Nvidia Secures $3.5B MediaTek Deal and Explores $250B OpenAI Infrastructure Guarantee

Semiconductor giant Nvidia is reportedly in discussions with OpenAI to provide funding guarantees for a data center located in Ohio. [1] The $250 billion deal would assist the ChatGPT owner in leasing a 10-gigawatt project currently being built by SoftBank subsidiary SB Energy in Piketon, Ohio. [1]

This project is expected to deliver up to 800 megawatts of electricity by 2028, enough to power 640,000 homes, and could cost more than $500 billion including the chips that would go inside it. [1] While the $250 billion guarantee covers the center's lease, it does not include the chips within the facility, which are valued at an additional $350 billion. [1] The power supply for the project stems from a separate $33 billion United States government agreement with Japan for a natural gas plant. [1]

Alongside its data center talks, Nvidia announced it has invested $3.5 billion in convertible bonds issued by Taiwanese chipmaker MediaTek. [4][5] As part of this expanded agreement, MediaTek will adopt Nvidia's NVLink Fusion platform to design custom chips for AI companies and cloud providers that can be plugged directly into Nvidia-based data centers. [4][5]

MediaTek and Nvidia share a vision for making advanced artificial intelligence computing pervasive, according to MediaTek vice chairman and CEO Rick Tsai. [5] The collaboration spans three major areas: AI infrastructure, local AI computing involving multiple generations of Nvidia RTX Spark and DGX Spark PC chips, and platforms for AI-powered, software-defined vehicles. [5]

A growing number of AI companies and major cloud providers, including Amazon, Google, Microsoft, OpenAI, and Anthropic, are investing in building their own chips to reduce reliance on Nvidia's GPUs. [4] Deals like the MediaTek partnership allow Nvidia to cede ground to custom silicon while maintaining its lead as the dominant data center scaffolding. [4]

Nvidia's financing efforts over the past few years have often involved investing in companies that flow back into its own ecosystem. [4] According to Dion Harris, Nvidia's senior director of HPC and AI hyperscaler infrastructure solutions, Nvidia expanded beyond pure computing chips years ago to become an AI infrastructure company. [4]

Nvidia’s multi-pronged investment strategy illustrates its shift from a pure GPU provider to the foundational layer of global AI infrastructure. By moving to guarantee OpenAI's massive 10-gigawatt facility in Ohio and funding MediaTek’s custom silicon designs via NVLink, Nvidia is embedding its architecture into the very facilities and alternative chips designed to compete with it. This defensive maneuvering directly addresses the trend of hyperscalers like Google and Amazon developing bespoke hardware, ensuring that even non-Nvidia processors will rely on Nvidia networking and data center scaffolding. What the sources don't address: How regulators will respond to Nvidia leveraging its balance sheet to finance the core infrastructure of the AI models heavily dependent on its technology.