NVIDIA global AI infrastructure expansion

NVIDIA is significantly expanding its global AI infrastructure through various strategic partnerships. In a collaboration with Amazon Web Services (AWS), the companies plan to deploy 2 million additional NVIDIA GPUs across AWS’s global infrastructure during 2027-2028. This initiative includes the integration of NVIDIA Vera CPU-based infrastructure and co-engineered solutions for robotics and physical AI.

Simultaneously, NVIDIA is targeting expansion in Australia by partnering with several data center operators and cloud providers. This project aims to build a network with up to 2-gigawatt capacity by 2027 to bolster Australian AI sovereignty. The collaboration involves multiple firms providing land, power, and cooling, while NVIDIA supplies its DSX AI factory platform, GPUs, and networking solutions. One partner, Sharon AI, is reportedly deploying up to 68,000 NVIDIA GPUs as part of this regional expansion.

Recent developments show NVIDIA is expanding its Australian footprint by partnering with eight infrastructure providers, including Firmus, NEXTDC, and AirTrunk, to reach the 2GW computing capacity goal. Beyond infrastructure, NVIDIA has invested $3.5 billion in MediaTek to integrate NVLink Fusion technology, enabling third-party chips to connect to its AI ecosystem.

The company is also scaling its software and specialized applications. Its DRIVE platform is currently powering robotaxi fleets for Uber, Lyft, and Waymo. In the media sector, a collaboration with NDI is developing AI-powered workflows for real-time multilingual broadcasting using NIM microservices, which reportedly reduced video bandwidth requirements eightfold in internal tests. Additionally, the NIM 2.0.12 update has demonstrated a 2.5x throughput increase for the Nemotron 3 Ultra system.