Nvidia is reinforcing its position at the center of AI infrastructure on two fronts: expanding its ecosystem through a major chip partnership and demonstrating a growing advantage in the systems that support its GPUs.
The company announced a $3.5 billion investment in Taiwanese chipmaker MediaTek, under which MediaTek will adopt Nvidia’s technology to help design custom chips for AI companies and hyperscalers that integrate directly into Nvidia-based data centers. The deal comes as major cloud providers, including Amazon, Google, Microsoft, OpenAI, and Anthropic, increasingly develop their own custom silicon to reduce dependence on Nvidia GPUs. Dion Harris, Nvidia’s senior director of HPC and AI hyperscaler infrastructure solutions, described Nvidia as an AI infrastructure company that expanded beyond core computing chips years ago. The partnership grants MediaTek access to Nvidia’s NVLink Fusion ecosystem, enabling faster communication between chips regardless of origin, and follows a similar arrangement last week with AWS, which will deploy two million additional Nvidia GPUs. MediaTek, which expects its custom data-center chip business to generate $2 billion in 2026, will also continue collaborating with Nvidia on developer hardware and automotive AI platforms. Nvidia founder and CEO Jensen Huang said the companies are building platforms that extend Nvidia’s accelerated computing into new markets.
In other Nvidia news, the company’s latest earnings revealed the extent of its advantage beyond GPUs alone, as AI compute scales toward gigawatt levels and data orchestration becomes increasingly complex. The company’s Vera Rubin architecture pairs its Rubin GPU with supporting systems, including the Vera CPU, designed to manage data movement efficiently. Nvidia’s Jason Hardy said the Vera CPU has delivered roughly threefold improvements in memory operations, addressing bottlenecks as data centers scale. Competitors, including OpenAI with its Jalapeño chip, are pursuing similar efficiency gains through different architectural approaches, signaling a new competitive layer beyond raw processing power, one where Nvidia currently holds a substantial lead.
