I don't bother debating the issue anymore. Because it doesn't matter when the correct technical time of death for the platform occurred. AM4 in the practical sense, is more alive in retail, than any Intel platform. Hell, the 3600 is not only still in the top 10 on Amazon U.S. it has moved up to 7th place. A CPU from the last decade, is outselling everything Intel makes, at the biggest U.S. retailer, in 2026. The 2 best-selling boards are both B550.
The investment in AM4 tech is fully amortized. It's all gravy now. For both AMD and its board partners. There is obviously more profit to be made selling the parts, so they will keep making them. Any plans they had to fully move on to AM5, have been foiled by DDR5 pricing. How it started: Just a few months ago, prognosticators were saying AM4 was finally cooked because DDR4 was almost the same price as DDR5. I showed a lot of U.S.shoppers were still picking it for new builds.
How it's going: DDR5 pricing exploded, and AM4 sales have benefitted from it. Focusing on upgraders, by talking about telemetry for how many are still on older Zen, is myopic. Shoppers are choosing it for new builds. Those shoppers are not going to pay $450 for a 5800X3D. They picked the platform because it is less expensive and gets the job done. Give them back a $250 5700X3D and they will snatch them up.
Aussie Steve just revisited vanilla Zen 3. It is aging well. 24H2 stretched its legs. My experience is RT can kneecap it in some games, in the most demanding areas. But I doubt most gamers building with it in 2026 are champing at the bit, to crank up RT.
Side note: That Zen 3 revisit demonstrated another important point. Concerns about needing core and thread counts above 6/12 for gaming, over 5 years later, turned out to be a nothingburger.