Global semiconductor leaders NVIDIA and MediaTek have announced a major expansion of their strategic partnership, introducing co-developed system-on-chip (SoC) architectures tailored for edge-to-cloud artificial intelligence integration. Unveiled during an executive technology briefing, the new hardware series fuses MediaTek’s low-power mobile processor architecture with NVIDIA’s high-performance Tensor Core GPU capabilities into a single silicon footprint. Designed specifically to run parameter-heavy generative AI models directly on localized hardware devices, the chips eliminate latency bottlenecks associated with continuous cloud server requests while maintaining high computational efficiency. Industry engineers highlight that the unified architecture allows smartphone platforms, autonomous automotive units, and internet-of-things (IoT) systems to execute complex multimodal AI tasks natively on-device. Corporate leadership from both chipmakers noted that the collaboration will standardize edge-computing developer frameworks, lowering energy overhead for enterprise deployments across global mobile and automotive ecosystems. Hardware analysts view the release as a pivotal milestone that accelerates real-time on-device AI processing without relying solely on remote data center infrastructure. Multi-Billion Expansion Targets End-to-End AI Infrastructure and Edge ComputingSemiconductor powerhouses NVIDIA and MediaTek officially announced a major deepening of their long-standing partnership to build multi-generational AI computing architectures. Backed by a $3.5 billion investment by NVIDIA in convertible bonds issued by MediaTek, the collaboration combines NVIDIA’s GPU acceleration, software ecosystems, and high-speed interconnects with MediaTek’s low-power System-on-Chip (SoC) design, custom silicon expertise, and advanced connectivity technologies. The joint initiative establishes an integrated product portfolio designed to process agentic and physical AI workloads seamlessly from hyperscale data centers down to edge devices. Strategic Overview: NVIDIA – MediaTek Co-Developed Chip ArchitecturePillar / PlatformCore Technologies & Integration SpecificationsOperational ScopeStrategic Investment$3.5 Billion convertible bond financing by NVIDIALong-term capital alignment & joint R&D fundingCloud & AI InfrastructureNVIDIA NVLink Fusion™ Platform integrationCustom XPUs for hyperscaler rack-scale AI factoriesLocal AI ComputingNVIDIA RTX Spark™ & DGX Spark™ (Grace Blackwell GB10)Enterprise workstations, developer PCs, & local AI agentsAutomotive & Physical AIDimensity Auto / AX C-X1 with integrated NVIDIA GPUsSoftware-defined vehicles & autonomous driving platformsInterconnect StandardHigh-density NVLink-C2C (Chip-to-Chip) packagingUltra-low latency edge-to-cloud data movementNVLink Fusion and Custom Hyperscale XPUsA centerpiece of the expanded agreement is MediaTek’s adoption of the NVIDIA NVLink Fusion platform. This architecture provides hyperscalers, cloud service providers, and frontier AI model developers with a prevalidated pathway to design custom accelerator processors (XPUs). By utilizing MediaTek’s custom silicon design alongside NVIDIA's rack-scale interconnect technology, cloud operators can rapidly deploy specialized enterprise silicon into NVIDIA NVLink-connected AI factories without re-engineering proprietary board and memory fabrics from scratch. Empowering Local PCs, Developer Supercomputers, and Autonomous VehiclesBeyond cloud infrastructure, the alliance advances co-developed processors across local computing and automotive sectors. The companies confirmed ongoing deliveries of multiple chip generations powering NVIDIA RTX Spark and DGX Spark desktop platforms. Utilizing the GB10 Grace Blackwell Superchip architecture—which couples an NVIDIA Blackwell GPU with a custom MediaTek/Grace CPU over NVLink-C2C—the chips deliver high-performance local LLM execution for AI developers and enterprise workstations. Concurrently, the partners are expanding their automotive footprint with the MediaTek Dimensity platform, embedding NVIDIA graphics and physical AI processing into next-generation software-defined vehicles.