NVIDIA is investing approximately $3.5 billion in Taiwanese company MediaTek in a deal that gives the latter access to NVIDIA technologies for designing custom AI chips that can run directly inside data centers built on NVIDIA platforms.
The agreement’s significance is not limited to the funding. Under the partnership, MediaTek will adopt the NVLink Fusion ecosystem, which includes NVLink technology to enable different chips, including chips not manufactured by NVIDIA, to exchange data at high speed within data center infrastructure. This allows MediaTek to offer custom chips to AI companies and cloud service providers while keeping them compatible with the NVIDIA-led operating infrastructure.
A Strategy to Remain at the Heart of Infrastructure
The deal comes as major companies such as Amazon, Google, Microsoft, OpenAI, and Anthropic work to develop their own chips to reduce reliance on NVIDIA’s graphics processing units. Rather than fully resisting this trend, NVIDIA is enabling companies to build custom chips that can be used alongside its technologies inside data centers.
Dion Harris, NVIDIA’s senior director of high-performance computing and AI infrastructure solutions, said that the company has for years been an AI infrastructure company, not merely a manufacturer of computing chips. This reflects a shift in competition from the performance of an individual chip to the design of an integrated ecosystem encompassing chip-to-chip connectivity, servers, racks, and software.
What Does MediaTek Get?
MediaTek will be able to develop custom chips for workloads designed by cloud companies or AI labs, benefiting from NVIDIA’s technologies for scaling within a server and across servers, in addition to rack-scale architecture. According to NVIDIA, this enables customers to run custom chips alongside NVIDIA chips within a unified platform.
MediaTek has not publicly disclosed details about its customer base in the custom AI chip sector, but it said in June that it expects its custom data center chip business to generate $2 billion in revenue during 2026, while targeting a larger share of the market in the coming years.
Expansion into Cars and Developer Devices
The partnership is not limited to data centers. The two companies will continue collaborating on DGX Spark, a small desktop computer aimed at developers, and have also extended their cooperation on RTX Spark as part of NVIDIA’s efforts to bring its technologies to consumer AI computers.
MediaTek and NVIDIA will also develop platforms for AI-powered vehicles whose functions can be defined through software. MediaTek’s automotive platforms rely on NVIDIA RTX graphics for smart cockpits and work with the NVIDIA Drive AGX platform for autonomous-driving workloads.
Why Does This News Matter?
The deal shows that the expansion of custom chips does not necessarily mean NVIDIA is leaving the value chain. The company is attempting to make its technologies a shared connectivity and operating layer even when customers choose chips designed specifically for them. In practice, this could give cloud providers and AI labs greater flexibility in building hardware suited to their workloads while retaining compatibility with NVIDIA’s architecture.
However, the source provides no details about the actual customers, the chips MediaTek will produce, or the timeline for turning the partnership into published products. Therefore, the investment’s ultimate commercial and technical impact remains uncertain, although its strategic direction is clear: expanding reliance on NVIDIA’s full ecosystem, rather than on GPUs alone.