"Nvidia CEO Huang Announces New Blackwell Ultra Chip for AI Reasoning"

Nvidia CEO Jensen Huang has just announced the next evolution of the NVIDIA Blackwell AI factory platform, the NVIDIA Blackwell Ultra. This new chip is set to revolutionize the AI industry by boosting training and test-time scaling inference, paving the way for the age of AI reasoning. The Blackwell Ultra is built on the groundbreaking Blackwell architecture introduced a year ago and includes the NVIDIA GB300 NVL72 rack-scale solution and the NVIDIA HGX™ B300 NVL16 system. The GB300 NVL72 delivers 1.5x more AI performance than the NVIDIA GB200 NVL72, as well as increases Blackwell’s revenue opportunity by 50x for AI factories, compared with those built with NVIDIA Hopper™.

The Blackwell Ultra platform is designed to handle complex AI tasks such as AI reasoning, agentic AI, and physical AI. The NVIDIA GB300 NVL72 connects 72 Blackwell Ultra GPUs and 36 Arm Neoverse-based NVIDIA Grace™ CPUs in a rack-scale design, acting as a single massive GPU built for test-time scaling. This configuration allows AI models to access increased compute capacity, enabling them to explore different solutions to problems and break down complex requests into multiple steps, resulting in higher-quality responses. This is a clear advancement over previous generations, which lacked such integrated and scalable solutions.

The NVIDIA HGX B300 NVL16 features 11x faster inference on large language models, 7x more compute, and 4x larger memory compared with the Hopper generation. This delivers breakthrough performance for the most complex workloads, such as AI reasoning. The platform is ideal for applications including agentic AI, which uses sophisticated reasoning and iterative planning to autonomously solve complex, multistep problems, and physical AI, enabling companies to generate synthetic, photorealistic videos in real time for the training of applications such as robots and autonomous vehicles at scale. These specific advancements highlight the Blackwell Ultra's superior capabilities in handling advanced AI tasks compared to its predecessors.

The Blackwell Ultra platform is supported by the full-stack NVIDIA AI platform. The NVIDIA Dynamo open-source inference framework — also announced today — scales up reasoning AI services, delivering leaps in throughput while reducing response times and model serving costs. This makes the Blackwell Ultra platform ideal for applications including agentic AI, which uses sophisticated reasoning and iterative planning to autonomously solve complex, multistep problems, and physical AI, enabling companies to generate synthetic, photorealistic videos in real time for the training of applications such as robots and autonomous vehicles at scale.

The Blackwell Ultra platform is expected to have a significant impact on the broader semiconductor industry. The chip's advanced capabilities and NVIDIA's dominance in the AI chip market could drive demand for AI chips, leading to increased revenue for the semiconductor industry. For example, NVIDIA's revenue was $130.50 billion in 2024, an increase of 114.20% compared to the previous year. This growth is likely to continue as AI becomes more prevalent in various industries.

The Blackwell Ultra platform is also expected to have a significant impact on the development and adoption of AI technologies in various sectors. The integration of Blackwell Ultra into various AI applications, such as agentic AI and physical AI, is poised to significantly influence the development and adoption of these technologies across different sectors. For instance, in healthcare, the integration of Blackwell Ultra can lead to the development of AI agents that can assist in diagnosing diseases, managing patient care, and even performing surgeries with greater precision and reliability. In the finance sector, AI agents powered by Blackwell Ultra can manage portfolios, execute trades, and provide financial advice with enhanced accuracy and speed, leading to better investment outcomes. In manufacturing, physical AI enabled by Blackwell Ultra can train robots to perform complex tasks, leading to increased efficiency and productivity on the factory floor. In the automotive industry, Blackwell Ultra can be used to train autonomous vehicles to navigate real-world scenarios more effectively, leading to safer and more reliable self-driving cars.



In conclusion, the Blackwell Ultra chip represents a significant leap in AI reasoning capabilities compared to its predecessors. The chip's advanced capabilities and NVIDIA's dominance in the AI chip market could drive demand for AI chips, leading to increased revenue for the semiconductor industry. The integration of Blackwell Ultra into various AI applications, such as agentic AI and physical AI, is poised to significantly influence the development and adoption of these technologies across different sectors. This could lead to more accurate diagnoses in healthcare, better investment outcomes in finance, increased productivity in manufacturing, and safer autonomous vehicles in the automotive industry.