CoreWeave announces Vera Rubin NVL72 availability and Vera CPU deployment at FullyConnected26; Cognition reported as first production customer on Vera Rubin NVL72 with 4.8x throughput gain.
Following·New·first covered 30 Sep 2026·2 sources
CoreWeave Launches Vera Rubin NVL72 and Vera CPU on Its Platform
The gist
The Vera Rubin NVL72 availability gives cloud customers access to NVIDIA's latest GPU generation, with a production customer already reporting significant throughput gains for agentic inference workloads. The Vera CPU launch adds a purpose-built compute tier for agent orchestration alongside GPU training, which CoreWeave describes as enabling the full AI loop on one platform.
The full picture
CoreWeave has made NVIDIA's Vera Rubin NVL72 systems available on its cloud, with AI lab Cognition running its Devin software engineering agent in production on the hardware at up to 4.8x the total token throughput of the GB200 NVL72. CoreWeave also announced it will deploy the NVIDIA Vera CPU, described as the first CPU built specifically for AI agents, supporting more than 11,000 concurrent isolated environments per rack and delivering more than 3x faster agent sandbox startup times compared to prior infrastructure. Both announcements were made at the FullyConnected26 event.
How it developed
Sources
CoreWeave Twitter
The daily email
Want this in your inbox?
I send one email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free