NVIDIA and MediaTek expand collaboration across AI infrastructure, PCs, and automotive
NVIDIA and MediaTek announced an expanded partnership on August 31, 2026, covering custom AI accelerators, local AI computing, and software-defined vehicles. NVIDIA has also invested $3.5 billion in convertible bonds issued by MediaTek, deepening a relationship that now spans three major product categories.
The centerpiece of the deal is MediaTek's adoption of the NVIDIA NVLink Fusion platform. This gives hyperscalers, cloud service providers, and frontier model developers a prevalidated path to build custom XPUs and connect them to NVIDIA's rack-scale AI factory infrastructure.
"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."
Three areas of collaboration
The companies are working together in three specific domains:
- AI infrastructure: MediaTek will integrate with NVIDIA's NVLink Fusion ecosystem, enabling customers to design custom AI accelerators that fit into NVIDIA rack-scale systems.
- Local AI computing: The two firms will continue developing multiple generations of NVIDIA RTX Spark and DGX Spark PC chips, pairing NVIDIA GPUs with MediaTek SoCs for consumer PCs, developer supercomputers, and workstations.
- Automotive: MediaTek Dimensity Auto platforms will keep integrating NVIDIA technologies for AI-powered vehicle cockpits and work alongside NVIDIA DRIVE AGX systems.
Rick Tsai, vice chairman and CEO of MediaTek, said the collaboration "spans cloud AI infrastructure, local AI computing and automotive in the era of physical AI." He pointed to the combination of NVIDIA's accelerated computing with MediaTek's custom silicon expertise as the foundation for customer innovation.
What the NVLink Fusion platform provides
For product teams building custom accelerators, the NVLink Fusion platform offers a prebuilt and system-prevalidated foundation for multi-die XPU development. It bundles three key technologies: the NVLink Fusion chiplet for scale-up fabric connectivity, NVLink-C2C for high-bandwidth processor interconnects, and NVHBM for customized memory integration.
The pitch is straightforward: instead of engineering every surrounding element from scratch, customers focus on their differentiated compute while NVIDIA and MediaTek handle connectivity, memory architecture, packaging, and manufacturing. MediaTek is part of the broader NVLink Fusion ecosystem, which lets partners use NVIDIA's scale-up and scale-out stack alongside the MGX rack-scale architecture.
This approach matters for teams working on generative AI and LLM infrastructure, where custom silicon decisions increasingly shape deployment timelines and total cost.
Why this matters for product development
For product development professionals, this announcement signals a shift in how custom AI hardware gets built. The NVLink Fusion platform reduces the chip-to-rack engineering burden that typically consumes 18-24 months of a custom accelerator program. If your roadmap includes custom XPUs, evaluating MediaTek as a design partner alongside NVIDIA's interconnect stack is now a concrete option rather than a theoretical one.
The automotive and PC collaboration lines also matter: the RTX Spark and DGX Spark chips point to a generation of edge devices with tighter GPU-SoC integration, while Dimensity Auto platforms suggest NVIDIA's AI for product development tooling will reach vehicle cockpits through MediaTek silicon. Teams planning AI features for PCs, workstations, or vehicles should track how these integrated platforms change the hardware assumptions underneath their software.
Your membership also unlocks:
