Is that comparable to a cray (maybe those are obsolete now) supercomputer as big as a city block cubed?
No, that's a standard 42U network rack, would fit in your average home closet. (Although you'd have to have some serious HVAC revisions and deliver 120 amps of power to the closet).
The servers & racks we build (highly available failover clusters) are based on Intel supercomputing hardware (we're an Intel Gold partner), but repurposed for smaller scale high performance clustering and highly available VM hosting for Fortune 500's. We have 64 physical XEON cores in each 2U server, with 512GB of RAM (upgradable to 2TB per 2U). Our limiting factor on rack density is (once again) not physical space but power delivery; I can only fit 4 full size PDU's in a rack, each handling 30 amps (120 total). Factoring in fault tolerance on the PDU's I can only have a maximum of 90 amps delivered to a rack - have to allow for a PDU failure without downtime. Each 2U compute cluster draws 11.93 amps at peak use, so we can only fit 5 of them in a rack and not exceed the power ratings (with assorted switches, etc).
We fill the rest of the rack with SAN's and $15,000 10gbps switches.
If you didn't understand any of that, your vision of a supercomputer (cray) that filled a warehouse 5 years ago can now fit in to a rack that's only about 2x4x6'. Our HPC tests on single racks exceed supercomputer ratings from less than a decade ago. If I had the money, filled a room with them, and could deliver the power, and keep it all cool and dry, we'd give modern supercomputer clusters a run for their money. Same exact technology in play.
EDIT: Although the Chinese are currently kicking our asses in HPC clusters...