Nvidia says its new Vera Rubin platform will account for one-fifth of data center revenue in a single quarter, and it has barely started shipping.
CFO Colette Kress told analysts on Nvidia's earnings call that Vera Rubin should make up about 20% of data center revenue in the third quarter of fiscal 2027. Nvidia projects roughly $108 billion in total revenue for the quarter, and with data center sales historically running around 92% of that total, Vera Rubin alone would generate close to $20 billion. Production of Vera Rubin components, including Vera CPUs, Rubin GPUs, and BlueField 4 DPUs, ramped up this spring, with first revenue shipments of VR200 NVL72 racks beginning in August; Microsoft has said it was first to deploy the systems commercially. Nvidia says it already has purchase orders from every major hyperscaler, AI cloud, and system OEM.
Going from near-zero revenue last quarter to $20 billion this one is an unusual ramp for any hardware line, and it shows hyperscalers committing to Vera Rubin before it has been proven at scale in the field. But the dollar figure flatters the underlying volume: at estimated prices of $5 million to $7.8 million per rack, that $20 billion could come from as few as roughly 2,550 to 3,975 systems.
Impressive as a growth chart, this looks less like mass adoption than a small number of customers writing very large checks.