center compute last quarter and the Hopper transition is largely complete, so any commentary on GB300 NVL72 rack shipment pace at hyperscalers is the cleanest signal of when production inference capacity loosens.
Margins are the tell
Analysts expect gross margin near 74.5%, against company guidance closer to 75%. The threshold to watch is 73% — a print below that signals pricing pressure as Blackwell scales and the cost curve of liquid-cooled rack-scale systems bites. Holding the mid-70s while ramping a more complex platform would confirm Nvidia still owns pricing power at the rack level, which directly shapes what enterprises pay per token at production scale.
The Rubin handoff
With GB300 already in volume production — roughly two-thirds of Blackwell revenue last quarter — the next leg is the Vera Rubin transition from sampling to production: samples have shipped and production shipments are on track for 2H 2026. Jensen Huang has said he expects $1 trillion in cumulative revenue from Blackwell and Vera Rubin across 2026 and 2027, so the guidance commentary on Rubin readiness is what underwrites the multi-year compute roadmap that OpenAI, Meta, and the neoclouds are building against.
China remains the wildcard
There is no clean China data-center line guided into this print. The previously disclosed $4.5 billion inventory charge — taken on the China-specific H20 — is the baseline. With the Trump administration having approved H200 exports to China in January 2026 under a 25% surcharge and volume caps, and chips back in the market by February, the swing variable on both revenue and inventory is how much volume actually clears under those caps — not whether shipments reopen.
What to watch on the guide
The forward number carries more weight than the trailing one: Q2 FY27 revenue consensus sits at $85–87 billion, with whisper numbers near $90B. For builders, that guide is a proxy for 2026 GPU availability. A guide at or above $87B implies supply is finally catching demand into the back half of the year; a softer guide means the GPU wall holds, and capacity planning — reserved instances, multi-cloud commitments, and inference-cost models — stays defensive through year-end.
