Nvidia Unveils Personal AI Computer and Blackwell Ultra Chip at GTC 2025

Blackwell Ultra and Dynamo: The Next Step in GPU AI

One of the most important announcements at GTC 2025 (USA) was the Blackwell Ultra platform, an upgraded version of the previous Blackwell architecture. Blackwell Ultra is designed to provide significantly higher performance than the Hopper generation of GPUs (graphics cards), helping to improve AI inference processing. According to Nvidia, this platform can increase the token processing speed per second by up to 25 times compared to the previous generation, helping AI models run faster and more efficiently.

Blackwell Ultra NVL72 is a data center GPU system that supports high-performance AI processing with high-speed NVLink connectivity

PHOTO: REVELATION

In addition, Nvidia also integrates the next generation NVLink, allowing multiple Blackwell Ultra GPUs to be connected to form high-performance computing systems. This is especially important as the demand for AI computing increases, especially in the fields of scientific research and business.

Alongside the hardware, Nvidia is introducing Dynamo, a new software platform that it describes as the “operating system for the AI ​​factory.” Dynamo helps manage the entire AI lifecycle, from training to deployment to model optimization. The platform makes it easy for enterprises to deploy AI at scale, optimize performance on Blackwell Ultra GPUs, and reduce model training times.

DGX and DGX Spark: Personal AI Supercomputers

One of the most notable announcements at GTC 2025 was the DGX Station and DGX Spark personal AI computers. These are compact supercomputers with powerful AI processing capabilities, allowing researchers and businesses to develop AI right in the office.

DGX Station uses the GB10 Grace Blackwell Superchip, capable of performing up to 1 quadrillion AI calculations per second. Meanwhile, DGX Spark is a smaller version, aimed at small and medium-sized businesses at a more reasonable price, expanding AI accessibility to a wider audience.

DGX Spark is designed to support AI tasks in small and medium-sized businesses, using Grace Blackwell architecture to handle machine learning models and big data

PHOTO: REVELATION

At GTC 2025, Nvidia also emphasized the role of CUDA-X, an accelerated computing platform applied in many fields from physics, data science, computer engineering, deep learning, quantum computing to medicine and telecommunications. CUDA-X includes specialized libraries such as cuDNN, cuBLAS, cuOPT, helping to optimize processing performance on Nvidia GPUs.

One of the highlights is CUDA-Q, a new toolkit for quantum computing that makes the technology more accessible to researchers and businesses. By continually expanding the CUDA-X ecosystem, Nvidia is strengthening its position not only in AI but also in many other cutting-edge technology areas.

Strategic Cooperation and AI Ecosystem Expansion at GTC 2025

Nvidia also announced several important partnerships with technology corporations such as Google, Cisco, HPE and General Motors. These agreements focus on developing AI computing infrastructure, optimizing telecommunications networks and supporting AI applications in enterprises.

Additionally, Nvidia continues to expand its AI robotics and autonomous driving platforms, with new enhancements to the Nvidia Isaac platform for robotics and DRIVE Thor for autonomous driving. These upgrades improve the environmental perception, navigation, and decision-making capabilities of AI systems.

Nvidia and General Motors Partnership Focuses on AI Applications in Self-Driving Cars, Driver Assistance Systems

PHOTO: REVELATION

GTC 2025 demonstrated that Nvidia continues to expand its influence in the field of AI and accelerated computing. With products such as Blackwell Ultra, Dynamo, DGX Station, DGX Spark and the CUDA-X ecosystem, Nvidia is not only focusing on hardware but also building a strong software and service platform, accelerating the development of AI globally.

The event will continue with many in-depth discussions on AI trends, deep learning, supercomputers and cloud computing, promising to bring important information to the technology world.

Source: https://thanhnien.vn/nvidia-ra-mat-may-tinh-ai-ca-nhan-va-chip-blackwell-ultra-tai-gtc-2025-185250319054344816.htm