I don't think this is like Bulldozer. Bulldozer was a whole new architecture on a whole new node and it performed worse than its predecessor in single -threaded tasks while also using more power. Moreover AMD was pretty much stuck with that node for 5 years. (AMD did eventually migrate from 32nm to 28nm, but the 28nm node was GPU-centric and killed AMD's desktop CPU line.) Bulldozer was so inefficient that AMD ported their old architecture to 32nm for the mobile market. Piledriver, Bulldozer's successor, pretty much fixed every issue, but Bulldozer competed against Sandy Bridge, Intel's wildly successful microarchitecture, and Piledriver competed against Ivy Bridge built on Intel's equally wildly successful 22nm node.
It is true that Meteor Lake is only coming to mobile (probably because it can't clock high enough for desktop) and it's a regression in single-threaded performance and it's late to market at a time when Intel needs a win. But it's efficient and has a great iGPU. It's not using a new CPU microarchitecture, so the regression in CPU performance is on the Intel 4 node and not on the design. (The E core microarchitecture is new but the multithreaded performance is better so I think those cores really have improved.) AMD isn't making stellar leaps; Zen 4 is good but not spectacular like what Bulldozer was against, and RDNA3 is probably why Meteor Lake's iGPU looks so good. And Intel won't be stuck on Intel 4 forever; Intel 3 and 20A are both due next year.
Bulldozer was a failure that AMD's entire future was built on. Meteor Lake is just another step down Intel's decline. I'd say it's more like Ice Lake, but at least this time Intel didn't precede it with a line that never really came to market (Cannon Lake). Ice Lake was followed by Tiger Lake which was great on mobile and Alder Lake which was great on desktop, so maybe there's hope for Arrow Lake and Lunar Lake. But the fact that Intel has been talking so much about Lunar Lake makes me think that Arrow Lake isn't looking much better than Meteor Lake.
