I wouldn't be surprised if mankind will evolve similar to an organism and use 20% of all energy it produces on AI. Which is about 10x of what we use for software at the moment.
But then more AI also means more physical activity. When robots drive cars, we will have more cars driving around. When robots build houses, we will have more houses being built, etc. So energy usage will probably go up exponentially.
At the moment, the sun sends more energy to earth in an hour than humans use in a year. So the sun alone will be able to power this for the foreseeable future.
But the article says that energy use by AI is 48% more carbon intensive than the US average. So talk of solar power is a red herring -- that's not what it is running on now.
You said "for the foreseeable future", which I interpret as being about now.
Anyway I hope you're right, but so far global CO2 output is still growing. All the other energy has only come on top of carbon intensive energy, it hasn't replaced any of it. Every time we build more, we find new ways of spending that much energy and more.
Seeing 20 years into the future is quite possible in some aspects.
I remember how me and my friends discovered email in 1999 and were like "Yay, in the future we'll all do this instead of sending letters!". And it took about 20 years until letters were largely replaced by email and the web. And when the first videos appeared on the web, it was quite clear to us that they would replace DVDs.
Similar with the advent of self driving cars and solar energy I think.
The energy use by AI probably is just as, if not more, carbon intensive, but the article never says that. It talks about the energy use of the general data center.
> The carbon intensity of electricity used by data centers was 48% higher than the US average.
In case anyone is wondering why that is, it's because they put data centers in the places with the cheapest electricity. Which, in the US, is in places like Virginia and Ohio, where they burn fossil fuels.
If the people always talking about how cheap solar is want to fix this, find a way to make that cheapness actually make it into the customer's electric bill.
I've always wondered why data centers aren't taking off more in places like Iceland (cheap geothermal) or Quebec (cheap hydro). Both of these places are also pretty cold and one would think this benefits cooling.
There are periodically news articles and such about data centers in Iceland, of course, but I get the impression it's mostly a fad, and the real build-outs are still in Northern Virginia as they've always been.
The typical answer I've seen is that Internet access and low latency matter more than cooling and power, but LLMs seem like they wouldn't care about that. I mean, you're literally interacting with them over text, and there's already plenty of latency - a few extra ms shouldn't matter?
I'd assume construction costs and costs of shipping in equipment also play a role, but Iceland and Canada aren't that far away.
How much bandwidth is there in Iceland? I suspect not much because the population is only 400K. You will need to lay new undersea fiber. And how are you going to build them? The construction alone would take a massive amount of resources and manpower not feasibly available there. And what about the power supply? In data center heavy areas like Virginia, data center power consumption is already 25% of the entire state power consumption, and VA has 22x more people than Iceland. So if you build even 1/5th the number of data centers in just Virginia, that will consume the entire power grid of Iceland. Therefore, in addition to the data centers themselves, you are also going to have to build an entirely new grid and distribution system.
I did already mention both of those: re. bandwidth, I can't imagine LLMs use that much of it? It's just text - absolutely peanuts compared to something like Netflix. That said, of course, there are multimodal models. Construction difficulty is a factor, but at the same time, it's not like Iceland or Quebec/Canada are backwater regions, they're developed countries. Building a warehouse with some wires in it isn't the most complicated thing ever.
As for power, that's what I was referring to with geothermal and hydro - Iceland and Quebec both have famously cheap electricity. The former would need a large increase in capacity, for sure, but Quebec already pumps out a lot of power (and regularly sells it to the Northeastern US).
Not saying it wouldn't be difficult, by any means, but it does seem like all the right incentives are there.
This assumes no technological adaptions towards efficiency. Consider yourself walking a mile and the energy expenditure. It isn't insignificant. Now imagine you have a bicycle. Some bicyclists will train and do century rides, a distance that were never possible merely walking for a day. But these are few bikers overall, most will not maximize capability to that extent but will still be taking advantage of the efficiency of the bike.
First-order effect is that quality of life will improve though as a result of all that work being done. People able to live more comfortably, relax more etc.
The main complaint about energy usage is it will damage the environment, which will (indirectly) reduce quality of life.
> When robots drive cars, we will have more cars driving around
This doesn't seem true. In SF, waymo with 300 cars does more rides than lyft with 45k drivers. If self driving cars interleave different tasks based on their routes I imagine they would be much more efficient per mile.
Seems like we are way too early in the adoption curve to tell. Currently the average number of passengers per trip is >1.0 across the whole fleet. Some day, I'd expect that to dip below 1.0, as people send an empty car to pick up the dog from the vet, or circle the block to avoid having to pay for parking, etc.
If waymo is doing more rides with 300 cars than 45k drivers on lyft, we can assume then that waymo cars are on the road serving customers at least 150x as long of time as a lyft driver. So yes it could really mean more cars are "around" even if the fleet is much smaller.
> With more than 700 vehicles in its fleet - 300 of which operate in San Francisco - Waymo is the only U.S. firm that runs uncrewed robotaxis that collect fares.
>We’ve also incrementally grown our commercial fleet as we’ve welcomed more riders, with over 1,500 vehicles across San Francisco, Los Angeles, Phoenix, and Austin.
Thank you for this data point. It massively lowers the embodied carbon footprint (carbon from manufacturing, supply chain, transportation, etc.). Operational carbon is a solved problem; it is easy to measure and can be supplied from renewable sources.
Most of the sunlight that hits a roof is already turned into heat. Whether you use that for calculations or not does not make a difference.
Not sure about the exact numbers, but I guess that at the moment normal roofs and solar panels absorb very roughly about the same percentage of sunlight.
So if in the future solar panels become more efficient, then yes, the amount of sunlight turned into heat could double.
Maybe that can be offset by covering other parts of earth with reflective materials or finding a way to send the heat back into the universe more effectively.
Because a mix of renewables deployed on a continent (most grids tend nowadays to extend on whole continents because it enhances continuity of service while offering many ways to optimize for emissions, costs...) is better (cheaper, less dependency towards any fuel, less risk related to accidents/neglects/mistakes/war/terrorism/hot waste/... ...).
Building and running a nuclear reactor involves a lot of physical activity. And if the past is an indicator, we always move from physical activity to the flow of electrons.
The discussion about nuclear vs solar remind me of the discussions about spinning HDs versus solid state drives when they were new.
I wouldn't be surprised if mankind will evolve similar to an organism and use 20% of all energy it produces on AI. Which is about 10x of what we use for software at the moment.
But then more AI also means more physical activity. When robots drive cars, we will have more cars driving around. When robots build houses, we will have more houses being built, etc. So energy usage will probably go up exponentially.
At the moment, the sun sends more energy to earth in an hour than humans use in a year. So the sun alone will be able to power this for the foreseeable future.