Workflow
英伟达NVIDIA:Computex 2025期间发布关键技术 向开放生态平台转型

Core Insights - NVIDIA is entering a new phase of development as a leading provider of AI infrastructure, with its market value continuing to rise [1] - The company announced the launch of new technologies and products, including the NVIDIA NVLink Fusion chip, during the Computex 2025 event [1][2] - NVIDIA is transitioning from a single hardware supplier to an open ecosystem platform by allowing third-party access to NVLink IP [1][2] Group 1: New Technologies and Products - The NVLink Fusion technology enhances ecosystem compatibility, enabling integration with CPUs from Fujitsu and Qualcomm to build high-performance AI infrastructures [1][2] - NVIDIA introduced the RTX PRO server for enterprise-level AI inference, capable of supporting up to 8 Blackwell RTX Pro Graphics 6000 cards [2] - The company updated its robot foundational model, Isaac GR00T, and introduced a synthetic data generation framework for humanoid robot training [2][4] Group 2: Partnerships and Collaborations - Initial adopters of NVLink Fusion include MediaTek, Marvell, and others, but no Chinese companies are among the first group [2] - NVIDIA is collaborating with Foxconn and Taiwan partners to build a supercomputer with 10,000 Blackwell GPUs, with TSMC as a primary customer [5][6] - The company plans to establish a new office in Taiwan to strengthen its ecosystem partnerships [5][6] Group 3: Market Presence and Future Plans - NVIDIA's revenue in China is projected to be approximately $17 billion in 2024, accounting for about 14% of its global total [6] - The company is leasing new office space in Shanghai to accommodate current employees and prepare for future expansion [6]