It is like comparing a banana to a chair.
> An analysis of the mammoth data centre proposed by celebrity investor Kevin O’Leary suggests it would release the equivalent of 23 atom bombs of heat waste into the boreal forest of northwestern Alberta every day.
a really obvious electricity in, waste heat out is aluminum smelting in quebec. at ~12GWh a day for the biggest ones, this would come out at a similar scale, but make clear that the data center industry in alberta produces far less waste heat than the aluminum industry in quebec, and we should start looking to shutter all those smelters for wasting too much power
bringing up nuclear bombs when the heat output of a nuclear bomb isnt what people are familiar with is intentionally misleading. it also elicits a "cool, this thing can survive 20 nuclear blasts a day?" response which is pretty obviously not intended and isnt true
I stand corrected.
A day has 86,400 seconds, so that’s about 11 orders of magnitude.
That makes this a fairly meaningless comparison.
Dimensional error.
Do they mean 18 GWh per day = 64.8 TJ / day = 750 MW ?
Or is this thing really going to use twice as much power as the largest nuclear power plant that currently exists?
Correction: 16.77 GW
A. my math was off
1 Little Boy = 63 terajoules https://en.wikipedia.org/wiki/Little_Boy
23 Little Boys / 24 hours = 16.77 Gigawatts
B. https://www.theenergymix.com/wp-content/uploads/2026/06/Wond...
"As recently described, the project would draw approximately 9 gigawatts (GW) of power, [...] Power would come from on-site, behind-the-meter natural-gas generation. At realistic generating efficiency, supplying 9 GW of electricity means burning fuel — and therefore rejecting heat — at a continuous rate on the order of 16 to 18 GW"
New York City uses 1 hiroshima bomb-energy every 3 hours.
A forest fire can release many hiroshimas of energy every day, and canada has more than a few of those each year.
Or at least economic measurements in terms of money/profits brought by companies/entities who used it?
I mean of course it's going to depend on the field and how it's being used, but there could be a comparison between how much money was spent making AI overall, and how much actual measurable good things it brought.
And I don't mean the amount of customers paying for AI, I mean an actual demonstration that AI is worth doing.