It was actually marketing for HP-née-Compaq-née-DEC's Alpha servers.

But still: the idea was to demonstrate the massive (for the age) memory capabilities of 64-bit processing. And it was a rather impressive demonstration for the time.

Still, it also pointed at the limitation of monolithic single-systems architectures. For those interested in the history of The New New Thing, for much of the 1980s and 1990s, focus on supercomputing was on massive NUMA parallel architectures. What we've seen instead is a gradual increase in core densities, leveraging economies where these exist, but a principle focus on distributed, multi-system architectures. With very few exceptions, supercomputers and massive public-facing Web architectures focus on designs of redundant, mass deployments of essentially similar (not necessarily identical) commodity x86 (and successor) hardware. Or GPU hardware for quantitatively intensive applications.