NVIDIA launches Enterprise Reference Architectures to help organizations design and scale AI data centers from four-node clusters to enterprise environments

NVIDIA Enterprise Reference Architectures combine certified compute, high-speed networking, observability tools, and software to enable organizations to design, deploy, and scale high-performance AI factories using validated, repeatable infrastructure.

Three primary cluster configurations are available: NVIDIA RTX PRO AI Factory (optimized for 16- and 32-node design points across generative AI, data analytics, and simulation); NVIDIA HGX AI Factory (32-, 64-, and 128-node design points for multi-node training and inference, achieving up to 15x higher token throughput); and NVIDIA NVL72 AI Factory (four- and eight-rack configurations designed to train and deploy trillion-parameter models within a single rack).

Palantir has teamed with NVIDIA to deliver the Sovereign AI OS Reference Architecture, which is based on NVIDIA Enterprise RAs and tested to run Palantir's complete software suite on NVIDIA AI infrastructure.