Equinix plans to offer a managed service featuring the Blackwell Ultra-powered Nvidia DGX SuperPOD from data centers located in 45 markets around the world. The 'Nvidia Instant AI Factory' is planned to be rolled out to other data center operators in the future.

Nvidia unveiled its next-generation DGX SuperPOD AI infrastructure at GTC 2025, powered by Blackwell Ultra GPUs. Engineered for agent-based AI inference, the system provides enterprises with scalable,

Nvidia will also offer a single rack called the GB300 NVL72 that offers 1.1 exaflops of FP4, 20TB of HBM memory, 40TB of “fast memory,” and 130TB/sec of NVLink bandwidth and 14.4 TB/sec networking.

The Blackwell Ultra B300, set for release in the second half of 2025, increases memory capacity from 192GB to 288GB of HBM3e and offers a 50%

The Nvidia Blackwell Ultra DGX SuperPOD features the latest Blackwell Ultra GPUs, and comes in two flavors - DGX GB300 and B300. The DGX B300 can still be air-cooled and relies on the B300 NVL16 architecture. The GB300, meanwhile, combines the 72 Blackwell Ultra GPUs with 36 Grace CPUs and requires liquid cooling.

According to Nvidia, Vera Rubin will deliver significant performance improvements over its predecessor, Grace Blackwell, particularly for AI training and inference. Vera Rubin features two GPUs together on one die that deliver 50 petaflops of FP4 inference performance per chip.