As an interesting benefit of building these things, I've learned a number of lessons I feel are applicable to future human habitats.
The first has to do with cleaning. The way we normally deal with garbage and accumulated grime assumes we can move freely between the inside and outside of a structure. You cannot do that underwater. I solve the problem by bringing the habitat up monthly to wash and scrub the interior, to replace the litter, replenish food/water, etc.
Can't do that with a human sized habitat. You need some specialized module set aside just for compacting garbage, storing cleaning supplies, rinsing out mops/sponges and processing the resulting grey water. Docked to this module you'll need a smaller secondary one which can seal, detatch and be raised to the surface.
The idea is, everybody in the colony would deliver gathered trash from their residential module to the waste management module, have it compacted, then stored in the detatchable pod. Once a month or so that pod is un-docked, surfaced and either emptied onto a barge or towed to shore for the trash to be removed and dealt with.
Stored biowaste is simpler as it can be pumped out of the colony's central septic tank into the tank of the same ship which carries the garbage module to shore, same way you empty the septic tank of a sailboat or RV.
The important thing is that the specialized waste management module be sealed off from the rest of the colony by a moisture and scent barrier. Something as simple as flimsy but airtight sliding doors, with a powerful air filtration system treating the air inside, like the one you have in your bathroom. This is because in an enclosed environment with recirculated atmosphere, even a small degree of stench buildup is a lot more noticeable.
Such barriers to the free movement of air were introduced after Sealab II to prevent humidity from spreading from the moon pool/wet porch room into the living space. This solved massive recurring problems like chronic ear infections, rashes, etc. relating to the unbearably high humidity. The simple addition of a divider wall and air tight door, plus a dehumidifier in the living area made conditions vastly more comfortable for the aquanauts.
Another issue is the bubble exhaust. If ambient pressure, the main escape for bubbles will be through the moon pool. Either as loud, reverberating "burps" if it escapes the moon pool conventionally or as a quiet, neat "plume" if you manage where/how air escapes. The best way I have seen to do this is to have an outlet built into the moon pool at the exact height where you want the water level to be.
This outlet leads, through a hose, to a "bubble chimney" mounted to the side of the structure. This ensures that as air builds up inside the structure, when it pushes the water in the moon pool far enough down, it escapes through that outlet rather than through the bottom of the moon pool. Conshelf 2 had this but for some stupid reason Aquarius and many other newer habitats neglected to include one.
Next: I have for some time been meaning to get my hands on a dredge pump. I speculate that sand from the bottom could be suctioned up this way and deposited into the empty ballast tray of a habitat you mean to sink.
Securing ballast weight in this manner, "in situ" from the spot where you mean to sink the habitat would wipe out the considerable expenses involved in buying tons and tons of pig iron or lead ballast weight, moving it over land to the coast, ferrying it out by ship, then shoveling it by hand into the ballast tray of the habitat. This is, again, an unexpectedly huge percentage of the cost of underwater habitats in general.
Lastly, modern habitats like Aquarius are designed to operate also as decompression chambers, so aquanauts can decompress inside at the end of their mission and surface without need of a diving bell and deckside decompression chamber. This is brilliant but costly. An ambient pressure habitat can be as simple as a welded steel box, since it only needs to be strong enough to hold down the trapped air.
A 1atm habitat you intend to decompress in, however, must be a cylindrical or spherical pressure vessel of considerably greater hull thickness depending on depth. It also constrains how large windows can be.
I propose an alternative: Instead of the entire habitat being a decompression chamber, make 90% of it cheaper ambient pressure only living space, but integrate a single decompression module just large enough for the crew to sleep in with stacked bunks. This would be about one quarter of Aquarius' internal volume.
Ideally, the other side of this module is a docking collar for the submersible they will travel to the surface with so they can move from the decompressed 1atm environment of the chamber into the 1atm environment of the sub without having to re-acclimate to the ambient pressure environment of the habitat in order to get to the moon pool, as Aquarius aquanauts must do (it's not easy on the eardrums.)
Minimizing the portion of the habitat that you can decompress in greatly decreases overall cost, and opens up interesting new possibilities. Ambient pressure habitat space is way more fault tolerant, because it doesn't need to resist any pressure differential. You don't need a precisely machined steel cylinder. You just need welded sheet metal boxes with some reinforcement against buoyancy stress.
This means you could employ underwater welders to actually build expanded habitat space onto an existing mostly-ambient habitat, while underwater. As opposed to building the complete habitat on land, towing it out (or transporting via A-frame ship) and sinking it.
The traditional method of constructing modules on land and sinking them limits you to ISS style modular expansion, one pitifully small cylinder at a time. Instead, structures could be built on-site from loads of materials lowered from the surface, closer to how traditional buildings are constructed.
You'd still build them with ballast containers but could fill them on the bottom by shoveling in sand from the sea bed. Driving pilings into the seafloor and securing it to those is also a possibility with this method.
Once the addition is complete, you'd pump air in to displace the seawater, cut a doorway into it from inside the existing dry habitat space, then set about hosing it down with fresh water (to get rid of the salt) and drying it off with towels and sponges. The dehumidifiers would also help. An Italian diving club established a small model colony consisting of four cylindrical 2 person habitats in a similar fashion.
They did not construct them on the seabed, but did allow them to fill with seawater (without electronics or furniture inside) so they would sink easily. Then secured them to ballast containers, shoveled sand into those, then pumped the habitats full of air and cleaned the salt water out of the interiors. After this, the crew ferried down loads of materials in water tight sacks to slowly build out the liveable interior like a ship in a bottle.
If we're ever to see truly large undersea habitats, they have to be ambient pressure because saturation diving is the only thing you actually need to put humans underwater for that can generate a profit or be of scientific use (facilitating sea farming or coral research for example).
But by minimizing the amount of the habitat built to the spec needed for decompression, making most of it the lower spec ambient only (like Aquarius' wet porch, a simple sheet metal cube) and employing the other methods listed so far, the price can be brought down to where permanent sea floor settlements in support of science or mariculture become economically attractive.