AI news videos

CoreWeave Opens Vera Rubin Cloud to Customers

CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X 102.4T Ethernet networking at CoreWeave Fully Connected in San Francisco, making it one of the first cloud providers to put the platform in customers' hands. Cognition, the applied AI lab behind the Devin AI software engineer, is the first customer running production workloads on Vera Rubin.

Watch on YouTube

Transcript

Sources

Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:

• "Now, CoreWeave is bringing the next generation of NVIDIA infrastructure to production." Backs the hook's "NVIDIA's next generation AI systems just reached a public cloud".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "At CoreWeave Fully Connected, running this week in San Francisco, CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X 102.4T Ethernet networking." Backs scene one's "CoreWeave announced availability of NVIDIA Vera Rubin systems this week" and establishes that the venue was CoreWeave's own customer conference.(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "Cognition, the applied AI lab behind the Devin AI software engineer, is the first customer running production workloads on Vera Rubin." Backs scene one's "Cognition, the lab behind Devin, is the first in production".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "In its early tests, Cognition saw Vera Rubin NVL72 deliver up to a 4.8x increase in total token throughput for SWE-2 inference workloads over GB200 NVL72." Backs scene two's "up to four point eight times more token throughput than NVIDIA's previous generation".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "Shortly after, Cognition benchmarked Vera Rubin's inference performance against a GB200 NVL72 baseline using a real-world software engineering workload. To generate this workload, it sampled a subset of tasks from FrontierCode and deployed AI agents to solve them." Establishes that scene two's comparison baseline is an earlier NVIDIA generation and that the workload was a sampled software engineering task set, and backs scene four's "early tests run by Cognition".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "CoreWeave will also offer NVIDIA Vera, the first CPU built for AI agents." and "NVIDIA Vera CPU is purpose-built for agentic workloads." Backs scene three's "The Vera processor targets agent sandboxes".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "CoreWeave's deployment of Vera puts 128 CPUs and 11,264 cores in a single rack, enough for more than 11,000 concurrent environments at one core each." Backs scene three's "a single rack supports more than eleven thousand concurrent environments".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• "In testing, CoreWeave achieved more than 3x faster agent sandbox startup times on NVIDIA Vera CPUs, accelerating and scaling its sandboxes, an execution layer for reinforcement learning (RL), agent tool use and model evaluation that let AI teams run code in isolated environments on CoreWeave." Backs scene three's "Startup times tested over three times faster" and backs scene four's "early tests run by ... CoreWeave".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• The article is hosted on NVIDIA's own corporate blog at blogs.nvidia.com and its only named outside voices are a CoreWeave partner customer and an NVIDIA executive: "NVIDIA accelerated computing delivers value across generations," said Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, and "Agentic coding is a complex workload: long contexts, high concurrency and token volumes where cost per token decides what we can ship," said Silas Alberti, founding team at Cognition. No third party testing organization, auditor or reviewer is cited anywhere in the post for the throughput or startup figures. Backs the hook's "The early performance claims are NVIDIA's own" and scene four's "published on NVIDIA's own blog" and "No independent party verified them".(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• Context not narrated, recorded for the hub page: "In addition, CoreWeave launched CoreWeave Forge, a connected environment for training, evaluating and improving models and agents on NVIDIA accelerated computing." Forge "unifies Weights & Biases, post-training expertise from OpenPipe and the open source marimo notebook project in one connected environment built for continuous model and agent improvement", and "Canva, Capital One and MasterClass are among the first companies building on Forge."(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)
• Further context not narrated, recorded for the hub page: "Earlier this month, CoreWeave received its first Vera Rubin NVL72 production racks", "The company scaled to thousands of GPUs on CoreWeave in nine months, powering Cognition inference workloads", "On Terminal-Bench, CoreWeave saw a 1.7x performance gain on Vera CPU across all passing tasks", and on Agent Lens, "It improves failure detection by 20% and fixes issues at half of the cost". These additional vendor measured figures were left out of narration to keep each spoken number to a single mention.(NVIDIA (blogs.nvidia.com), first party blog post titled "From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI" by Stuart Pitts, dated September 30, 2026, alert time snapshot captured 2026-09-30T15:05:41.909Z used as primary source text.)