Nvidia's $81.6 Billion Quarter Comes With a $91 Billion Forecast and No China

Nvidia reported fiscal first-quarter revenue of $81.6 billion on May 20, up 85% from a year earlier and 20% from the previous quarter. It guided the current quarter to $91 billion, give or take 2%. The company's forecast leaves out data center compute sales to China entirely, which means the growth it is promising does not depend on a market that politics keeps opening and closing.

What the quarter showed

The quarter ended April 26. Data center revenue reached $75.2 billion, up 92%. Within it, compute (the GPUs themselves) brought in $60.4 billion, up 77%, and networking, the switches and interconnects that tie thousands of GPUs together into one machine, brought in $14.8 billion, up 199% and 35% from the prior quarter. Nvidia also changed how it reports: it now splits results into Data Center and Edge Computing, and divides data center into hyperscale customers, at about $38 billion, and a second group labeled ACIE, at about $37 billion.

The reporting change is telling. Hyperscalers, the big cloud companies, are now roughly half of data center sales. The rest comes from AI labs, sovereign projects, enterprises and other buyers, a broader base than the "four customers" narrative that has haunted the stock. EPS figures vary across outlets, with $1.87 the most widely cited, so check Nvidia's release if the exact figure matters.

China as a rounding error, by design

Export restrictions on advanced AI processors cost Nvidia about $2.5 billion in revenue last quarter, according to Techzine, and the company expects to lose roughly another $8 billion in the current one. By writing China out of the guidance, management turns a recurring source of upside and downside into a clean baseline. If licenses or tariffs shift, the change shows up as a surprise on the upside rather than a cut to forecasts.

That framing follows the January decision to permit limited H200 sales while taxing them with a 25% tariff. Nvidia is effectively saying it will not plan around policy it cannot control.

Why networking matters more than it looks

A 199% rise in networking is the most interesting line in the report. AI clusters are limited as much by how fast chips can talk to each other as by how many exist, so the more GPUs a customer deploys, the more switching and cabling it needs. Networking is rising faster than compute because racks are growing larger and the interconnect is becoming the bottleneck. It is also a stickier product: once a data center is wired around Nvidia's fabric, swapping in another vendor's accelerators gets harder.

The wider backdrop is the $725 billion that the largest cloud companies plan to spend this year. Nvidia's results are where that spending shows up as revenue, and the quarter shows it landing on schedule, with room for more.

How big is big

A year ago, Nvidia's quarterly revenue was less than half of what it posted now, and the sequential gain alone, about $14 billion in a single quarter, exceeds the full-quarter revenue of most large semiconductor companies. The Q2 guide implies another jump of roughly $9 billion. Growth at that scale is without precedent in chips, and it is the reason Nvidia's market value has kept pace with the largest companies in the world, topping $5.5 trillion at one point.

The scale also changes the risk. Nvidia's product cycle is now the pacing item for the industry. If a new platform slips, or if a key supplier such as an advanced packaging partner or a memory maker falls short, the delay shows up in the capex plans and revenue forecasts of its customers. Management said little to suggest that is happening, but the margin for error narrows as expectations rise.

Sovereign buyers and enterprises also deserve attention. Their share in the ACIE group is nearly equal to that of the hyperscalers, and, unlike the cloud giants, they do not build competing chips. If that segment keeps growing, it eases the worry that Nvidia depends on customers who are also trying to replace it.

What to watch

• Whether Nvidia lands at or above the $91 billion guide in August.
• The balance between hyperscale customers and the broader ACIE group, a measure of how diversified demand has become.
• Any change in China licensing that would add revenue the guidance excludes.
• Gross margin and component costs, given memory price pressure across the supply chain.

Sources

1. Nvidia Q1 earnings call highlights — MarketBeat
2. AI data centers drive Nvidia to record revenue — Techzine
3. Nvidia guides $91B Q2: AI factory era fuels record earnings — HotHardware
4. Nvidia results for the first quarter and fiscal 2027 — Jon Peddie Research