The article says “propylene glycol closed loop” so my most sane interpretation is that whatever cools the actual machines is that. Then whatever cools that propylene glycol should be the tap water. Kinda how some nukes are cooled. Inner water loop with that goes into the reactor, cools the core, gets irradiated, then that’s cooled by outer water loop that cools the inner loop water. That way radiation is kept from contaminating the water coming out of the reactor. The heat is tranferred from one loop into the other via heat exchanger.

If the DC is keeps the water and recycles it, they gotta reject the heat from it somehow, that would mean heat pumps. Maybe? That’s more energy and money tho.

If there’s a huge nerd with too much time among us, they could compute how much heat 100000 liters of cold water can absorb till they heat to “hot water.” Then cross reference that against the expected heat produced by the initial DC chip capacity and check whether the two numbers are within the realm of comparable per day. 🤭