Supported by Super Micro Computer, Argentum AI landed $7.8 billion in contracts to deploy 47,000 Nvidia GB300 chips across Poland. The agreements rank among Europe’s top three largest AI GPU deployments.

What got signed and what it means

The first contract, announced on April 29, 2026, is valued at $1.5 billion and covers GPU cloud capacity equivalent to roughly 10,500 Nvidia GB300 GPUs.

The second deal is the headline-grabber. On May 15, 2026, Argentum AI signed a $2.5 billion agreement with Boosteroid and DL Invest Group to build a 300 megawatt AI data center in Bielsko-Biała, Poland. The facility is designed to house tens of thousands of next-generation GPUs, spanning Nvidia’s GB300 and Blackwell series, and has already commenced operations.

The company behind the contracts

Argentum AI operates a decentralized compute marketplace, connecting buyers who need GPU capacity with providers who have idle resources to sell. The platform uses a native utility token called AGP, an ERC-20 and SPL-compatible token with a fixed supply of one billion. AGP is used for payments, staking, and governance within the ecosystem. Rafay Systems serves as a strategic partner for orchestration, handling the infrastructure management layer that keeps workloads running across distributed compute nodes.

The company counts Goldman Sachs, J.P. Morgan, and Blackstone among its financial backers. The Poland data center, developed in partnership with DL Invest Group and Boosteroid, supports both AI model training and inference.