CoreWeave just locked in a contract for Nvidia A100 GPUs that runs through 2029. For chips that first launched in 2020, that’s an unusually long shelf life, and CEO Michael Intrator wants you to know these machines are being booked at “full freight.”
The deal was disclosed during CoreWeave’s Q2 2026 earnings call on August 11, where the company also reported revenue of $2.575 billion, a 112% increase year-over-year. CFO Nitin Agrawal described the A100 contract as secured at an attractive price point, which is a polite way of saying the company expects to make very good money renting out hardware that most people assumed would be obsolete by now.
Old GPUs, new tricks
In July 2026, CoreWeave implemented a 25% price increase for older-generation GPUs. Average selling prices for these earlier models are holding steady or actually rising compared to the prior year. That’s the kind of pricing power most companies dream about for their newest products, let alone their legacy hardware.
The numbers behind the momentum
CoreWeave’s Q2 results paint the picture of a company riding a demand wave that shows no sign of cresting. Beyond the $2.575 billion revenue figure, the company disclosed a revenue backlog of $104.2 billion. That number excludes over $25 billion in early Q3 commitments that hadn’t been finalized at the time of reporting.
Management raised full-year 2026 revenue guidance to a range of $12.4 billion to $13.2 billion.
The relationship with Nvidia goes beyond a buyer-seller dynamic. Nvidia invested $2 billion in CoreWeave back in January 2026, cementing a partnership where Nvidia gets a reliable, high-volume customer for its chips and CoreWeave gets preferential access to supply.
From Ethereum rigs to AI empire
CoreWeave’s origin story is one of the more remarkable pivots in recent tech history. Founded in 2017, the company started life as an Ethereum mining operation. When it became clear that GPU compute had a far larger addressable market than cryptocurrency mining, especially as Ethereum transitioned to proof-of-stake, the company repositioned itself as a cloud infrastructure provider focused on GPU workloads.