Nvidia has invested $3.5 billion in MediaTek to deepen their partnership, integrating Nvidia’s advanced GPU, networking, and AI technologies with MediaTek’s SoCs to develop powerful AI computing solutions like the DGX Spark desktop supercomputer. This collaboration aims to accelerate AI adoption across various industries by providing scalable, high-performance, and flexible AI hardware and software platforms, supported by a long-term commitment to joint innovation.
Nvidia and MediaTek have announced a significant expansion of their partnership, highlighted by Nvidia’s $3.5 billion investment in MediaTek through the purchase of convertible bonds. This collaboration builds on their previous work integrating Nvidia’s NVLink chip-to-chip interface with MediaTek’s SoCs, enabling seamless connectivity between MediaTek’s processors and Nvidia GPUs. A key outcome of this partnership is the development of the DGX Spark, a powerful desktop AI supercomputer capable of delivering one petaflop of performance, which will serve as the foundation for new AI-driven PCs from Microsoft and others.
The partnership extends beyond hardware integration to include Nvidia’s advanced networking technologies, such as NVLink Fusion and Spectrum X switches, which will be incorporated into MediaTek’s systems. This integration allows MediaTek to leverage Nvidia’s comprehensive AI factory platform, encompassing GPUs, networking, and CPUs, to provide customers with a scalable, high-performance AI computing stack. Both companies emphasize the importance of speed to market and flexibility, enabling customers to build custom accelerators without reinventing the entire architecture, thus accelerating AI adoption across various sectors including cloud, data centers, robotics, and autonomous vehicles.
MediaTek CEO Rick Tsai highlighted the strategic use of proceeds from their bond program to enhance supply chain capabilities and invest further in technology development, particularly in advanced packaging and interconnects. Despite global supply chain challenges, MediaTek benefits from its location in Taiwan, a hub for semiconductor infrastructure, and its close collaboration with Nvidia and customers. The partnership also addresses the complexity of integrating specialized AI chips by providing a standardized, end-to-end solution that simplifies networking and system architecture for customers deploying AI workloads.
Nvidia CEO Jensen Huang underscored the superiority and versatility of GPU-based architectures in accelerating the entire AI lifecycle, from data processing to inference across diverse AI models. While acknowledging the presence of specialized accelerators like XPUs in the market, Huang emphasized Nvidia’s openness to integrating these specialized chips through NVLink Fusion, enhancing the overall AI ecosystem. This approach not only strengthens Nvidia’s position across cloud, on-premises, and edge computing but also fosters collaboration with partners and customers seeking differentiated AI solutions.
Finally, the partnership represents a long-term, decade-spanning commitment grounded in mutual trust and shared technological vision. Nvidia’s investment formalizes and deepens this relationship, enabling joint innovation in AI hardware and software ecosystems. By combining MediaTek’s leadership in SoC manufacturing with Nvidia’s expertise in GPUs, networking, and AI infrastructure, the collaboration aims to accelerate AI adoption globally, delivering scalable, high-performance solutions that meet the rapidly evolving demands of the AI market.