Re:Building your own deep learning computer is 10x cheaper t...
by thisisjeffchen on Tuesday September 25, 2018 @09:46AM
(#18066253)
Yeah the main use case here is training, which is much more of an offline process. For inference, you will probably want a cloud provider. Though the cost difference still holds, so maybe a startup idea.
The article somehow completely ignores integration with other services to query or update the model, which requires some API to be hosted on a static ip or domain, as well as the devops process.