NVIDIA (NVDA) GTC 2026 Keynote summary

Keynote highlights and strategic vision

• Introduced three core platforms: CUDA-X, systems, and AI Factories, emphasizing ecosystem growth and vertical integration.

• Celebrated 20 years of CUDA, highlighting its foundational role in AI and the global developer ecosystem.

• Stressed the importance of installed base and the accelerating flywheel effect driving adoption, innovation, and cost reduction.

• Outlined the transition from data centers to AI factories, with token generation as the new metric for value and revenue.

• Announced the Vera Rubin platform, a vertically integrated, liquid-cooled AI supercomputer designed for agentic AI and massive throughput.

Key announcements and product launches

• Introduced NVIDIA Vera Rubin NVL72, a new AI inference platform with 10X performance per watt and 3.6 exaflops NVFP4, launching with major partners in 2H26.

• Announced NVIDIA Vera CPU with 256 CPUs and 300 TB/s LPDDR5X bandwidth, and BlueField-4 STX with 5X tokens/sec and 50 Tb/s networking bandwidth.

• Launched NVIDIA NemoClaw, an open agent toolkit, and announced the Nemotron 3 Ultra model with 5X efficiency and highest reasoning accuracy on GB200 NVL72.

• Showcased the expansion of the DSX AI Factory Platform for scalable, energy-efficient AI factory deployment.

• Highlighted the formation of the NVIDIA Nemotron Coalition, bringing together global AI leaders to advance open frontier models.

Platform and ecosystem advancements

• Announced deep integration of CUDA libraries with major partners like IBM, Dell, and Google Cloud, accelerating data processing and AI workloads.

• Showcased partnerships with hyperscalers (AWS, Azure, Oracle, CoreWeave) and the expansion of AI-native cloud services.

• Emphasized vertical integration with horizontal openness, supporting domain-specific acceleration across industries.

• Noted the rise of AI-native startups and a $150 billion surge in venture investment, driven by unprecedented compute demand.

• Expanded support for open-source AI, with NVIDIA recognized as the world's largest contributor to open-source AI models.