Nvidia to invest $3.5 billion in MediaTek to retain cloud-infrastructure lead

Nvidia announced a $3.5 billion investment in Taiwanese chipmaker MediaTek. Under the agreement MediaTek will adopt Nvidia technology to design custom AI chips for large cloud providers, enabling those chips to connect directly to data-center servers built on Nvidia’s architecture. The move comes as Amazon, Google, Microsoft, OpenAI and Anthropic accelerate efforts to develop their own silicon and reduce reliance on Nvidia GPU. The deal lets Nvidia forgo a share of the custom-silicon market while preserving its role as the backbone of data-center infrastructure.

“Nvidia is an AI infrastructure company,” said Dion Harris, senior manager for HPC and hyperscaler infrastructure at Nvidia, in an interview with reporters. He noted that Nvidia expanded beyond pure compute chips years ago. The partnership illustrates the circular nature of Nvidia’s recent investment strategy: the chip giant channels capital to companies that ultimately feed back into its ecosystem. MediaTek will gain access to Nvidia’s NVLink Fusion stack, which includes NVLink—the technology that lets any chip, even non-Nvidia ones, communicate with other chips at high speed.

Last week Nvidia disclosed a similar collaboration with Amazon Web Services (AWS), though without a direct cash investment. AWS will deploy an additional 2 million Nvidia GPUs across its infrastructure and will also integrate NVLink Fusion. For MediaTek, the partnership allows continued development of custom ASICs designed by cloud operators or AI labs, leveraging “the proven technology stack and ecosystem of Nvidia for large-scale deployment,” according to the company. Harris added that every cloud provider and model builder ships Nvidia’s platform in some form, so MediaTek can offer customers a standardized rack-scale infrastructure.

MediaTek has not publicly disclosed its AI-chip customer base, but it is clear that Nvidia aims to help the firm grow that business as long as demand persists. In June the company said ASIC deals for data centers are expected to generate $2 billion in revenue by 2026, with a view to capturing a larger market share in subsequent years. MediaTek’s expertise in custom chips for smartphones, smart homes, automotive and wireless communications complements Nvidia’s direction.

The firms will also continue joint work on DGX Spark, the compact desktop system for AI developers, and on RTX Spark, Nvidia’s push to bring its AI technology to consumer PCs. In the automotive arena they will “develop platforms for software-defined, AI-driven vehicles in the physical AI era,” the press release said. MediaTek’s automotive platforms will rely on Nvidia’s RTX graphics for smart-driver cells and will operate alongside Nvidia Drive AGX, the in-vehicle computing platform for autonomous driving. “AI is changing every computing platform, from the world’s largest AI factories to PC and cars,” said Jensen Huang, founder and CEO of Nvidia. “Together we are building platforms that bring Nvidia-accelerated computing to new markets and give customers the freedom to create isolated AI systems at massive scale.”