In today’s issue:
- AI’s biggest bottleneck is energy
- Our power grid is failing before AI spikes demand
- How to power an AI data centre
Ever since AI burst onto the scene, we’ve all been trying to figure out how to profit from it.
Will the companies building AI programs profit the most? Or those using AI to cut their costs?
Perhaps the companies making the hardware that AI runs on are the best way to play the trend? The likes of Nvidia…
That was Sam Volkering’s take back in 2013, long before AI became a mainstream topic. Nvidia has dominated stock market returns ever since.
And now Sam reckons he’s found something closer to home with the same potential he delivered on his Nvidia recommendation.
Perhaps the most profitable AI bottleneck will be in the ravenous energy demand of AI data centres? That was the angle I took in The Fleet Street Letter back in March.
Over the past few weeks, many of the predictions I made in that monthly issue have played out. Today, I’ve decided to reveal some of that analysis here in Fortune & Freedom. You need to understand why our energy world is rapidly changing.
Tomorrow, we’ll dig into how you could profit from AI’s voracious appetite for energy. But first, you need to know how precarious it really is. Here’s how I covered it back in March…
Imagine you are put in charge of implementing Google’s artificial intelligence project.
Congratulations.
Being an ambitious sort, with one eye on the CEO’s job, you would like things at your new job to go very well indeed. Even if you’re all too successful and your AI creation ends up ruling the planet, triggering a global war between man and machine in the process…
Your first challenge is to build a vast data centre to run the AI on. I’m talking a lot of hardware. Huge buildings, terrifying amounts of computer chips, limitless cooling capacity – that sort of thing. It takes astronomical amounts of computing power to train and then run AI.
Your second task would be to power this data centre with the electricity it needs. We’re talking a small country’s worth of power, by some estimates. So, how are you going to do it?
Do you a) rely on the government’s idea of an energy grid to provide enough reliable green electricity at an economical and stable price for the next 20 years while it shuts down baseload fossil fuel power?
Or b) seek out independent sources of power you can build on site to ensure your AI’s operations are impervious to the shortcomings of the country’s failing energy system, no matter what happens?
Your Google’s head of AI. It’s your choice…
But I suspect choosing option a) will get you fired within a year or two. There simply won’t be enough energy to go around. Local politicians will soon demand shutting down your AI site because of how much power it uses while the rest of the grid flickers. And nobody on the board will accept your excuses then. I mean, what sort of industrial leader would rely on the electricity grid at large after the last few years of complete chaos? Not to mention the many warnings from institutions and analysts about the future shortages of power.
Big companies don’t have a choice
Leaving our corporate fantasy behind, I think we all know what the likes of Google and Amazon will choose. They will need to control their own power source to ensure their AI does not evaporate periodically due to intermittent energy or unpredictable political winds.
And they won’t want to share their priceless energy with the rest of the economy either. Especially on a grid that loses a lot of the power on route.
They won’t risk having to switch off their AI mid-global domination because the governor of California wants to charge his rather large electric vehicle on a cloudy day.
They won’t accept uncertain power bills that fluctuate dangerously minute by minute and occasionally surge out of control, turning their whole AI project into a liability.
They won’t want to throttle back their AI due to a temporary lack of energy, making it as unreliable as the rest of the economy constantly on the cusp of an energy crisis.
Just as every tech billionaire has his AI impervious bunker in New Zealand, and a fuelled-up fossil fuel burning private jet to get them there, every tech billionaire’s AI project will have their own dedicated power source to provide cheap energy security to their latest innovation.
But how are they going to come up with enough power without relying on the grid?
We are on the cusp of the commercialisation of a plausible solution. And, if it works, it will change the rest of the energy system with it. There’s nothing like seeing your neighbour’s lights on during a blackout…
I’m talking about a true energy transition. One so fundamental and foundational that it’ll upend the entire energy system as we know it. And replace it with something we have never seen before. A decentralised and fragmented energy grid full of independent power sources. Sources that are privately owned and operated by their own electricity users. Sources that rarely share their power, because they don’t have to – they are each in control of how much power they produce.
But why will all this happen?
Our electricity system is failing before electrification and AI
As things stand, the energy transition is not going very well. We’ve discussed many of the issues in the past. And next month’s issue is expected to cover the question of the electricity grid in more detail. But it is worth reviewing one particular angle that is growing especially concerning.
