Moreover, if such solutions work today, will they continue to work as CPU dies shrink even further in future generations? Perhaps case design will have to change to the point where the entire case acts as giant heatsink.
There is probably an upper limit to how much power a consumer CPU can draw mostly for practical real world reasons. The first main thing I can think of is the limitation of a standard household circuit: 120 V x 15 A = 1800 W. I'll get to the second one later.
With a monitor and typical accessories (like a set of powered computer speakers + subwoofer, maybe an external drive or USB hub), the typical desktop PC -- even the enthusiast grade builds -- are probably have a soft limit about 1500 or 1600 watts peak draw.
Proper airflow appears to be more effective than case material choice. Don't forget Lian Li originally focused on all aluminum computer cases (back in the Nineties) which many believed would help dissipate heat. A flat panel doesn't scale very well which is why typical desktop PC heatsinks are designed with spikes or ridges to increase surface area. And if you were using the exterior case panels, you'd still need a way to transfer the heat from the source to those panels.
The radiator is really just an extension of this large surface area for cooling. You're just using water (or other liquid coolant) to move the heat from the source (silicon chip) to a place where the heat can be exhausted fairly easily because water has great thermal capacity.
Remember that all of these cooling solutions are compromises. Thermal performance, ease of installation, acoustics, size, ease of construction, durability, materials, maintenance requirements, cost, etc.
Could there be 400-450W PPT CPUs one day? Possibly. Could they be cooled? Sure, why not? GPU chips already put out that PPT and most of them are air cooled at this point. Can it be done quietly? Maybe. GPU cooling systems are constrained to certain dimensions. We're seeing larger and larger AIB coolers to handle increased GPU heat production. In some of the high-end graphics card products, there are hybrid AIO liquid/fan cooling systems.
There's an upper limit to what Joe Consumer will find as acceptable fan noise for a consumer electronics product. Hell, some people gripe about fan noise from their video game consoles, it's not just about enthusiast PCs.
And now the second reason: from a total cost of ownership perspective, I'm not sure how many people really want a 1500 W PC running for a long time. It's not just the component cost or how much power a residential circuit can handle: there's also the electricity charges to consider. For sure, there will always be enthusiasts who would want such a machine but I would expect them to be in a very small minority of even enthusiasts.
There aren't many CPU workloads that will max out a typical gaming enthusiast's CPU. Most games are still heavily reliant on IPC/single threaded performance. I look at the performance analyses at DSOGaming and usually peak gaming CPU performance is achieved with 6 cores/12 threads or occasionally 8 cores/16 threads.
There are always unusual edge cases and I'm sure someone here will say "I use Linux on a 64-core CPU to do blah blah blah" which is probably something that 0.001% of people on this planet do but they think is common enough to mention.