The Pathetic Reality At The Heart Of The AI Revolution
This proves we were right all along.
There are a million and one reasons why the AI revolution is a step in the wrong direction, and by now, I feel like I have written an article discussing every single one. But arguably, one of the worst reasons is its energy use. AI is predicted to use up to 1,050 TWh per year by 2026, equivalent to the current energy use of Germany or more than twice its already ballooning rate. This is threatening to undo the tiny amount of climate progress we have managed to achieve and allow big tech companies to become some of the worst polluters on the planet. As a result, the major AI companies are turning to nuclear power, and every serious climate scientist has collectively held their head in their hands.
Right now, Amazon, Google, and Microsoft are the most notorious AI corporations because they not only develop their own AI but also build, maintain, and operate AI infrastructure, such as data centres, for both themselves and other AI companies. For example, while OpenAI is the largest AI company in the world, it uses Microsoft infrastructure to run. As such, the vast majority of the AI industry’s infrastructure expansion and exponentially growing energy use are thanks to these three companies.
But Amazon, Google, and Microsoft, along with every other big tech company, are also under huge scrutiny over their emissions. Their industry is arguably the easiest to decarbonise, as they can easily adopt sustainable energy and efficiency improvements to reach net zero carbon. It’s not like aviation, heavy industry, or shipping, which require revolutionary new technology and industry-wide radical restructuring in order to meet this target. And so, these three companies are in a desperate race to reduce their emissions before they are forced to, and their business model is compromised.
AI stands in the way of this. As I have covered numerous times now, for AI development to continue at the rate it has, it needs exponentially more data, infrastructure, energy, and funds. What’s more, Amazon, Google, and Microsoft are aware that, right now, AI is simply not profitable. They know that its value and investment drive are entirely based on it becoming significantly better than it currently is, as only then can it have justifiable and useful applications.
So, in order for these tech giants to ensure the tens of billions of dollars they have sunk into AI isn’t totally wasted, they need to secure vast amounts of round-the-clock, on-tap, ultra-low carbon energy incredibly quickly. Not only that, but the energy also needs to be incredibly safe, as they are private companies and therefore have the issue of public liability.
So, what have they turned to? Solar? Wind? Hydro?
No, they have all turned to nuclear power.
Amazon has invested deeply in X-energy, a nuclear company developing next-gen Small Modular Reactors (SMR), which promise to be quicker to deploy and cheaper than current nuclear technology in order to power, or even offset, its AI emissions. Microsoft has paid for the Three Mile Island nuclear power station to restart operations to power its AI data centres. Google has signed a deal to buy six SMRs from Californian nuclear start-up Kairos Power.
These actions alone have vindicated almost every single serious climate scientist of the past 30 years. Why? They have known that nuclear power was the best solution to decarbonisation for decades.
Endless research has found that nuclear power has a smaller carbon footprint than wind or solar power. In fact, one study even found that solar power produces 50% more emissions per unit of energy than modern nuclear technology. It also produces a fraction of the waste of solar or wind power, and said waste is far better managed when you consider that solar panels going to waste can leak toxic materials. Not only that, but nuclear is one of the most compact forms of energy we have, even when you take into account uranium mining and refining, meaning it produces less habitat loss per unit of capacity. And, as a cherry on top, nuclear power actually produces less environmental radiation per unit of energy than coal power!
But nuclear is unsafe. Chornobyl and Fukushima showed us that, right?
Well, if you actually crunch the numbers, you will quickly find this isn’t the case. Nuclear power only causes, on average, 90 deaths per 1,000 TWh (which includes the Chornobyl and Fukushima incidents), compared to wind power’s 150, solar power’s 440, natural gas’s 4,000, and coal power’s 100,000 deaths! This is because these other energy sources require more mining, more industry, and handling higher volumes of toxic materials, as well as fossil fuel energy producing deadly emissions.
This is why climate scientists have been calling for more nuclear power for decades. It is not only safer and lower carbon than any other widely applicable form of energy we have, but it is also far easier to integrate into our current infrastructure. There would be no need for grid batteries, long-distance energy transmission, or decentralised grids.
The only genuine potential downside is price and deployment time, as nuclear power costs around $170 per MWh and nuclear plants cost billions upon billions of dollars to build, as well as taking decades to fully implement. But we were working on solutions to this back in the ’90s when the first SMRs were developed (learn more about SMRs here). These promised to halve the cost of building nuclear power plants and reduce the time to deploy them from decades to just a few years.
Unfortunately, funding for nuclear power was dramatically cut in the 1990s and early 2000s for a number of reasons. The most significant reason was that fossil fuel companies started pushing natural gas as a cleaner “transitional fuel,” soaking up the investment that nuclear power needed. As such, SMR and other next-gen nuclear technology haven’t really moved forward since the mid-’90s. That’s why current SMRs currently cost as much and take as long to build as traditional reactors, as they are still underfunded technology in their infancy. Consequently, our nuclear industry has dramatically shrunk and is now a shell of what it once was.
This is why AI’s (or at least generative AI’s) rapid adoption of nuclear technology is deeply ironic. This is a nonsense technology that isn’t profitable, has no genuinely useful applications, is entirely based on stealing copyrighted work and individuals’ labour, and uses so much energy that it threatens to derail our pathetic efforts to save the planet. Yet, this is enough for us to finally take nuclear technology seriously?!
Oh, the environments we depend on will disintegrate and our economies will crumble thanks to planet-wide climate change caused by our carbon emissions? No need to invest in the technology that is an obvious solution to the issue; we will instead fall for the propaganda and lobbying of the oil billionaires that got us into this mess in the first place and make it our lives’ mission to continue to enrich them.
Monopolistic tech companies need copious amounts of safe ultra-low carbon energy to continue growing their power-hungry bullshit borderline scam technology they have sunk tens of billions of dollars into without being held accountable for their ballooning emissions? Sure, we can invest in nuclear technology. Only it’s several decades too late to actually make a difference.
This shows that there was nothing in the way of us widely adopting nuclear technology decades ago and getting ahead of the climate action curve. The money was always available. It is more than safe enough. There was never really any solid political stance against it.
How pathetic are we?
Protecting tech-bros’ moronic multi-billion dollar investments in snake-oil technology is apparently more of a motive than saving the only planet we can ever call home.
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and help get the word out by hitting the share button below.
Sources: The Guardian, The Guardian, CNBC, Amazon, Will Lockett, Will Lockett, The Verge, BBC, Change Oracle