NVIDIA unveils NVLink Fusion to drive next-gen AI factories

NVIDIA has introduced NVLink Fusion, a new silicon technology intended to enable industries to build semi-custom AI infrastructure in collaboration with partners leveraging the NVIDIA NVLink ecosystem.

Partner companies including MediaTek, Marvell, Alchip Technologies, Astera Labs, Synopsys and Cadence are among the first organisations to adopt NVLink Fusion. This approach is designed to support the creation of custom silicon and to meet increasingly complex demands for model training and agentic AI inference. NVLink Fusion further allows integration of CPUs from Fujitsu and Qualcomm Technologies with NVIDIA GPUs to develop high-performance AI factories.

Jensen Huang, Founder and Chief Executive Officer of NVIDIA, said, "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. NVLink Fusion opens NVIDIA's AI platform and rich ecosystem for partners to build specialized AI infrastructures."

NVLink Fusion provides cloud service providers with the capability to scale out AI factories to incorporate millions of GPUs. The platform supports any ASIC, as well as NVIDIA's rack-scale systems and its end-to-end networking platform, which offers up to 800Gb/s of throughput and incorporates ConnectX-8 SuperNICs, Spectrum-X Ethernet and Quantum-X800 InfiniBand switches. Co-packaged optical components are expected to be available soon.

MediaTek, Marvell, Alchip, Astera Labs, Synopsys and Cadence are developing custom AI compute solutions deployable with NVLink Fusion as part of NVIDIA's approach to integrating specialised hardware into data centre infrastructure.

Rick Tsai, Vice Chairman and Chief Executive Officer of MediaTek, stated, "By leveraging our world-class ASIC design services and deep expertise in high-speed interconnects, MediaTek is collaborating with NVIDIA to build the next generation of AI infrastructure. Our collaboration, which began in the automotive segment, now extends even further, enabling us to deliver scalable, efficient and flexible technologies that address the rapidly evolving needs of cloud-scale AI."

Matt Murphy, Chairman and Chief Executive Officer of Marvell, commented, "Marvell is collaborating with NVIDIA to redefine what's possible for AI factory integration. Marvell custom silicon with NVLink Fusion gives customers a flexible, high-performance foundation to build advanced AI infrastructure — delivering the bandwidth, reliability and agility required for the next generation of trillion-parameter AI models."

Johnny Shen, Chief Executive Officer of Alchip, said, "Alchip is supporting adoption of NVLink Fusion by broadening its availability through a design and manufacturing ecosystem, encompassing advanced processes and proven packaging and supported by the ASIC industry's most flexible engagement. It's our contribution to ensuring that the next generation of AI models can be trained and deployed efficiently to meet the demands of tomorrow's intelligent applications."

Jitendra Mohan, Chief Executive Officer of Astera Labs, expressed, "Building on our rich history of close collaboration with NVIDIA, we are thrilled to add purpose-built connectivity solutions to address the NVLink Fusion ecosystem. Low-latency and high-bandwidth scale-up interconnects with native support for memory semantics is critical for maximizing AI server utilization and performance. By expanding our scale-up connectivity portfolio with NVLink solutions, we are providing more optionality with faster time to market for our hyperscaler and enterprise AI customers."

Sassine Ghazi, President and Chief Executive Officer of Synopsys, remarked, "Data centers are transforming into AI factories, and Synopsys' industry-leading AI chip design solutions and standards-based interface IP are mission-critical enablers. Our support for NVIDIA NVLink Fusion reflects our commitment to fostering an open and scalable ecosystem for next-generation AI and high-performance computing."

Boyd Phelps, Senior Vice President and General Manager of the Silicon Solutions Group at Cadence, added, "HPC and AI workload demands are unique and evolving rapidly, and hyperscalers architecting the most advanced custom AI systems rely on Cadence to deliver enabling technology from data centers to the edge. Our comprehensive IP portfolio, including design IP, chiplet infrastructure, subsystems and other critical IP, complements the NVIDIA NVLink ecosystem, accelerating the delivery of AI factories that are powerful, energy-efficient and production-ready at scale."

NVLink Fusion also enables companies such as Fujitsu and Qualcomm Technologies to connect their custom CPUs with NVIDIA GPUs in a rack-scale architecture, aimed at enhancing AI system performance.

Vivek Mahajan, Chief Technology Officer at Fujitsu, said, "Combining Fujitsu's advanced CPU technology with NVIDIA's full-stack AI infrastructure delivers new levels of performance. Fujitsu's next-generation processor, FUJITSU-MONAKA, is a 2-nanometer, Arm-based CPU aiming to achieve extreme power efficiency. Directly connecting our technologies to NVIDIA's architecture marks a monumental step forward in our vision to drive the evolution of AI through world-leading computing technology — paving the way for a new class of scalable, sovereign and sustainable AI systems."

Cristiano Amon, President and Chief Executive Officer of Qualcomm Technologies, commented, "Qualcomm Technologies' advanced custom CPU technology with NVIDIA's full-stack AI platform brings powerful, efficient intelligence to data center infrastructure. With the ability to connect our custom processors to NVIDIA's rack-scale architecture, we're advancing our vision of high-performance, energy-efficient computing to the data center."

NVLink's fifth-generation platform includes NVIDIA GB200 NVL72 and GB300 NVL72 racks, providing up to 1.8 TB/s per GPU — which the company notes is fourteen times the speed of PCIe Gen5. Several hyperscale companies are already deploying full-rack solutions using NVLink and are able to reduce the time to deployment by standardising on the NVIDIA rack architecture with NVLink Fusion.

Supporting this infrastructure is NVIDIA Mission Control, a unified software platform that automates the operations and orchestration of AI data centres and workloads. Mission Control manages aspects ranging from configuration to validation and orchestration of critical workloads, with an aim to enable faster deployment of advanced AI models.

NVLink Fusion silicon design services and solutions are now available from MediaTek, Marvell, Alchip, Astera Labs, Synopsys and Cadence.