In 2019 every PE investor in London told me the same thing: everything is going to public cloud. Your private cloud is dead. Your margins will erode.
I was running Sentia, a Cloud MSP in Europe. We ran workloads in our own multi-tenant private cloud and supported public cloud too. The investors were convinced we were on the wrong side of history.
We sold Sentia to Accenture a few years later. By then it was clear that local European clouds weren't going anywhere. The fact we could support any cloud was a key reason Accenture wanted to acquire us.
I'm not writing this to say I told you so. I'm writing it because the same pattern is repeating.
Everyone assumes AI means sending everything to the big model providers. That the future is API calls to Silicon Valley. I think that's wrong.
If you want to run LLMs yourself, you want them in your own environment. It's most likely cheaper. If you want to use the biggest frontier models but need a layer that monitors, logs, and filters out sensitive data before it leaves? That layer belongs on-prem. Not in someone else's cloud.
With Aimable we're building exactly for this reality. A trusted AI platform where organisations work with frontier models and local models through one platform. It protects sensitive data, logs everything, enforces your ground rules. And it gives AI the right knowledge for the context you're working in, so responses are actually grounded.
The person working? Could be a human, could be an agent. Both are held against the same rules.
Even if geography and sovereignty aren't your concern, there's a simpler question worth sitting with:
Do you feel comfortable sharing all your sensitive data with a company that can only get better by understanding more of the world?