NVIDIA (NVDA) COMPUTEX 2025 Keynote summary

Industry vision and transformation

• AI is positioned as the next essential infrastructure, following electricity and the internet, with every region and industry needing AI integration.

• Data centers are evolving into AI factories, producing valuable tokens and representing a multi-trillion dollar industry opportunity.

• Accelerated computing, AI, and specialized libraries (CUDA-X) are central to driving new markets and applications.

• The company is shifting from a technology provider to an essential infrastructure company, revealing detailed multi-year roadmaps to enable global planning.

• AI's evolution is highlighted from perception to generative and agentic AI, with reasoning and multimodal capabilities enabling digital and physical agents.

Product launches and ecosystem expansion

• Announced the Grace Blackwell GB300 system, offering 1.5x inference performance and 2x networking, with full production and global rollout in Q3.

• Introduced NVLink Fusion, enabling semi-custom AI infrastructure by integrating custom CPUs, ASICs, and accelerators into the NVLink ecosystem.

• Launched DGX Spark and DGX Station, personal AI-native supercomputers for developers and researchers, available through major partners.

• RTX Pro Enterprise and Omniverse server unveiled for enterprise IT, supporting both traditional and AI-native workloads with high performance.

• Announced NVIDIA Constellation, a new office in Beitou-Shilin, Taipei, to support regional growth and partnerships.

Partnerships and industrial impact

• Collaborations with TSMC, Foxconn, and the Taiwanese government to build a giant AI supercomputer for Taiwan's AI ecosystem.

• Over 150 companies contributed to the Blackwell ecosystem, with partners like Dell, HPE, ASUS, and others supporting new product categories.

• NVLink Fusion ecosystem includes partners such as MediaTek, Fujitsu, Qualcomm, Cadence, and Synopsys, enabling broad industry integration.

• Storage and AI data platforms developed with Dell, Hitachi, IBM, NetApp, and VAST, integrating advanced AI models for enterprise use.

• Robotics and digital twin initiatives involve partners like Google DeepMind, Disney Research, and major Taiwanese manufacturers using Omniverse.