NVIDIA expands AI hardware with Windows-based DGX Station

NVIDIA has announced the DGX Station for Windows, a deskside AI supercomputer designed to run trillion-parameter models locally. Unveiled at GTC Taipei, the system utilizes GB300 Grace Blackwell-class infrastructure and is expected to be available in the fourth quarter of 2026. The hardware is intended for enterprise developers and researchers who require high-performance AI capabilities within a Windows ecosystem.

In a separate development, CIQ has announced the CIQ Supercomputing Local AI Hackathon, scheduled for November 13–15 in Chicago. The event, held ahead of the Supercomputing 2026 (SC26) conference, will challenge up to 20 teams to scale AI workloads across multi-node clusters. Participants will use Lenovo ThinkStation PGX systems powered by NVIDIA GB10 Grace Blackwell Superchips, orchestrated via CIQ Fuzzball, to compete in tracks including multi-node training, inference at scale, scientific AI, and cluster scheduling.