I love the idea of underground infrastructure. Unfortunately, after playing around with it a little I will probably not use it for the same reason I'm not using orbital habitats - it's too expensive.
First, let's compere it to orbital habitats. A single habitat for one million people costs 6800 BP. However, it is mobile and can be built on heavily developed planet, allowing relatively rapid construction and deployment. UI on the other hand costs 10BP per unit and you need 300 units for one million people, for a total of 3000BPs, or almost half as much as a space habitat. And of course, all of this has to be constructed on site, which means that the development is slow. And to top it all off, the UI behaves like colonization cost 3 infrastructure, which means full 20% of the population is needed for life support. IIRC you don't need any people for life support in the case of space habitats. And while this isn't such a big difference at the beginning, for very developed objects (100+ million people) this means you have much, much fewer available workers in the case of UI. I haven't done the math, but I wouldn't be surprised if, as far as manning installations go, orbital habitats were actually more cost effective.
And now let's compere it to normal infrastructure. First, it's five times cheaper. Second, it can be transported. Third it can be created automatically, without the need for resources or industrial capacity (that can be spent for more important items). Fourth, the infrastructure requirements can be lowered by partial or full terraforming. Fifth the infrastructure requirements can be lowered by researching "colonization cost reduction technology", which does not affect underground infrastructure.
Overall, the new UI is great when setting up new campaign, but as an actual mechanic within the game it's not really all that practical. What I'd like to see is some of the following (not necessary all of them, but some):
Lower cost (preferably 3-5 BP)
Self replication when there isn't enough UI (same as with normal infrastructure) or lack of population growth when current habitat is filled to capacity (same as with orbital habitats)
Possibility of lowering colonization cost by researching tech, either currently existing one or by adding new tech tree.
Prefabricating UI on developed planets and moving it to target body (like PDCs)
Making the UI equivalent to colonization cost 2, or preferably 1.5 instead of col. cost 3 (so more workers are available)