Rafay is a platform for orchestrating AI infrastructure and managing Kubernetes. Organisations such as telecom operators, so-called neoclouds and large enterprises can use it to turn GPU capacity into self-service offerings with governance, multi-tenancy and cost controls. The offering includes Kubernetes clusters, virtual machines, bare-metal GPU servers, SLURM clusters and token-metered model inference, with observability and automated remediation.
Rafay Systems, Inc. is based in Sunnyvale, California, and is therefore a US vendor; no parent company is known and the company is venture-funded. According to its website the platform can run in-region, in private or air-gapped environments, including on-premises and hybrid. Where the vendor hosts the management layer is not stated, and the privacy policy refers to servers in the United States without mentioning the GDPR. European organisations building sovereign AI clouds should therefore have the management location, processor terms and the effect of US law explicitly settled.