NVIDIA Revolutionizes AI Infrastructure with NVLink Fusion
At COMPUTEX, NVIDIA unveiled its latest innovation: NVIDIA NVLink Fusion™, a groundbreaking technology that empowers industries to create semi-custom AI infrastructure. This development leverages the extensive ecosystem of partners building with NVIDIA NVLink™, the world’s most advanced and widely adopted computing fabric. Early adopters include MediaTek, Marvell, Alchip Technologies, Astera Labs, Synopsys, and Cadence, who are utilizing NVLink Fusion to scale up custom silicon for demanding AI workloads.
Empowering Custom AI Solutions
NVLink Fusion allows companies like Fujitsu and Qualcomm Technologies to integrate their CPUs with NVIDIA GPUs, creating high-performance NVIDIA AI factories. Jensen Huang, NVIDIA’s founder and CEO, emphasized the significance of this technology, stating, “A tectonic shift is underway: for the first time in decades, data centers must be fundamentally rearchitected — AI is being fused into every computing platform.” Huang added, “NVLink Fusion opens NVIDIA’s AI platform and rich ecosystem for partners to build specialized AI infrastructures.”
Industry Leaders Embrace NVLink Fusion
Several industry leaders have expressed their enthusiasm for NVLink Fusion:
- Rick Tsai, MediaTek’s vice chairman and CEO, highlighted the company’s collaboration with NVIDIA to build the next generation of AI infrastructure.
- Matt Murphy, Marvell’s chairman and CEO, noted that NVLink Fusion provides a flexible, high-performance foundation for advanced AI infrastructure.
- Johnny Shen, Alchip’s CEO, emphasized the company’s commitment to broadening NVLink Fusion’s availability through its design and manufacturing ecosystem.
Enhancing AI Factory Capabilities
NVLink Fusion enables hyperscalers to work seamlessly with NVIDIA’s partner ecosystem, integrating rack-scale solutions into data center infrastructure. The technology also allows for the creation of AI factories with millions of GPUs, using any ASIC, NVIDIA’s rack-scale systems, and the NVIDIA end-to-end networking platform. This platform delivers up to 800Gb/s of throughput and features NVIDIA ConnectX®-8 SuperNICs, NVIDIA Spectrum-X™ Ethernet, and NVIDIA Quantum-X800 InfiniBand switches.
Maximizing AI Performance
The fifth-generation NVIDIA NVLink platform, including NVIDIA GB200 NVL72 and GB300 NVL72, offers a total bandwidth of 1.8 TB/s per GPU — 14 times faster than PCIe Gen5. This technology is already being deployed by leading hyperscalers, who can accelerate their time to availability by standardizing on NVIDIA’s rack architecture with NVLink Fusion.
Software for AI Factories
NVIDIA Mission Control™, a unified operations and orchestration software platform, powers AI factories connected with NVLink Fusion. This software automates the complex management of AI data centers and workloads, enhancing every aspect of AI factory operations.
Availability
NVIDIA NVLink Fusion silicon design services and solutions are now available from MediaTek, Marvell, Alchip, Astera Labs, Synopsys, and Cadence. For more information, watch NVIDIA’s COMPUTEX keynote featuring Jensen Huang and visit NVIDIA GTC Taipei.
