NVIDIA announced that GPU cloud platform CoreWeave is among the first cloud providers to bring NVIDIA GB200 NVL72 systems online at scale, with Cohere, IBM and Mistral AI using them for model training and deployment.
The CoreWeave-NVIDIA relationship is well-known – NVIDIA has invested heavily in CoreWeave when CoreWeave was a private company and now, as a publicly held one, and CoreWeave, which is an NVIDIA preferred cloud services provider, has adopted NVIDIA GPUs in its AI cloud infrastructure. Last year, the company was among the first to offer NVIDIA H100 and NVIDIA H200 GPUs, and was one of the first to demo NVIDIA GB200 NVL72 systems.
CoreWeave said its portfolio of cloud services are optimized for the GB200 NVL72, offering CoreWeave’s Kubernetes Service, Slurm on Kubernetes (SUNK), Mission Control, and other services. CoreWeave’s Blackwell instances scale to up to 110,000 Blackwell GPUs with NVIDIA Quantum-2 InfiniBand networking.
NVIDIA said Cohere is using its Grace Blackwell Superchips to help develop secure enterprise AI applications. Its enterprise AI platform, North, enables teams to build personalized AI agents to automate enterprise workflows, surface real-time insights and more. The company said Cohere is experiencing up to 3x more performance in training for 100 billion-parameter models compared with previous-generation NVIDIA Hopper GPUs — even without Blackwell-specific optimizations.
IBM‘s deployment is scaling to thousands of Blackwell GPUs on CoreWeave to train its Granite open-source AI models used for IBM watsonx Orchestrate to build and deploy AI agents. The deployment also works with the IBM Storage Scale System for AI.
Mistral AI, a Paris-based open-source AI company , is getting its first thousand Blackwell GPUs to build the next generation of open-source AI models, according to NVIDIA. The company said this requires GPU clusters with NVIDIA Quantum InfiniBand networking and infrastructure management capabilities, such as CoreWeave Mission Control.
The company saw a 2x improvement in performance for dense model training, according to Thimothee Lacroix, cofounder and chief technology officer at Mistral AI. “What’s exciting about NVIDIA GB200 NVL72 is the new possibilities it opens up for model development and inference.”
“Enterprises and organizations around the world are racing to turn reasoning models into agentic AI applications that will transform the way people work and play,” said Ian Buck, vice president of Hyperscale and HPC at NVIDIA. “CoreWeave’s rapid deployment of NVIDIA GB200 systems delivers the AI infrastructure and software that are making AI factories a reality.”
The company recently reprted an industry record in AI inference with NVIDIA GB200 Grace Blackwell Superchips, reported in the latest MLPerf v5.0 results. MLPerf Inference is a benchmark suite for measuring machine learning performance across realistic deployment scenarios.
CoreWeave also offers instances with rack-scale NVIDIA NVLink across 72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs, scaling to up to 110,000 GPUs with NVIDIA Quantum-2 InfiniBand networking.
These instances, accelerated by the NVIDIA GB200 NVL72 rack-scale accelerated computing platform, provide the scale and performance needed to build and deploy the next generation of AI reasoning models and agents.