Antimatter Launches Distributed Neocloud for AI Inference

Antimatter has launched what it calls the world's first vertically integrated neocloud platform for AI inference, announced in a press release. The company combines energy sourcing, modular micro data centers, and distributed cloud software into a single infrastructure system designed to deliver faster and more cost-efficient AI processing.

Antimatter is the result of merging three firms: Datafactory, Policloud, and Hivenet. The company has secured more than one gigawatt of power capacity through grid agreements and site reservations across the United States, Europe, and the Gulf Cooperation Council region. It plans to deploy 1,000 distributed micro data centers by 2030, providing over 36 exaFLOPS of compute capacity and 400,000 GPUs.

The company has raised €300 million to fund the first 100 Policloud units by 2027, representing 40,000 GPUs and 3.6 exaFLOPS of active compute capacity. Each modular data center can be deployed in about five months, significantly shorter than traditional hyperscale timelines. Antimatter currently operates 10 sites and has a pipeline of more than 500 additional units.

Led by CEO David Gurle, Antimatter positions itself as an energy-first infrastructure provider, placing micro data centers near renewable power sources such as wind, solar, and hydro. The platform’s distributed software layer connects these units into a unified cloud fabric with sub-10 millisecond latency for edge workloads and data sovereignty features for regulated industries.