RE: LeoThread 2025-05-09 13:17

AI requires compute to first train a model, and then to run inference on it (like when you ask it a question, or generate an image).

Nosana is a DePIN project that aggregates underutilized consumer-grade NVIDIA GPUs worldwide, so that you can run inference on an AI model.

Nosana is 5x-10x cheaper than AWS, because they don't have the overhead costs of centralized corporations.

Their NOS token is required to pay for compute, and is what the GPU owners earn for lending their equipment to the network.

Check out Nosana's presentation (linked below) at Solana Build Station if you want to know more!