What kind of server has 416 cores?

I was looking for something completely unrelated and found the gcloud m2-node-416-11776 which apparently has 416 cores and up to 12TiB of Memory.

I then did some digging and found the Azure Mv2-series with the M416ms v2 being the biggest one of them which also has 416 cores and 12 TiB of Memory.

The Google machines only say that they use skylake CPUs, the Azure ones say that the CPUs are Intel® Xeon® Platinum 8180M CPUs, which according to Intel have 56 cores.

Now, 416 cores are not easily divided into 56 core CPUs, I figured that there might be 13x 32 Core CPUs in there but that also doesn't really fit my logic.

There are some high memory instances at AWS as well, but they offer 448 CPU Cores, which more evenly divides between the hyperthreaded 56 core CPUs. Do Microsoft and Google disable some cores? And if yes, why?

So now I want to know what type of servers they use and how the CPU configuration is set up.

A big box, possibly for scale up.

Google Cloud's sole tenant page explicitly states m2-node-416-11776 is 8 sockets x 28 core Skylake. Note that Intel can ship custom CPU SKUs that aren't available at retail. However these is probably Platinum 8176.

Hypervisor overheads take up multiple CPUs full time at this scale. At least if response time is a priority at high utilization. 224 cores available minus 208 cores for guests (I prefer to count physical CPUs and not threads) is 16 cores not available for guests, about 7%. Two reserved cores per socket. Possibly conservative, although on the other hand I wouldn't want to cut that in half.

Azure M416ms v2 isn't documented as a sole tenant node, rather it is an enormous VM. I would hope paying that much rents you a similar dedicated 224 core box.