The Nvidia sign outside its Voyager building in Santa Clara, California. Nvidia.
Gear / The story

Data centers brought in 93% of Nvidia’s $115.8 billion quarter

Nvidia expects $130 billion in sales next quarter. An extra week and a lower projected margin change what that figure says about growth.

GeekRadar3 min read

Of every $100 Nvidia recorded in sales last quarter, about $93 came from data centers. The company reported $107.6 billion in Data Center revenue out of $115.8 billion overall for the quarter ended October 25. Sales into that market alone brought in nearly twice as much as the whole company did in the same quarter a year earlier. Nvidia published its fiscal third-quarter results on November 17.

That concentration changes how to read the numbers. Nvidia is familiar to millions of people as a graphics-card company, yet even a careful assessment of demand for gaming hardware would tell you remarkably little about why its total sales doubled. The business now depends much more heavily on whether customers can finance, equip and bring large computing facilities into service.

Where the growth came from

Data Center sales rose 110% from a year earlier, compared with 103% for Nvidia as a whole. Put another way, company revenue grew by $58.8 billion year over year, from $57.0 billion to $115.8 billion. Using the reported growth rate to estimate the earlier Data Center figure, roughly $56 billion of that increase came from this one market. The estimate is approximate because Nvidia rounds both the sales figures and the percentage.

The remaining businesses together generated about $8.2 billion this quarter, by subtraction. That is a substantial business in its own right, but the total includes several markets. Assigning the entire remainder to gaming would overstate what players spent with Nvidia. The same caution applies in reverse: Data Center revenue measures what Nvidia sold into that market. It cannot tell us how many hours customers used the equipment or how much money they earned from running models on it.

For a cloud operator, the difference is tangible. Equipment can arrive before a facility has enough power or before the operator has paying customers for all its capacity. Nvidia can report strong sales while the operators buying those systems still face very different economics. In her earnings-call comments, interim CEO Colette Kress identified site readiness and customers’ access to capital among the assumptions behind the company’s longer-term outlook. TechCrunch’s account of the call follows our coverage of Kress’s appointment.

Colette Kress, Nvidia’s interim chief executive and chief financial officer. Nvidia.
Colette Kress, Nvidia’s interim chief executive and chief financial officer. Nvidia.

An extra week changes the comparison

Nvidia expects fourth-quarter revenue of $130 billion, give or take 3%. The midpoint implies an increase of about 12.3% over the third quarter. The fourth quarter, however, has an extra week in which to recognise sales. Anyone using that 12.3% as a measure of acceleration is comparing two different lengths of time. Nor can the value of the extra week be read straight off the calendar: large equipment deliveries need not occur at a uniform weekly pace.

Those two forecasts can coexist. At the revenue midpoint and a 71.5% margin, the figures imply roughly $93 billion in gross profit, compared with about $87 billion obtained from the rounded third-quarter figures. This is an illustration using management’s assumptions. Gross profit still has to cover operating expenses, and the quarter has yet to close.

One geographical qualification matters here. Nvidia assumes no Data Center compute revenue from China in the outlook. The wording is specific to that activity and that forecast period. The two figures to revisit when the actual results arrive are how many additional dollars of gross profit the longer quarter produced and how much Nvidia retained from each dollar of sales. A larger total can come with a smaller share of every sale left over.