NVIDIA and MediaTek are expanding their long-standing collaboration across AI infrastructure, local AI computing and automotive platforms, bringing together NVIDIA’s accelerated computing and AI technologies with MediaTek’s expertise in custom silicon, high-performance computing and power-efficient system-on-chip design.
As part of the expanded partnership, MediaTek will adopt the NVIDIA NVLink Fusion™ platform, giving hyperscalers, cloud service providers and developers of frontier AI models a prevalidated route for developing custom XPUs and connecting them to NVIDIA NVLink™ based, rack-scale AI infrastructure.
NVIDIA has also invested $3.5 billion in convertible bonds issued by MediaTek, further strengthening the relationship between the two companies as demand for customized and scalable AI computing infrastructure continues to grow.
The expanded collaboration between NVIDIA and MediaTek covers three major areas: AI infrastructure, local AI computing and automotive systems.
MediaTek joins NVIDIA NVLink Fusion ecosystem
A major focus of the collaboration is custom AI infrastructure.
MediaTek will use the NVLink Fusion platform as a design foundation for customers developing custom AI accelerators. The approach is intended to allow these platforms to evolve alongside future NVIDIA architectures while reducing some of the complexity associated with building production-ready AI systems.
The NVLink Fusion platform provides a prebuilt, prequalified and system-validated foundation for multi-die XPU development. It brings together key technologies spanning silicon, advanced packaging, connectivity, memory and rack-scale infrastructure.
At the chip level, the platform includes the NVIDIA NVLink Fusion Chiplet, which connects custom XPUs to the NVIDIA NVLink scale-up fabric through NVIDIA Photonics or electrical interconnects.
It also incorporates NVIDIA NVLink-C2C, providing high-bandwidth and energy-efficient connectivity between XPUs, NVIDIA Rosa CPUs and other compatible processors.
Another component is NVIDIA NVHBM, which enables customized memory configurations aimed at improving bandwidth and energy efficiency while allowing more silicon area to be dedicated to compute.
Reducing the complexity of custom AI accelerators
Developing a custom AI accelerator is only one part of building a rack-scale AI system.
Production deployment requires the integration and qualification of multiple technologies, including multi-die architectures, advanced packaging, high-speed SerDes, high-bandwidth memory, I/O and scale-up networking. These elements must also work together reliably at the rack level and support the manufacturing and supply chain requirements of large-scale deployments.
The NVIDIA and MediaTek approach is designed to reduce the amount of engineering required around the custom XPU.
Instead of developing and qualifying every surrounding component independently, customers can concentrate on the compute architecture that differentiates their systems while relying on NVIDIA and MediaTek for key areas such as NVLink connectivity, memory architecture, packaging, manufacturing and rack-scale technologies.
Customers will be able to work with MediaTek to tailor connectivity, memory, packaging, performance and power characteristics according to their workloads and infrastructure requirements.
MediaTek’s participation in the broader NVIDIA NVLink Fusion ecosystem also provides a route for hyperscalers, cloud providers and frontier model developers to connect custom XPUs with NVIDIA AI infrastructure.
Through NVLink Fusion, partners can make use of NVIDIA’s scale-up and scale-out technologies and ecosystem, along with the MGX™ rack-scale architecture. The companies say this can reduce development complexity, improve performance and accelerate the deployment of semi-custom AI factories.
Taking AI computing closer to users
The collaboration also extends to local AI computing, where the companies are targeting the growing demand for generative and agentic AI capabilities on PCs and other edge systems.
The two companies are combining NVIDIA’s accelerated computing technologies with MediaTek’s experience in designing high-performance, power-efficient SoCs.
MediaTek previously collaborated with NVIDIA on the GB10 Grace Blackwell Superchip, which powers NVIDIA DGX Spark. The platform combines an NVIDIA Blackwell GPU with a Grace CPU, connected through NVLink-C2C, to deliver substantial AI computing capabilities in a compact system designed for local and edge environments.
The companies are now extending this work through NVIDIA RTX Spark, targeting the next generation of AI-enabled consumer PCs.
The broader objective is to bring more capable AI computing closer to users rather than relying entirely on centralized cloud infrastructure. This could support applications involving generative AI, AI agents and other compute-intensive workloads that benefit from local processing.
Expanding AI into software-defined vehicles
Automotive technology is the third major pillar of the expanded collaboration.
NVIDIA and MediaTek are working across multiple generations of automotive platforms to support the development of increasingly intelligent, software-defined vehicles.
The partnership combines MediaTek’s automotive SoC expertise with NVIDIA’s accelerated computing, AI, graphics and software platforms.
MediaTek’s Dimensity Auto platforms integrate NVIDIA technologies to provide advanced AI capabilities and NVIDIA RTX™ graphics for intelligent vehicle cockpits. These platforms can also operate alongside NVIDIA DRIVE AGX™.
The companies plan to build on this architecture through future generations of automotive platforms as vehicles increasingly rely on AI for cockpit experiences, intelligent functions and software-defined features.
Expanding Customized AI Computing
The expanded NVIDIA and MediaTek collaboration comes as AI computing architectures become increasingly diverse.
Large-scale AI systems are demanding more compute performance, memory bandwidth and high-speed interconnectivity, while cloud providers and other major infrastructure operators are looking for greater flexibility to customize systems around specific workloads.
At the same time, AI is moving beyond traditional data centers into PCs, workstations, edge systems and vehicles.
The companies’ expanded partnership addresses these different requirements through a combination of custom silicon, accelerated computing, high-speed interconnects, advanced packaging and software.
For AI infrastructure customers, the NVLink Fusion platform provides a pathway to develop differentiated XPUs without having to build the complete surrounding rack-scale architecture from the ground up.
For local computing, the collaboration is bringing NVIDIA accelerated computing capabilities together with MediaTek’s power-efficient SoC expertise.
In automotive, the two companies are extending the same strategy toward increasingly intelligent and software-defined vehicles.
Together, these efforts position NVIDIA and MediaTek to address AI computing across the data center, edge and automotive markets as the industry moves toward more customized and distributed AI architectures.
Leadership Comments
“AI is transforming every computing platform — from the world’s largest AI factories to the PC and the car,” said Jensen Huang, founder and CEO of NVIDIA. “MediaTek is one of the world’s great semiconductor companies, with exceptional expertise in system-on-chip design, connectivity, leading performance and power efficiency. Together, we’re building platforms that bring NVIDIA accelerated computing to new markets and give customers the freedom to create differentiated AI systems at enormous scale.”
“MediaTek and NVIDIA share a vision for making advanced AI computing pervasive across the technology landscape,” said Rick Tsai, vice chairman and CEO of MediaTek. “NVIDIA’s investment strengthens a collaboration that spans cloud AI infrastructure, local AI computing and automotive in the era of physical AI. By combining NVIDIA’s leadership in accelerated computing and AI software ecosystem with MediaTek’s expertise in a diverse AI technology portfolio from edge to cloud, and our leadership position in custom silicon, we can accelerate innovation for our customers.”
