We have network volumes in all s3 compatible regions. All volumes are connected to serverless endpoints.
I would expect to have some kind - any kind - of GPU availability, but no, all gpu types are now either unavailable or with low availability. It directly impacts our ability to scale.
And even worse, Runpod seem to have added a new limitation in serverless pool, where we can pick only 3 GPU types. Seriously?
Any solution in sight?
Any recommendation for an alternative serverless GPU provider?
I know about the GPU shortages. But any sane company would stop onboarding new customers before getting to that point. Maybe once runpod start loosing their big stable customers, they will do some change.