Regarding Rocket-Lake, ouch that's really underwhelming! New microcode might help things a bit but it won't turn the thing around.
The first thing Intel has to fix with Alder Lake is to get it out in time.
I'm not even talking about 10nm. All their 14nm desktop CPUs since Coffee-Lake have slipped at least a quarter. All of the Server models as well, starting from and including Skylake-X (that's why EPYC was compared to Broadwell-X on release). And that's with all of them being essentially the same architecture with minor tweaks.
Intel is unbelievably consistent in their inability to execute with every single CPU launchsince 2017 (most of them essentially the same arch on the same node). Apple, for comparison, managed to release a new architecture on time every year since then, often on new node, going from A10 to A14, essentially doubling their SpecINT2006 score from 36.8 to 63.34.
Even if they manage to get a good hardware stepping of Alder Lake in time, they still need to get their microcode and OS Big-Little support in order. Based on past performance I'm not holding my breath they'll get it done in within 9 months after Rocket Lake.
I wouldn't mind at all if they prove me wrong, but so far 2021 looks like smooth saling for AMD, if they only manage to get more supply out.