It's the same old story: The problem with APUs is that currently they sell for a comparable prices as CPUs while manufacturing costs are significantly higher. The only way you get better (as in bigger) iGPUs is by paying more for it on top of comparable CPU prices. AMD so far doesn't deem it realistic that the audience willing to do that exists in a sizeable enough amount.
The balance used to be that you get less CPU and in exchange, more GPU. This was true on Picasso, they made it more "equal" on Renoir, by giving more cpu and reducing the gpu and now they are prioritizing cpu over gpu, in a APU.
If you try to put a HUGE igp along with a HUGE cpu its no wonder it will be expensive. With that in mind i think it was a mistake to move to 8c/16t APU, if you need that much cpu you probably want a dgpu.
Maybe it made sense on AM4 because Intel sold a lot of i5, i7 and i9 with igps, that probably will never get a dgpu (this is far more common that some people realise), but in AM5 when all CPU have a igp the priority should be GPU first in APUs.
All that said, i dont belive the problem is cost, the problem is that we have zero competition, AMD is sitting on the same position as Intel on the 6th gen right now, at least in APUs. When they need it, they came out with Picasso in 2018, giving a huge APU (for that time) at a very competitive price, hell they even used the same die for $50 Athlons 200GE.
Now in 2024 they are going to try to sell what in a normal situation would be the Athlon 3000G die of 2024, as a Ryzen 5 7500G ? Something is very very wrong here.
I can understand inflation and that prices cant be the same as 5 year ago, but here we are talking about selling what should be a entry level product as midrange and the IGP Phoenix 2 has is just not up to the task.