Browse Comments — Clean (de-noised)

Tarbell link is broken, but here's some info re. funding source for this research:

Nicole Anderson Sandeep Dama

The important question to ask is how does the overall human energy consumption compare with the available energy stores. There isn't an infinite amount of coal, oil, etc. Despite the obvious political and socioeconomic leanings of societal factions, the reality is that humans do not have an infinite supply of energy sources readily available and many are opposed to developing renewable sources. Do these data centers accelerate the rate of depletion? There's no debating the answer. They do. So, how do we compensate?

Kyle Bouwknecht Jill Doucette 👀 bringing your expertise into this for future discussions!

Everyone is talking about AI models. Far fewer are talking about the fact we’re effectively rebuilding the energy grid, data infrastructure, and industrial stack around them. That’s the real story. We already know that, AI is no longer just software. It’s becoming a civilisation-scale infrastructure layer. And the eye opening part? Today’s energy footprint is probably the smallest AI will ever be.

Do the math on ROI as well, when you get a chance.

Maria Hernandez

The single biggest carbon saving from AI is to not use it for things it's not needed for.

No, you didn't; you did the maths.

We need to put data centers in earth orbit as Jeff Bezos has suggested and they can get energy from the sun - then it is zero emissions and they are not using water for cooling, because it is cold in space anyway. This is a very serious problem, a potential threat to human species surviving long term, so moving data centers to space needs to be a top priority. Bezos, Musk, etc. who have the money should fund this themselves.

This is Malthus with a GPU. Project demand forward on a straight line, freeze supply and efficiency where they are today, and you get a catastrophe. It's compelling maths. It's also exactly the same mistake people have been making for 200 years. Energy per computation is falling. Model efficiency is improving with every generation. Inference costs are dropping faster than usage is scaling. And the supply side isn't standing still either. SMRs, next-gen solar, and fusion investment are all accelerating precisely because the demand signal is there. MIT did the maths on today's footprint and drew a line into the future. Malthus did the maths on population and predicted mass famine. He couldn't see the combine harvester coming. Same error, different century.

Philip Bell

Data centres could provide solar for every building in the area to offset their energy consumption, and could provide rainwater recovery either on their own buildings or for locals, to accommodate their water needs. Grownups do this kind of thing.

Can someone explain why some believe we are the smartest species on earth?

When it comes to pushing all those electrons around one thing is for sure ... Overcoming Entropy is a bitch. The Technorati among us always were awae of this. Its high time that the rest of the world comes to term with the fact that AI is not a free lunch

Annalie Killian this may be of interest to you...

Can we cite these findings in environmental reviews? Asking for a friend.

My Perspective on AI’s Energy Footprint The issue is not a single AI query—it’s the scale. Billions of interactions are driving significant growth in electricity demand and data center infrastructure. Three concerns stand out: Transparency: Limited visibility into actual energy consumption. Infrastructure: Massive investments are reshaping energy planning. Accountability: Questions remain about who pays for the growing energy demands. Key Takeaway AI will deliver tremendous value, but long-term success requires balancing innovation with sustainability, efficiency, and accountability.

In high tech expect the pitch to go across the plate but also watch out for "in your ear."

Monica Poling bone dry? Yikes. That’s pretty scary. Seeing the big rivers in the world on a shrinking trend, is heart breaking.