I wonder how that compares to the cost of booking time on a system of similar class, assuming said systems can be booked?

> (1 759.00 / 41.82) * 880 * 1.60 = 59222.1903

And for almost 60k per hour, you will be number 1! How long till AWS as the capacity to top the chart?

If it's a one-off computation then a few thousand for a relatively hassle-free setup is bearable for many use cases. I've seen many clusters running idle just because customers initially bought more metal than they could handle. Peak load != continuous load.

Considering how much you would have to pay for the power alone, that doesn't seem unfair.

Using extra large high cpu instances is actually cheaper per compute unit. You just lose io and memory.