AI cloud compute provider CoreWeave activated a multi-rack Nvidia Vera Rubin NVL72 cluster on its CoreWeave Cloud platform, linking hundreds of Rubin GPUs into a horizontally scaled system for agentic AI workloads. CoreWeave Executive Vice President and Head of Product and Engineering Chen Goldberg said the company is, to the best of his knowledge, the first AI cloud service provider to complete Vera Rubin NVL72 validation and put the system into production. CoreWeave shares rose nearly 5% intraday to $84.72, with trading volume reaching $283 million. Each rack contains 72 Rubin GPUs and 36 Vera CPUs, supported by Nvidia NVLink 6, ConnectX-9 SuperNIC network interface cards and BlueField-4 data processing units (DPUs). Multiple racks connect through Nvidia Spectrum-X Ethernet, allowing training and inference workloads to run across hundreds of GPUs rather than being limited to one rack. CoreWeave also launched cross-region write acceleration and an Archive storage tier for long-term retention without data retrieval, early deletion or Archive-tier reading fees. Its Local Object Transport Accelerator (LOTA) caches data closer to workloads, cutting read latency by 8x versus traditional storage clusters and providing throughput of up to 7 GB/s per GPU. CoreWeave shares are up approximately 2% year-to-date, compared with gains of more than 6% for the ARK Innovation ETF and over 25% for the iShares AI Innovation and Tech Active ETF, while Stocktwits retail sentiment remains bearish. The infrastructure announcement comes amid an intensifying debate over AI safety regulation involving Bill Gurley, Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, Tesla CEO Elon Musk, U.S. President Donald Trump and members of Congress. The broader question for investors is how the AI industry will balance accelerating compute investment with regulatory uncertainty.