NVIDIA AI Updates: June 11, 2026

1. NVIDIA Introduces Halos OS as a Safety-Certifiable Foundation for Robotaxis

NVIDIA. NVIDIA introduced the Halos Operating System, a safety-certifiable software foundation for autonomous vehicles that combines a certified OS architecture, standardized hardware and software interfaces, AI safety guardrails, and validation frameworks aimed at Level 4 robotaxis. The platform spans three computing layers, using DGX systems for training, Omniverse for simulation, and AGX in-vehicle computers for real-time processing across the development lifecycle. NVIDIA also announced robotaxi partnerships built on its DRIVE Hyperion platform, including Uber and Autobrains in Munich, Foxconn in Taiwan, VinFast in Southeast Asia, and HUMAIN in Saudi Arabia. Source

2. NVIDIA Accelerates Google DeepMind’s DiffusionGemma for Local AI

NVIDIA. NVIDIA optimized Google DeepMind’s DiffusionGemma model to run on NVIDIA GPUs, using parallel generation that denoises up to 256 tokens per step rather than predicting one token at a time. Built on the Gemma 4 architecture, the model generates text blocks simultaneously and delivers up to 4x faster performance than traditional autoregressive language models in single-user scenarios. DiffusionGemma is available under the Apache 2.0 license and runs locally on GeForce RTX GPUs, RTX PRO workstations, and DGX Spark systems, with support from Hugging Face Transformers, vLLM, and Unsloth. Source

3. NVIDIA Publishes Developer Guidance for High-Throughput DiffusionGemma Deployment

NVIDIA. NVIDIA published a developer-focused guide for running DiffusionGemma in high-throughput text generation workloads on NVIDIA hardware. The post details how the diffusion-based language model’s parallel token generation can be deployed for developer-ready inference, complementing the consumer-oriented local AI release. It targets teams looking to integrate the Gemma 4-based diffusion model into production text generation pipelines. Source

4. NVIDIA Details Production-Ready Battery Energy Storage Systems for AI Factories

NVIDIA. NVIDIA published guidance on designing production-ready battery energy storage systems for AI factories, addressing the power demands of large-scale GPU data centers. The post covers how energy storage can support the high and variable power profiles of AI training and inference infrastructure. It is aimed at operators planning the electrical and energy resilience requirements of facilities housing dense accelerated computing deployments. Source