NVIDIA GTC 2026: Agentic AI Infrastructure Across Models, Agents, and Robotics

Vera Rubin GPU announced as next-generation platform for agentic AI

Vera Rubin is NVIDIA's next-generation GPU architecture designed specifically for the high-throughput, low-latency demands of agentic AI workloads. It marks a deliberate shift in GPU design philosophy toward serving AI agents rather than purely training large models.

Vera CPU introduced as purpose-built processor for agentic AI orchestration

The Vera CPU complements the Vera Rubin GPU by handling orchestration tasks that agentic AI systems require — managing multi-step reasoning, tool calls, and memory — workloads that general-purpose CPUs handle inefficiently at scale.

Dynamo inference operating system enters production for AI factories

Dynamo is NVIDIA's system-level software for managing inference at scale inside AI factories. Its move to production availability means enterprises can now deploy it in live environments, making it a critical piece of infrastructure for organizations running large fleets of AI agents.

BYD, Geely, Isuzu, and Nissan adopt DRIVE Hyperion for Level 4 vehicles; Hyundai/Kia expand partnership

Multiple major global automakers are standardizing on NVIDIA's DRIVE Hyperion platform for Level 4 autonomous vehicle programs. The breadth of adoption — spanning Chinese, Japanese, and Korean OEMs — indicates DRIVE Hyperion is emerging as a de facto industry platform for autonomous driving compute.

Adobe partners with NVIDIA to integrate Firefly models into agentic creative workflows

The Adobe-NVIDIA strategic partnership connects Adobe Firefly generative models with NVIDIA's agentic AI infrastructure, targeting creative and marketing automation use cases. This extends NVIDIA's agentic AI stack into the enterprise creative software market beyond traditional data center and robotics applications.

Nemotron Coalition launched to coordinate open frontier model development across global AI labs

NVIDIA formed the Nemotron Coalition, a consortium of global AI labs collaborating on open frontier model development. A separate NemoClaw release targets the OpenClaw open-source community, positioning NVIDIA as a supporter of open-model ecosystems alongside its proprietary stack.

Open Physical AI Data Factory Blueprint released for robotics and autonomous vehicles

The blueprint provides a reference architecture for building data pipelines that physical AI systems — robots, autonomous vehicles, vision agents — require for training and continuous improvement. Open-sourcing this infrastructure lowers barriers for developers building real-world AI applications and expands NVIDIA's developer ecosystem.