We are looking at a terrifying energy shortfall in coming years due to the lack of green energy investment. In its Energy Transition 2024 report, energy company Shell explained just how far behind we are:
Current global investment in low- and zero-carbon energy is around $1.7 trillion a year. To reach net zero by 2050, scenarios suggest that $3-4 trillion of commercially viable investment in low-carbon energy is required each year.
The International Energy Agency forecasts an outright drop in clean energy investment. And electricity grids can’t handle the amount of renewables coming online as it is.
Labour cut its intended green energy spending in half despite warnings the amount was not enough to begin with. Bloomberg estimates that California will build less than a quarter of the offshore wind generation it needs. And the UK’s solar ambitions are proving a disaster according to energy consultant Huw Evans, with the target of a fivefold increase “obliterated by inflated costs and gridlock”.
While future energy supply struggles to expand, future demand is growing dangerously fast.
AI is estimated to require about the same amount of power an electrified American electric vehicle fleet would need. We’re talking about adding another Netherlands to the global energy mix by 2027 alone.
The electrification of everything threatens to add impossible amounts of electricity demand to this. The electrification of industrial activity in the US could double US electricity use, for example. Electrifying the UK’s transport industry alone would add about 40% more electricity demand. Using electricity to produce alternative fuels such as hydrogen only makes the challenge more daunting.
Of course, all this masks the nature of that electricity demand, which creates a far bigger problem.
Aggregates hide the extremes
For a functioning electricity grid, we must be able to meet peaks in demand, not just the total amount of electricity needed. And, as there is more electrification, the peaks are likely to grow relative to the average supply.
For example, according to a study by Technical University of Denmark, overall electricity demand could spike by 550% if everyone tries to charge their EV at the same time.
While demand for electricity inconveniently peaks, supply is beyond our control in a renewables-based energy system. That is not a good combination. If a lull in supply meets a peak in demand, we risk blackouts.
Between the vast boom in electricity demand due to electrification and AI, the failure to produce enough electricity reliably as the green dream comes up short, and the implausibility of building a truly extraordinary grid to shuffle electricity around effectively, the electricity system is not going to be fit for purpose much longer… if it is now.
This is not theorising. We have seen the consequences of such shortcomings in places like Germany, Texas and California already. The German version of the OBR, the Bundesrechnungshof, recently released a report that it considers the country’s electricity supply to be “insecure” in the medium term. It also pointed out that this clashed with stated geopolitical aims.
In the political arena, none of this is a secret. Political leaders are already trying to poach industries off each other by pointing out that their competitors can’t keep the lights on.
The point is, any major energy consumer out there must be wondering how on earth they are going to secure enough electricity for themselves in the electrified future. We can’t all move to the US like so many German firms are doing already. (Including the last manufacturer of solar panels in Germany, ironically enough.)
If you’re waiting for a grand national- or international-scale solution to our failing energy system, good luck. It’s not coming. At least, not from governments…
The solution in search of the problem we’ve created for ourselves
There’s nothing like seeing your neighbour’s lights on during a blackout. Or smelling the woodsmoke of their heater while you sit in the cold. In a future without enough electricity, most of us are doomed to wait for our smart meter to turn the power back on. But that’s just not good enough for some of us.
A German documentary I watched about a year ago reported on a spectacular boom in diesel generator sales amongst German professionals. Vets, doctors, car mechanics and countless other small business owners decided they needed to be able to keep the lights on, whatever mad energy policies the government comes up with next.
I believe this trend is about to go global amongst large corporations, cities and mid-sized towns. A revolution in energy supply that identifies government mismanagement of the energy system as the threat and personally secured power as the only plausible solution. But they can’t all run on diesel generators…
So, how are they going to secure the power they need in a zero-emissions world?
***
Find out, tomorrow…
But if you can’t wait to profit from AI, and want to get your hands on Sam Volkering’s latest angle to potentially profit, click here.
Until next time,
Nick Hubble
Editor, Fortune & Freedom
PS AI may be taking the spotlight right now when it comes to investment, but precious metals can be just as valuable an asset in your portfolio. You may think precious metals investing doesn’t pay you an income – but that’s not actually the case. It is possible to generate other types of returns on your precious metals. To find out how, just click here.
[Please note, Southbank Investment Research Ltd receives a commission when accounts are made and investments are deposited with Monetary Metals.]
