AI Will Run Out Of Electricity, According To An Expert


The premise of a new study from Bain is that the cost of electricity, and even its availability, will not keep up with data center growth if AI grows at the high end of projections. 

Bain says the cost of data centers could reach $500 billion in capital investment. They add it certainly is above any expected government support. Their experts add, “Bain’s analysis of sustainable ratios of capex to revenue for cloud service providers suggests that $500 billion of annual capex corresponds to $2 trillion in annual revenue.”

The world’s largest tech companies have an average of about $100 billion each on their balance sheets. Taken as a group, they might be able to afford the $500 billion cost. However, if capex rises above that, they would need to turn to the capital markets. It is uncertain whether institutional money would risk huge sums on an uncertain future. They would have to question whether there would be enough AI customers who would spend large sums on a tech which, although wildly powerful, may not be a huge commercial success.

Bain Report

Bain points out that there may be several solutions. One is improved algorithms. Another concern is that not enough electricity will come online to support rapid growth. It is still not clear that US electricity generation can come close to covering AI’s needs when supply is also needed for residences and legacy businesses

What this means, taken together, is that AI may indeed run too low on energy to grow at the high end of expectations.


  1. Finding a fiduciary financial advisor doesn't have to be hard. SmartAsset's free tool matches you with up to 3 financial advisors that serve your area in 5 minutes.
  2. Each advisor has been vetted by SmartAsset and is held to a fiduciary standard to act in your best interests. Get on the path toward achieving your financial goals!
More from ClimateCrisis 247

Similar Posts