The PS1 was built on a 1.2Āµm process. The tech was over 5 years old at the time.
The PS2 was built on a 250nm process. That tech was 3 years old at the time, and offered a 4.8x density win over the previous machine.
The PS3 was built on a 90nm process. The tech was 3 years old at the time, and t this point, the "size" of the process was already a marketing term, not reflecting actual transistor density. Sony put 6 times the number of transistors in there, but the chip was twice as large, reflecting a doubling of cost.
The PS4 was built on a 28nm process. The tech was only 2 years old at the time, and only offered a 3x improvement over the 90nm process, but again, Sony just made it larger, and lost more money on the hardware to get to the expected performance leap.
The PS5 was built on a 7nm process. The tech was only 18 months old at the time. This was, again, closer to a 3x improvement, and Sony simply made the APU larger, again at additional cost. Microsoft, believing that node shrinks would come so slow, and at such high cost, that a node shrink wouldn't be possible after launch (and thus no cheaper Slim model) made the Series S.
2nm process. which is not yet available. only offers a 45% improvement over 7nm. If Sony is willing to take an even bigger loss on PS6 than they take on PS5, they'll still need a 3x transistor density improvement to get there. There is no node on the horizon that will offer that over 7nm and if it does come, it will not be more than a year old as PS6's 2028 launch window.