Alright time to add my 2 cents.
Its not that simple. By which i mean you cant make any definite statements here because so many factors come into play. Everything from Gravity, Water salinity, Buckling forces, Vessel purpose and more. Its not a trivial thing to answer.
For example, if you classify as Submarine as a sphere that can sink to the bottom of a 30km deep ocean, you can build that. Make a sphere 10 meters in radius with a small cavity in the middle and you are good to go. But that's not a very useful DSV now is it ?
Water Density
Water is not incompressible. It is just very hard to compress. But that little bit of compression is responsible for 99% of your problems. For instance, why do Submarines implode supersonically ? After all, if your pressure vessel fails the Water should just "fall" in right ? Well no, because water compresses it becomes a spring. When you have a pressure vessel, what it is actually resisting is the spring force of all the water around it. The instance the pressure vessel fails, this spring force is released, which happens at the speed of sound in Water. ~1500 m/s. Which is why implosions are instant.
But ok, how do you model fluid density at depth ? This equation;
$$p_d = \frac{1}{1-\left(\frac{pgh}{B}\right)}p$$
gives us a good starting point. It computes the density of Water or really any medium at depth. The B is the bulk modulus, or how hard it is to compress something. Now there is an immediate problem with this, to figure out the pressure at depth, we have to run this equation for many different depths and add up the total. Which still does not take into account Temperature or salt contents. So while this equation is a good start, its not perfect.
So here is a graph i made. You can see that the Pressure up to a depth of 100km varies widely between the classical $P = phg$ and more accurate approach. To the point where at the maximum model depth the pressure suggested by the classical model is ~25% smaller than what it would more realistically be. Of course, this has an exponential nature to it. If the depth is 200 kilometers the projected pressure by the classical model is ~$2GPa$, where as in actuality it would be closer to $5.5 GPa$
This does of course leave out phase changes, which will occur. My main point here is that pressure is not a simple subject. And Water would never under natural circumstances turn into any other phase. Only in very weird situations could this even begin to happen. The diagram L.Dutch showed is correct but imo a bit misleading in presentation. Because they leave out the density differences. For Instance Ice II is less dense than liquid water and would just shoot up the moment it was made, to quickly dissolve again. Afaik, we don't know the exact densities for other forms of Ice or the bulk modulus. So it is a bit hard to be super accurate here. But from the Diagram, where Ice VII turns into Ice X after a 100 fold pressure increase, we can guess the bulk modulus is pretty high.
Oceans cannot achieve these pressures
Ill skim over this a bit, but these pressures are entirely theoretical for one main reason. Gravity. Oceans are just valleys. And while they can get quiet deep, they cant get massively deeper as the largest mountain on a planet is. On Earth, you cant get a 20km deep ocean because the ground couldn't support that depth. Same reason why you cant get a 20km tall Skyscraper.
If you want deeper oceans, you need lower gravity. But since gravity is a term in both models, if you make it lower the pressure decreases. I talked about this in a different comment but the TLDR is that you cant get exotic forms of Water in really any natural configuration. Either the oceans cant physically get this deep, or the temperature is to high or the gravity is to low.
Generally speaking, take the Gravity of your world, divide it by Earths gravity and that fraction is how high your tallest mountain and deep your deepest ocean should be.
For example, lets look at Ganymede. Ganymede's surface gravity is 1.4 m/s² While Earth is obviously 9.8 m/s². The fraction of this (Earth/Ganymede) is 7. The deepest point on Earth is ~12 kilometers deep, so this would suggest an Ocean depth for Ganymede of like 80-90 kilometers. Which is around the right ballpark. We estimate the depth to be ~100 kilometers. Note, the Pressure at the bottom of Ganymede deepest ocean is going to be about the same as in the Challenger Deep. Obviously with some variations because of the different temperature and salinity etc.
So, to answer the question, we can build DSVs for basically any Earth like ocean that will work. Duo the likely certainty that it really does not get more extreme than on Earth. Sure the oceans might be deeper which can lead to issues where it takes a week to get down there. But those are no physical limitations. If you took the Trieste and dropped it into Ganymede's oceans, it would probably still fail because the descend rate is to low but conceptually it could survive on the bottom.
Other Planets
So this is all for Oceans on planets. On Gas giants / dropping a sub into one you will get ripped apart by the winds long before reaching any solids. The Temperature will also become an issue. Titanium is pretty tough, as long as it is cold. Titanium is also surprisingly moldable once it gets hot.
Sadly we just dont have any data on this. We dont know what the temperature deep inside a gas giant looks like.
