Details
- NVIDIA and MediaTek announced an expanded partnership in August 2026 spanning AI infrastructure, local AI computing and automotive platforms, anchored by NVIDIA’s $3.5 billion investment in MediaTek-issued convertible bonds.
- The collaboration targets hyperscalers, cloud providers and frontier model developers, combining NVIDIA’s accelerated computing, AI and graphics stack with MediaTek’s custom silicon, high-performance SoC design, advanced packaging and connectivity capabilities.
- MediaTek will adopt the NVIDIA NVLink Fusion platform as a prevalidated foundation for custom XPUs, integrating NVLink Fusion chiplets, NVLink-C2C interconnects and NVHBM memory to plug third-party accelerators into NVIDIA NVLink-connected, MGX-based rack-scale AI factories.
- For local AI computing, the companies will co-develop multiple generations of NVIDIA RTX Spark and DGX Spark PC chips, including GB10 Grace Blackwell-based systems that pair Blackwell GPUs with Grace CPUs and MediaTek SoCs to bring supercomputer-class AI capabilities closer to edge PCs and workstations.
- In automotive, MediaTek’s Dimensity Auto platforms will continue to integrate NVIDIA accelerated computing and RTX graphics, interoperating with NVIDIA DRIVE AGX to create a scalable architecture for AI-powered, software-defined vehicles in the emerging “physical AI” era.
Impact
This deal strengthens NVIDIA’s strategy to keep custom data-center silicon and AI accelerators within its NVLink ecosystem, countering hyperscaler moves toward in-house chips while broadening edge and automotive reach. Over the next 12–24 months, it is likely to accelerate semi-custom AI factory deployments, deepen AI PC and on-device inference roadmaps, and push auto OEMs toward unified, GPU-centric compute platforms for software-defined, AI-first vehicles.
