Everything has a price, and that includes AI. And that cost may well be only indirectly financial. The bigger problem is its energy demands. AI is computationally expensive, and that means energy. Just as crypto mining has been a serious energy issue, AI is gobbling up electricity at an alarming rate.

Robinson Mayer, writing for Heatmap:

According to its supporters and critics alike, generative artificial intelligence like ChatGPT is about to devour huge amounts of electricity, enough to threaten the grid itself. “We still don’t appreciate the energy needs of this technology,” Sam Altman, the CEO of OpenAI, has said, arguing that the world needs a clean energy breakthrough to meet AI’s voracious energy needs. (He is investing in nuclear fusion and fission companies to meet this demand.)

The likely energy needs of AI are a serious concern in many quarters, not just among AI enthusiasts. Here’s Michael Khoo, climate disinformation program director at Friends of the Earth, quoted in The Guardian:

“It’s not like AI is ridding us of the internal combustion engine. People will be outraged to see how much more energy is being consumed by AI in the coming years, as well as how it will flood the zone with disinformation about climate change.”

And here’s Brian Calvert writing for Vox:

If ChatGPT were integrated into the 9 billion searches done each day, the IEA says, the electricity demand would increase by 10 terawatt-hours a year — the amount consumed by about 1.5 million European Union residents.

And that’s just for searches, let alone any other use of generative AI. These are worrying numbers at a time when we’re looking hard at our energy use, in the face of the climate crisis.

How real are the AI energy predictions?

Now, to be fair, there are some reasons to think that the demands might not be quite as high as predicted. Here’s Mayer again:

Jonathan Koomey, a researcher who studies how the internet and data centers use energy (and the namesake of Koomey’s Law) told me that many estimates of Nvidia’s most important AI chips assume that their energy use is the same as their advertised “rated” power. In reality, Nvidia chips probably use half of that amount, he said, because chipmakers engineer their chips to withstand more electricity than is necessary for safety reasons.
And this is just the current generation of chips: Nvidia’s next generation of AI-training chips, called “Blackwell,” use 25 times less energy to do the same amount of computation as the previous generation of chips.

So, technological advances are more than likely to offset some of the demands generated by increasing AI use — but probably not all of it. And that leaves us with a double quandary: how to both decarbonise our power, and provide more of it at the same time.

There’s nothing like a challenge.

Geothermal energy rocks

Now, one obvious solution would be to go searching for an ever wider range of sustainable power. We’ve seen big advances in solar, wave and wind power recently, but other forms could have a significant role to play. Take geothermal energy, for example:

Some posit that geothermal could be a valuable tool in transitioning the energy system off of fossil fuels, because it can provide a continuous backup to intermittent energy sources like solar and wind. “It’s been, to me, the most promising energy source for a long time,” says energy engineer Roland Horne of Stanford University. “But now that we’re moving towards a carbon-free grid, geothermal is very important.”

But how realistic is that? This might surprise you, but for some countries, it’s already a considerable part of their energy mix. Iceland already gets 25% of its power from geothermal energy, and Kenya nearly half:

In places where molten rock sizzles close to the surface, water will seep through porous volcanic rock, warm up and bubble upward as hot water, steam or both. If the water or steam is hot enough — ideally at least around 300 degrees Fahrenheit — it can be extracted from the ground and used to power generators for electricity. In Kenya, nearly 50 percent of electricity generated comes from geothermal. Iceland gets 25 percent of its electricity from this source, while New Zealand gets about 18 percent and the state of California, 6 percent.

However, these are places that have a happy combination of molten rock near the surface, and porous rock about it. That’s not found everywhere in the world — and we’re close to tapping out the areas where this is easy to do.

New energy from EGS

To spread geothermal power further, more work is needed on a technology called Enhanced Geothermal Systems (EGS):

Water is pumped at high pressure into wells, up to several miles deep, to blast cracks into the rocks. The cracked rock and water create an underground radiator where water heats before rising to the surface through a second well.

There have been some problems with this in the past—including an earthquake. However, some promising trials of new approaches just need some more investment to see if they could be scalable to useful levels.

Making better use of what we’ve got

Another approach might be to make use of resources that otherwise go to waste — like the heat that’s produced as a by-product of all that AI activity. And that’s precisely what’s happening in Ireland:

At our Tallaght data center, hot air from servers is captured by a heat exchanger in the air handling unit, which increases the temperature of water running through the building to approximately 23-28° Celsius (that’s 73-82° Fahrenheit for my American readers) before it’s pumped to Heat Works’ energy center via insulated pipes. The system takes this warm water and passes it through the first of two heat pumps, where it is condensed until the temperature is approximately 45° C (113° F). The hot water then passes through a second heat pump, where the water temperature is increased again to between 70° – 85° C (158 – 185° F) depending on the season. At this point, the piping hot water leaves the energy center in insulated pipes and is available for customers connected to the heating system.

That’s the CTO of Amazon, talking about how they’re using the heat generated by the AWS servers in that Irish town to provide heating to a district. That takes heat, a waste product generated from electricity by the computation process, and turns it into a resource. And by turning it into a resource, you massively reduce the need for energy to heat the town.

In other words, in an almost judo-like move, you turn the increased demand for energy into a way of reducing a different demand for energy. It is, if you like, finding new energy by hugging the system.

Solving the twin challenges of increased demand for energy and a need to decarbonise that power is going to require that level of creativity. And where there’s a need for creativity, there’s an opportunity for innovation — and an opening for new business models in saving the world.

Photo by Appolinary Kalashnikova on Unsplash.