Remix.run Logo
▲ preg_match a day ago

The energy used by AI is, frankly, negligible. The water too. It’s a rounding error as compared to agriculture and manufacturing. The water use of all AI data centers combined doesn’t even reach the change in water usage over the past year for agriculture. Never mind the base usage, which is orders of magnitude greater.

Moreover, big picture, software is getting built anyway. It’s not like if you don’t use AI then that energy use is gone. You need devs no matter what.

▲scott_weber an hour ago | parent | next [-]

For now. Atm all compute is like 1% of global energy usage. Until hyper scalers want to 100x their energy spend, which they'll might need to get AGI (to get a step change in capability for along scaling law lines, you need both 10x model size and 10x training data, 100x as much computation. They probably want at least 2 more step changes, so even assuming they 100x efficiency again, there's another 100x that needs raw compute hence energy). Then we're doubling global energy usage.

▲whiplash451 a day ago | parent | prev [-]

AI is a rounding error next to steel, cement, chemicals and food but calling its energy usage negligible is wrong in two ways: (1) it’s one of the fastest-growing sources of demand, and grid planning is driven by growth, not by share (2) it's highly concentrated -- in places like Ireland or Northern Virginia, data centres already use a double-digit share of local electricity.