My server has 48 cores, currently openlane only uses a maximum of 1 core to run, resulting in the running process being very slow. Is there any way to take full advantage of CPU cores to run openlane?

Most tasks are not parallelizable. The only one that is in OpenLane 1 is detailed routing for which you can set the variable

ROUTING_CORES

.

Even a 10mm2 design tapeout takes 1 hour to complete on a laptop. So, single core performance is pretty much satisfying.

I want to synthesize a very large AI design, SRAM size is 2048x32. Therefore, running 1 core to synthesize the design will take a long time. So is there any way to increase synthesis performance?. I have 48 cores and I really want to use them all for this process

The Synthesis tool, Yosys, doesn't support multi-threading, so them's the breaks