Fly.io has announced that users can now do AI workloads on the edge with their GPUs. The company aims to enable developers to do “hilarious things with AI” by providing industrial-grade muscle. With Fly.io GPUs, users can attach an Nvidia A100 to their applications and harness the power of CUDA. These GPUs allow for tasks such as speech recognition, text segmentation, article summarization, and image synthesis at high speeds. Fly.io’s edge inference capabilities ensure that users receive instant results, even in locations far from the datacenter. The company provides GPUs in datacenters worldwide and offers pricing options that allow users to pay only when the GPUs are in use.
https://fly.io/blog/fly-io-has-gpus-now/