August 2026 Summaries

During August, Lightning AI reported infrastructure, product, and community updates, including the rollout of H200e upgrades across thousands of racks as part of a broader fleet modernization effort and early access for Lightning VMs, which provide raw managed compute without the integrated development environment and tooling offered by Studios. Therna Biosciences described using Lightning Cloud to replace lengthy, costly H100 wait times and build a Boltz-2 structure-prediction web application in one day. The company also introduced Lightning Connections in early access, claiming up to 200-fold faster persistence for small-file datasets to improve ML workload resilience and reduce experiment reruns after GPU failures or memory interruptions. At the Ai4 conference, Lightning AI discussed enterprise AI infrastructure, expanding North American capacity, and custom clusters for partners, while citing forecasts that the AI-as-a-service market could approach $386 billion by 2035. It also announced a Santa Clara networking event for AI infrastructure builders during the AI Infra Summit.

Nemotron 3 Nano’s serverless Model API will be retired at 5 p.m. PT on August 19, after which the model will be disabled and requests will return HTTP 400 errors. Users are advised to update integrations before the deadline to use lightning-ai/Qwen3.8-27B, an alternative positioned as offering competitive general reasoning, coding, and long-context capabilities at lower cost. Organizations requiring continued access to Nemotron 3 Nano Omni can contact the provider to discuss a dedicated deployment.