GEMM isn't compute bound in ML in practice. If you do naive GEMM based attention, then you will have to write the output matrix into HBM and in the worst case you might even have to reload the output from HBM!
So what is done in practice is an algorithm that doesn't calculate the same result, but is imperceptibly close to doing classic attention, namely flash attention. Flash attention lets you fuse the kernel so that you can multiply against the V matrix and therefore write the condensed output to HBM. As an additional benefit you also go from quadratic memory usage to linear memory usage. But here is the problem: Your SRAM is limited and even flash attention is still O(n^2) in compute. This means if you tile your K and V cache into j and k tiles. You will have to load j*k times from memory. Meanwhile compute tends to consume very little silicon area. So you end up in a situation where you have excessive compute vs your SRAM. In the compute > SRAM regime, doubling SRAM size also doubles performance. You're memory bound again for super long contexts.
Now let's assume the opposite. Your compute resources are improperly sized with regards to your SRAM, you have too much SRAM but not enough compute resources e.g. a CPU. You will be compute bound with regard to a linear factor vs your SRAM, but always memory bound vs main memory. You could add the matrix cores to the CPU and the problem would disappear in thin air.