Cloud environments should be perfect for running Optimization and OR algorithms, with resource auto-scaling, GPUs and TPUs, etc...
Optimization and OR algorithms are just as important as ML algorithms, or even more so, for various corporate analytics and planning tasks.
Cloud environments are already running a ton of optimization as part of the tools they use for training ML models, or for the various Auto-ML frameworks.
So clearly there's a market for Optimization APIs on the various Cloud platforms, and there should be no technical issues with running them, since several optimization routines are part of ML training frameworks.
So why are there no APIs for solving Optimization problems on Cloud platforms?
The only thing I could find so far is one Mixed Integer Programming Option on Google AI Hub and it is an advanced experiment that's not available to the general public. Nothing on AWS, nothing on Azure.
This is very strange? Is there are reason for this?