How is a server different from a workstation computer?

Classically, the term "workstation" was reserved for computers that contained server-grade hardware, but ran only locally-accessed software that required high performance.

Basically, a workstation was a computer that was so powerful that it could have been used as a server, but was instead used for single-user tasks -- as a "client" or "PC".

The lines have blurred somewhat, now that typical high-end consumer PCs can do most of the things a workstation was typically built for (3D design, etc). Those are now often advertised as "workstations," and the word is tossed around flagrantly, even for hardware that isn't particularly specialized. There's also the fact that high-end PCs today can also function as servers, at least in some scenarios.

Purists will still say that a bona-fide workstation should be built on hardware one normally wouldn't see in a consumer PC. Nevertheless, the fact is that "workstation" is a fairly ambiguous, highly subjective term, with relatively little technical meaning today.

So, basically, workstations classically were not supposed to be any different from servers in terms of their hardware. Workstations were supposed to be servers “on the inside”, and their only true difference from servers was in the tasks they were used for, ie. their software.

The most fundamental differences between actual servers and high-end PCs/workstations is the motherboard/processor/memory. These part in a server are built to handle concurrent tasks more smoothly versus performing singular tasks faster, and with more error correction.

The Intel “Xeon” line of CPUs are commonly referred to as server CPUs, and motherboards with multiple CPU sockets are generally reserved for servers. Their RAM memory is typically the rarer/pricier ECC type (Error Code Correcting). Servers often have more elaborate storage schemes, with redundant storage drives (in an arrangement called RAID) that provide faster access to multiple users accessing multiple things simultaneously. All of these components offer more error correction and failsafes, which make the overall system faster, more stable, and less prone to system hangs that could require reboots.

The modern workstation is merely a higher-power version of a PC. For the CPU, you might have an Intel X-series, or an AMD Ryzen Threadripper, each with 10 or more CPU cores. Rather than a special RAM type, you’d see the same type of RAM that PCs have, just perhaps slightly faster and more of it — say 32–64 GB, instead of the 8–16 you’d see in home PCs. For storage, you’d have a large NVMe (a faster type of SSD), which are again also being in PCs, but something top-of-the-line like the Samsung Pro line.

My own current computer could be considered a lower-end workstation: an Intel 9900k (8 cores overclocked to a flat 5GHz on all cores) with 32GB of RAM and a 1TB NVMe. It’s simply more powerful than most consumer PCs, which is really the only “requirement” for calling a computer a “workstation” these days.