The AI Power Crunch
While SA tries (and has largely succeeded) to solve load shedding, other countries are scrambling to build out power generators to feed the AI monster. We are nowhere to be seen in this race.
The US and China are probably going to lead the scramble to dominate artificial intelligence. The race is to build Artificial General Intelligence (AGI) that is leaps ahead of current versions of AI in that it will allow machines to solve problems better than a human being and without any form of manual intervention.
There are dangers in this which we will explore in a future article: superintelligence can – and will – become weaponised. Rogue states could use it to threaten annihilation or to sabotage competing nation states, even taking out any nuclear deterrents. Those with just a few months’ lead in AGI will have a huge competitive advantage.
To do that, the AGI promoters are going to need lots of power. It takes a decade or so to build a gigawatt nuclear power plant (sorry, solar panels and wind turbines won’t do it). US electricity generation has grown about 5% in the last decade, but this will have to accelerate at much faster rates if the potential of AI is going to be realised. The US could use its huge natural gas reserves to get the power it needs in about 12 months, but that would need $100 billion in capex for 100 gigawatts of natural power plants, according to this article by Leopold Aschenbrenner.
The revenues from AI are massive, and this is where capex is going in the next few decades.
Look at the graph below, showing datacentre revenue for chip maker Nvidia.
Big tech companies are looking at these revenues with mouths agape. Microsoft and Google will likely do $50 billion-plus, and AWS and Meta $40 billion-plus in capex this year, much of it into AI.
The revenue return rates from AI investment will become a huge part of these businesses. Microsoft has 350 million subscribers for its Office suite. How many of these would be willing to pay $100 a month for an AI add-on that vastly improves productivity?
“It’s hard to understate the ensuing reverberations. This would make AI products the biggest revenue driver for America’s largest corporations, and by far their biggest area of growth. Forecasts of overall revenue growth for these companies would skyrocket. Stock markets would follow; we might see our first $10T company soon thereafter,” argues Aschenbrenner.
Big tech at this point would be willing to go all out, each investing many hundreds of billions (at least) into further AI scaleout. We probably see our first many-hundred-billion dollar corporate bond sale.
All this is happening while SA scrambles to keep the lights on, and while we exported our modular nuclear reactor tech and expertise for want of a bit of vision and investment from government.
