Trump-backed WLF is collaborating with Hong Kong-based AI platform WorldClaw; 43 of its 90 AI models are developed by Chinese entities flagged as security risks (Lawrence Delevingne/Reuters)

A new Hong Kong-based company, Antimatter, is capitalizing on the growing popularity of Chinese open-weight models to challenge the dominance of US AI infrastructure providers like CoreWeave. As the global demand for AI continues to surge, businesses are seeking alternatives to US tech giants, with many opting for closed frontier AI models that offer potential cost savings and greater data control.

Antimatter, founded in May, is a collaborative effort between Data Factory, a US energy provider, PoliCloud, a French modular data centre operator, and HiveNet, a Swiss AI inference platform. The company's goal is to provide specialized cloud infrastructure tailored to AI workloads running on high-performance GPUs, with a focus on the neo-cloud sector.

Antimatter expects to see substantial growth in the coming years, with revenue projected to reach €1 billion by 2028. The company is currently raising €300 million in equity financing and has already secured a significant deal with a French accounting firm, which reduced its AI costs by 40% by deploying Antimatter's infrastructure.

By the end of the year, Antimatter aims to deploy 100 Policloud data centres, housing over 40,000 GPUs, and plans to expand to 1,000 Policlouds with more than 400,000 GPUs by 2030. Antimatter's research and development team is actively testing chips from various Chinese suppliers and working to optimize them for different large language models.