Ask HN: Are any startups working on Neural Architecture Search?

It seems like while there are a lot of research papers coming out about automated architecture searches, including DARTS and NAS, there aren't many products or companies focused on an offering. Is the compute bottleneck really the only reason this is the case?

I feel that automated architecture search is probably the most useful way to democratize machine learning, and can likely be packaged as an API that Software Engineers can easily understand. Am I wrong on this?

I'm aware of Darwin and Google's AutoML, specifically. Anyone else?