NVIDIA Announces America’s Most Powerful Open Model “Nemotron 3 Ultra” and Reports Start of Mass Production for AI Server “Vera Rubin”

NVIDIA unveiled Nemotron 3 Ultra, a 550-billion-parameter open model, at GTC Taipei 2026.

The company said Vera Rubin AI servers are now in full production, boosting data-center capacity.

New Windows-focused systems, including RTX Spark and DGX Station, expand NVIDIA’s enterprise AI lineup.

The launches are meant to improve model performance, inference speed, and overall AI infrastructure.