AMD will be in trouble soon IMO. They were starting to make some headway with laptop GPUs (in the mid-tier), but that seems to be reversing this year. I don't know what the issue is exactly, but they've always struggled with drivers. Nvidia toys with them as usual; they just make their GPUs good enough to beat AMD (I'm guessing they have good spies), but unlike Intel's recent past, they are always prepared to up the ante in a hurry if AMD makes a significant advance. The RTX 4000 GPUs are a good example. Nvidia could have given us a huge generational boost, but they are holding back and banking. The Vram, bandwidth and shaders are a big reduction vs 3000 for all but the very top card (which is mega $$$). They are heavily leveraging the smaller node, frequency boost, and DLSS3 to still deliver a nice fps/$ improvement anyway. Very calculated.
I do kinda hate Intel and Nvidia, but think they will do well vs the competition, and put AMD in a tough place soon.
Part of AMD's strength is also their weakness; they're on record stating that in order to rapidly innovate in CPUs, they do run two teams focused on building upon previous generations and lightly competing against each other, and sometimes that includes throwing out some of the old code and basically redoing the code up to get those improvements. IIRC, it's the same for their GPU division. And it's also partly why every generation from AMD has teething issues for the first few months (except Zen+, which was just a straight refinement of Zen 1).
At the same time, AMD is still suffering from growth pains (needing to strategically spend their money where it's needed, vs just going all-out, and right now that market is the very lucrative enterprise/corporate sector) and long-held myths among the crowd that bias the average person away from AMD products, to say nothing of rivals manipulating deals and making it hard for AMD to be offered on the same brand platforms, often relegated to a secondary, slightly inferior one, esp. in laptops.
On the other hand, AMD's purchase of Xilinx could help them further rise with the right strategies, as Xilinx IP would let AMD quickly offer more counters in old entrenched spaces such as networking, aerospace, cellular/tel/comm, industrial, and automotive, given the pervasiveness of Xilinx IP within those sectors, while also helping to boost AMD's own products. I recall AMD teasing the possibility of FPGAs added to their GPUs and CPUs as a way to add more flexibility in things such as AI learning or a reprogrammable element set up for each client. In addition, replacing a bunch of 3rd party IP used in their products with Xilinx options could actually save them money, and allow them to field more custom solutions since more of it is in-house.
In the near term, AMD is likely to continue suffering in the consumer space while they focus on the enterprise and prosumer space, but their goal is taking over the much more lucrative markets that Intel has been trying to hold (and as of last year, was still losing ground to AMD in). With Intel now dropping yet another area that they mismanaged (networking switches), AMD could easily move into that market with leveraged Xilinx IP. They even mentioned sometime last year that they were looking into entering the networking sector via Xilinx, and we might one day see AMD NICs and switches in the near future.