I wrote earlier this month about why a perfect storm is about to hit global power grids.
Investors and companies spend very little time thinking about electricity. For most companies and consumers — at least outside of South Africa — simply hook up to the grid and turn o the lights.
The same was true of the data centre business for much of its existence but that’s changing quickly. The Xbox network in California alone already uses more electricity than entire African countries and things like generative AI will crunch grids further.
Here is a (slightly dated) chart showing the acceleration in power demand from tech companies.
Here was Sam Altman last month:
“We still don’t appreciate the energy needs of this technology. There’s no way to get there without a breakthrough. We need fusion, or we need like radically cheaper solar plus storage or something at massive scale.”
On Friday, Mark Zuckerberg gave an interview where he made the same point.
One of the things he highlights is the slow pace of energy permitting, including generation and transmission lines. Add into that the difficulties of getting green metals like copper and lithium out of the ground and you have a problem. Then factor in the need to limit carbon dioxide emissions and it’s compounded.
Right now, the market is turning to natural gas and coal to feed into new demand but that’s not going to work in the long term.
Zuckerberg argues that it’s a good idea to invest $10 billion or $100 billion into securing power supply but he also talks about the uncertainty around how much is needed and why it’s a tough decision to commit that capital. That hesitancy and the long lead times will lead to an inevitable crunch.
Investing around that theme is tricky. If we’re power-limited then perhaps there’s a case for the utilities themselves but they’re heavily regulated and that means that the payoff is moderate at best. There are companies building out transformers and the like but some have already moved and that’s still a relatively small industry.
Natural gas is cheap right now and that may ultimately be the beneficiary even though many people would rather the demand go to green energy instead. What’s happening is that places like Texas with abundant natural gas are courting data centres.
I’m sure there are other avenues I haven’t considered but I think many roads lead to copper because there simply isn’t enough supply and lead times for new mines are extraordinary. The risk is that China’s demand for copper drops on a soft economy, creating some breathing room. However the risks run both ways as demand for power, EVs and robotics could supercharge the supply deficit.