On March 18th, at the SAP Center in San Jose, California, the GTC conference captured global attention with its live broadcast, hosted by NVIDIA.

As one of the most prominent entities in the AI sphere, NVIDIA's unveiling of new products at the conference felt like dropping several seismic breakthroughs in the industry, shaking practitioners still feeling the aftershocks of recent AI technological revolutions like Open AI and Sora.

Major Breakthrough One: Unveiling the all-new Blackwell architecture and the latest B200 GPU

• The B200 GPU, the first product featuring the NVIDIA Blackwell architecture, adopts a dual-chip design and is hailed as the most powerful GPU ever. Combining two B200 GPUs with the Grace CPU forms the GB200 super chip.

• These chips, when integrated into a Blackwell computing node and interconnected via NVIDIA Quantum-X800 InfiniBand and Spectrum™-X800 Ethernet platforms released at the conference, result in the GB200 NVL72, boasting 36 CPUs and 72 GPUs. This configuration enhances large model inference performance by up to 30 times while reducing costs and energy consumption by nearly 25 times.

• The GB200 NVL72 further connects via Quantum InfiniBand switches and cooling systems to form the DGX GB200 SuperPod supercomputing platform, comprising 32,000 GPUs in a distributed cluster, capable of elastic scaling and providing 11.5 Exaflops of computing power and 240TB of high-speed memory.

Major Breakthrough Two: Introduction of the integrated AI development software microservice system NIM

• NVIDIA introduced NIM (NVIDIA INFERENCE MICROSERVICE), an integrated AI development software microservice system, allowing traditional industries lacking AI development experience to run pre-packaged and optimized pre-trained models on NVIDIA's infrastructure. This system finds wide applications in healthcare, enabling medical institutions to screen trillions of drug compounds for drug development and collect patient data for early disease screening and prediction.

Major Breakthrough Three: Unveiling the humanoid robot basic model named GROOT

• NVIDIA unveiled GROOT, a basic model of a humanoid robot, along with the new Thor computer designed for robots. NVIDIA optimized its performance, power consumption, and size for robot applications. The project also includes the ISAAC Lab development toolkit and the Jetson Thor SoC system-on-chip development kit.

Major Breakthrough Four: Introduction of the new "DRIVE Thor" automotive computing platform

• As the successor to the previous generation "DRIVE Orin," this platform, equipped with the latest NVIDIA Blackwell architecture, offers enhanced performance for autonomous driving systems.

Major Breakthrough Five: Release of communication, climate governance, quantum computing, and 6G-related technologies

• During the conference, Huang Renxun stated that the collaboration with BYD to build automotive and industrial digital twins based on Omniverse would extend beyond the fields of robotics and digital twins to future 6G communication technology. In addition to 6G communication networks, these technologies can be applied to meteorological and climate observations. NVIDIA announced the NVIDIA Earth 2 Earth Climate Digital Twin Suite to accelerate weather forecasting, leveraging Omniverse's unified platform for computer graphics, AI, scientific computing, and physical simulation.

Major Breakthrough Six: Introduction of the latest digital human technology

• Originating from the gaming industry, NVIDIA showcased its digital human platform at the conference, featuring three main technologies: NVIDIA ACE, NVIDIA NeMo, and RTX ray tracing technology. These technologies drive game character language, speech, animation, and graphics through AI. NVIDIA ACE aids developers in driving facial animations using Audio2Face and voice communication driven by Riva's automatic speech recognition (ASR) and text-to-speech (TTS) capabilities. NVIDIA NeMo helps developers deliver enterprise-grade generative AI models through precise data management, personalized model customization, enhanced retrieval generation, and accelerated performance. Finally, the collection includes rendering technologies such as RTX global lighting DLSS 3.5, enabling real-time ray tracing in applications.

The release of NVIDIA's new product series undoubtedly brings humanity closer to the envisioned AI world. NVIDIA has built a robust technological moat around the ecosystem domain, significantly accelerating the pace of AI industry exploration and driving the global AI industry chain's renewal and iteration.

Gooxi, a leading server solutions provider in China, will leverage the latest GPU technology in the AI era to accelerate the renewal of AI servers and solutions in the industry. Additionally, based on its own industry advantages, Gooxi will collaborate with all participants in the AI arena to promote industry progress and development, striding forward in the tide of the times.