About rentagpu

rentagpu is a decentralized GPU marketplace that lets people rent out idle compute and run AI workloads directly on real machines without relying on a centralized cloud platform.

Instead of routing workloads through company-owned infrastructure, rentagpu connects customers directly to partner nodes over secure networking using technologies like Tailscale. Jobs run inside isolated Docker containers with restricted privileges and automatic cleanup, helping keep workloads separated and secure for both customers and hardware providers.

We support heterogeneous compute across NVIDIA, AMD, and Apple Silicon systems, making it possible to use everything from high-end workstation GPUs to spare home-lab hardware. Partners can securely contribute idle compute capacity, while customers get flexible access to GPUs without long-term contracts or enterprise cloud complexity.

rentagpu is designed for researchers, developers, hobbyists, and small teams running AI inference, fine-tuning, automation, agent workloads, and general GPU compute tasks. Our goal is to make distributed compute simpler, more accessible, and less dependent on centralized cloud providers.