Technology have advanced a lot since the PS5 released. In fact, it advanced much more than PS4 to PS5.
In 2026 we can expect to see these technologies to mature. That will dramatically make the gap between PS5 and PS6 larger than we think. And these are just a few.
...
I'm big on BSPDN / Backside Power Delivery and really hope to see PS6 use it; but the main issue with it is the TSMC Foundries implementation has slipped from N2/N2P to A16 which won't go into production until end of 2026 or more likely early 2027 and Sony probably won't get first dibs, instead having to wait 12-18mths while Apple fill their pockets with iPhone silicon, not to mention the node cost being much higher (though hopefully chiplets + indirect efficiencies of bspdn on the thermal design of the console can offset the bulk of that). Thinking worst case scenario, A16 entering mass production in Q1 2017 + 18mths to mature and have viable volume available, then PS6 go into production in July 2028 for release in Nov 2028.
Samsung Foundries have S2FZ but that's just loosely pegged as "2027" and Sony haven't used Samsung's foundries since the 65/45nm PS3 Cell B.E. revisions.
Then there's Intel Foundries who are way ahead on BSPDN with their PowerVia but it'd be an odd partnership given it'd almost certainly be semicustom AMD silicon.
The wildcard choice could be using TSMC for CPU and MCD/Memory Bus chiplets with something like N2 and N5 respectively to keep costs low, then 1 or 2 GPU chiplets from Samsung or Intel on their latest nodes (where they won't have to compete with Apple for volume); then packaging it on a substrate after with infinity fanout links. They could even use Samsung or Intel temporarily until TSMC are ready with BSPDN, but then I think Intel's implementation is a bit different (and a bit better), so it'd probably complicate it further.
Personally I'd prefer 2028 as I'd not only like to see BSPDN, but I'd like to see mature chiplet/substrate tech, mature ML & RT capabilities in terms of both hardware, software and dev experience. I'd like to see game engines have time to really implement stuff such as microgeometry and/or mesh shaders; and see the SSD+I/O systems be integral to those engines. Then with PS6 it can all be scaled up dramatically. Plus of course there's just the matter of raw power.
In addition, I'll be getting a Pro and I think a reasonable lifespan for that would be 3-4yrs; given the pace of game dev plus the fact that developers will only just be getting their engines set up for current gen featuresets (rather than just scaling up from -- or tacking stuff on to -- what are effectively last gen games). I also worry that if we don't hit a baseline level of RT and ML hardware throughput going into the next gen then it may still limit at a fundamental level the degree to which those game-changing features can be adopted; and then that means another 6-8yrs of games being functionally limited at the ground level.
Finally, it sounds like "RDNA5" will be a big departure from the existing RDNA, but those first outings of GPU architectures always have teething problems, so it'd be nice to get the first revision in RDNA6 or at least have some backported improvements/features.
Given the Pro is almost certainly coming Q4 this year I think it's very unlikely that PS6 comes any earlier than Q4 2027, but I genuinely believe the entirety of the next gen will be better served with a late 2028 release.
