Digital Event Horizon
NVIDIA's NVLink Fusion technology has revolutionized the way AI factories are constructed by combining custom silicon with proven, mature infrastructure. This enables hyperscalers and AI-native companies to build flexible AI factories that combine specialization with scale. With its high-performance and low-latency networking, NVLink Fusion also manages risk with infrastructure standardization, making it an ideal solution for modern AI workloads. Learn more about how NVLink Fusion is changing the game for AI factory construction.
NVLink Fusion enables hyperscalers and AI-native companies to build flexible AI factories that combine specialization with scale. NVLink Fusion addresses the challenges of building a custom AI factory by providing a unified architecture for XPU- and GPU-based systems. NVLink Fusion offers high-performance and low-latency networking, with an end-to-end latency that is 3x lower than alternative solutions. NVLink Fusion manages risk with infrastructure standardization, allowing operators to deploy their rack-level solution with the NVIDIA GPU before decoupling XPU development. NVLink Fusion includes NVIDIA NVLink-C2C for connecting XPUs to NVIDIA Vera CPUs or other ecosystem CPUs, delivering up to 6x the energy efficiency of a PCIe interface. NVLink Fusion aligns with the NVIDIA DSX reference architecture for AI factories, providing a digital twin and open reference design for gigawatt-scale AI factories.
NVIDIA's latest innovation, NVLink Fusion, has revolutionized the way AI factories are constructed. By combining custom silicon with proven, mature infrastructure, this technology enables hyperscalers and AI-native companies to build flexible AI factories that combine specialization with scale. This is particularly important for generating intelligence at scale, as AI factories run continuously and their economics are defined by delivered output.
The challenges of building a custom AI factory are significant. Teams developing custom XPUs often underestimate the effort and complexity of turning XPU innovation into data center deployment. This includes integrating high-speed CPU and scale-up interfaces, sourcing and validating a scale-up network solution, designing compute and switch trays, designing and validating a rack architecture, including cooling and power, integrating security and storage, and managing a complex supplier ecosystem.
To address these challenges, NVIDIA has developed NVLink Fusion, a unified architecture that enables XPU- and GPU-based systems to share rack footprints, networking, cooling, power delivery and management systems. This allows operators to move forward with buildout while deferring the precise silicon mix, then reprovision capacity as workload demand, silicon supply and business priorities change.
NVLink Fusion is supported by an ecosystem designed for rapid development, integration and deployment, spanning ASIC design, CPU, and IP and optical interconnect partners. This means that teams can focus on targeted innovation while using proven solutions for the rest.
One of the key benefits of NVLink Fusion is its ability to deliver high-performance and low-latency networking. Sixth-generation NVLink provides leading high-bandwidth, low-latency networking across a 72-XPU domain, with an end-to-end latency that is 3x lower than alternative solutions based on off-the-shelf Ethernet. The packet rate is also 10x higher, making it an ideal solution for modern workloads such as running trillion-parameter models, mixture-of-experts architectures and agentic AI.
Another key benefit of NVLink Fusion is its ability to manage risk with infrastructure standardization. AI factory planning doesn't wait for silicon, and power procurement, facility design, cooling, rack layout and network architecture begin long before the final accelerator mix is available. By using a unified architecture, operators can avoid schedule risk and deploy their rack-level solution with the NVIDIA GPU, then decouple the development of their XPU and put it at a different pace.
NVLink Fusion also includes NVIDIA NVLink-C2C for connecting XPUs to NVIDIA Vera CPUs or other ecosystem CPUs, delivering up to 6x the energy efficiency of a PCIe interface. This helps remove barriers between control and compute for agentic systems.
In addition, NVLink Fusion aligns with the NVIDIA DSX reference architecture for AI factories: codesigning buildings, power, cooling, compute and networking. The NVIDIA Omniverse DSX AI Factory Blueprint provides a digital twin and open reference design for gigawatt-scale AI factories, enabling partners to model facilities and technology together before deployment.
"The value of the NVLink Fusion program is that customers can deploy their rack-level solution with the NVIDIA GPU, and then they can decouple the development of their XPU and put it at a different pace," said Vince Hu, corporate senior vice president and general manager of the data center and computing business group at MediaTek.
"NVLink Fusion gives customers the ability to choose the CPU architecture, the performance level, the software capabilities that best meet their needs for the workloads that they care about," said Tim Wilson, vice president and general manager of data center silicon engineering at Intel.
With NVLink Fusion, XPUs can now meet a world-class AI platform, enabling hyperscalers and AI-native companies to build unified, semi-custom AI factories that combine the strengths of many builders into infrastructure no one company could build alone.
Related Information:
https://www.digitaleventhorizon.com/articles/Revolutionizing-AI-Factory-Construction-How-NVIDIAs-NVLink-Fusion-Bridges-the-Gap-Between-Custom-XPUs-and-Proven-Infrastructure-deh.shtml
https://blogs.nvidia.com/blog/nvlink-fusion-xpu-ai-factory/
Published: Mon Aug 24 14:59:44 2026 by llama3.2 3B Q4_K_M
