Immersion cooling is one of those technologies that's easy to be impressed by on paper, but I don't see many hosting providers talking about real-world deployments outside of AI, HPC, and some crypto environments.
I'm curious where people think the tipping point is. Is it driven by power density (say, 50-100kW+ racks), electricity costs, sustainability goals, or simply customer demand? And for anyone who's actually deployed it, has the reduction in cooling costs offset the added complexity of maintenance and hardware servicing?
It would be interesting to hear from providers who have run it in production for a year or more rather than just lab testing.