Not 100% sure what you're asking, but chip development like this, based on what we've seen from AMD and Nvidia, takes 3 years. So if you wanted to replace T239 with something for a 2025 release, you would have needed to start in 2022.
But let's pretend for a second that development is instantaneous, that I can package up a SOC from any existing technologies right this very second. The differences between the chip you'd make today right this very second and the leaked T239 specs would be little more than a footnote. And even those differences might not be good as they'd represent potentially increased costs for very little (or even no) improvements. Let's write them out even, so we can see what we're talking about.
We could replace T239's modified Ampere with stock Ada. This is almost no change at all, the architectures are so similar. You could get the larger Optical Flow Accelerator, but because of limitations of being a handheld, Frame Generation is not likely to ever work. The larger OFA might be a literal waste of space, so this could effectively be a downgrade.
You could update from LPDDR5 to LPDDR5X. It's possible this has already happened. 5X would decrease battery life for the sake of extra bandwidth. Benchmarking suggests that extra bandwidth might be much of a performance win. This is a shrug.
ARM processors have actually been progressing nicely. You could replace the A78C cores with A715s. This would break Switch 1 backwards compat for what ARM advertises as a whopping 10% increase in performance. You could use A710s instead. This would keep BC, and ARM advertises the same 10% performance increase, but benchmarks suggest the opposite with an actual drop in performance.
You could put your chip on TSMC N3. Going by existing products on that process node, Nintendo could probably do that and launch this thing at $750. It's a non-starter.
The technology to make T239 significantly better does not yet exist and is unlikely to till late 2027, at which point it will still be prohibitively expensive. That's just the reality on the ground.
No one is making games with PS6 specs in mind while they're still desperate to figure out how to make games with PS5 specs profitable.
Specs matter. Features matter more. Switch wasn't much more powerful than the 360, but games didn't look like the 360, and they certainly didn't only come from the 360 era. Why? Because Maxwell (the GPU arch) and the rest of the console (solid state storage, the RAM pool, the CPU design) were all much more modern.
If AMD delivers a truly innovative GPU design that causes Xbox Series 2 to completely alter how DirectX works, the Microsoft will push that version of DirectX into Windows, which will force Intel and Nvidia to redesign their GPU pipelines to be DirectX14 (or whatever) compatible, which will alter how PC software works, which will start to leave T239 behind.
Tellingly, Microsoft seems to be driving the updates and forcing AMD to keep up. And what MS is pushing are things like RT and ML, forcing AMD to keep up with Nvidia's design. Right now it looks like T239's feature set will have a long life.
Truth
No clue. Custom SOCs for mass market devices that last more than a couple years are pretty rare, and we only have knowledge on T239's tapeout because the Lapsus hack gave us data we could cross-check elsewhere. That was an unprecedented event.