My background is in aerospace and while I’m not an expert on COPVs, I’m familiar with their design. Carbon fiber overwrap is really good for internal pressure applications or in other words, the typical pressure vessel due to carbon fiber’s very high tensile strength. Think of those videos where folks put rubber bands on watermelons to make them explode; the concept is similar to a COPV in that the rubber bands (carbon fiber) “compress” the watermelon (pressure vessel) to “contain” the pressure. COPVs can take high internal pressures of 6000 PSI (which is the same internal pressure of the ISS NORS tank used to recharge oxygen and nitrogen on the ISS and coincidentally is the same *external* pressure experienced at the depths of the Titanic. (Some helium bottles have higher internal pressure). While in aerospace applications there are some load cases where the COPV has an external pressure load on them like a helium bottle stored in a propellant tank on a SpaceX Falcon 9, those pressures are nowhere near deep sea pressures.
The reason why I don’t think COPVs are a good design for an external pressure application is the load direction; carbon fiber tow doesn’t really do well maintaining an external load. Think back to the watermelon example; apply an increased external pressure and the rubber bands really don’t help with withstanding that increased pressure.
Carbon fiber itself is very difficult to verify that you have the correct properties once they're wound onto the tank. The properties are anisotropic meaning depending on the direction of the fiber you’re gonna get different mechanical properties. Defects like delamination (when a wind unwraps) or voids between the tank wall and fiber are common if you don’t have a qualified winding operation and really really good procedures. In short, while not needing software, repeatability in manufacturing a COPV is incredibly difficult. Part of the reason SpaceX switched to metal tanks is that the mechanical properties of stainless actually increase in a cryogenic environment so there’s an added benefit.
In 2016, SpaceX’s Falcon 9 carrying AMOS-6 exploded during a routine static fire due to a buckled liner of a COPV on the 2nd stage LOX tank. The liner buckled which created a void between the liner and the fiber overwrap which then developed solid oxygen (or an ingress of LOX) that initiated an explosion due to friction. Falcon 9 had been flying for 6 years at that point so I’d consider that a pretty mature or at least an “operational” vehicle so for it to explode on the pad like that is how tricky COPVs can be.
EDIT: Spelling and clarification on delamination