Nvidia has made a strategic pivot in its approach to AI infrastructure by opening up its proprietary NVLink technology to competitors and partners. The announcement, made by CEO Jensen Huang at Computex 2025 on May 19, represents a significant departure from the company's traditionally closed ecosystem.
NVLink Fusion enables the integration of non-Nvidia central processing units (CPUs) and graphics processing units (GPUs) with Nvidia's products. This technology will tie chips together to significantly enhance inter-chip communication speed, which is crucial for building and deploying sophisticated AI systems. According to industry analysts, this move could further solidify Nvidia's dominance in the AI chip ecosystem, where it already commands over 80% market share in the data center AI chip sector.
"NV link fusion is so that you can build semi-custom AI infrastructure, not just semi-custom chips," Huang explained during his keynote. The technology allows AI infrastructures to combine Nvidia processors with different CPUs and application-specific integrated circuits (ASICs), giving customers the benefit of using Nvidia's infrastructure and ecosystem even when not using exclusively Nvidia chips.
The fifth-generation NVLink platform includes impressive specifications, delivering up to 1.8 TB/s of bidirectional bandwidth per GPU—14 times faster than PCIe Gen5. A single NVLink spine can transfer data at rates up to 130 TB/s, which Huang claimed exceeds the entire Internet's peak data flow.
MediaTek, Marvell, Alchip Technologies, Astera Labs, Synopsys, and Cadence are among the first chipmakers to adopt NVLink Fusion. Additionally, Fujitsu and Qualcomm Technologies plan to integrate their custom CPUs with Nvidia GPUs using this technology. Notably absent from the partnership list are Nvidia's direct competitors Broadcom, AMD, and Intel, who are members of the rival Ultra Accelerator Link (UALink) consortium.
While NVLink Fusion potentially risks lowering demand for Nvidia's own CPUs by allowing customers to use alternatives, analysts believe the added flexibility ultimately improves the competitiveness of Nvidia's GPU-based solutions against emerging architectures. As Ray Wang, a semiconductor analyst, noted, "NVLink Fusion consolidates Nvidia as the center of next-generation AI factories—even when those systems aren't built entirely with Nvidia chips."