Databricks Launches AIR so Teams Train NVIDIA Models Without Managing Clusters
Databricks AI Runtime (AIR) has hit general availability, offering serverless NVIDIA GPUs, NeMo integration, and day-one support for Nemotron 3.5 models, so teams can train NVIDIA models without managing clusters. Why it matters: it pushes the ops burden of enterprise-grade GPU training close to zero, accelerating training-as-a-service into the mainstream, and the day-one Nemotron 3.5 support shows how deep the Databricks–NVIDIA tie-in runs.
SOURCES
- Alpha Signal2026-10-21 16:00
- Hacker News (AI)2026-10-22 00:29