0:00
/

Paid episode

The full episode is only available to paid subscribers of The official Ryan McBeth Substack

AI’s Biggest Problem Isn’t Chips—It’s Electricity

The bubble may be in for a slow deflate rather than a pop

Talking about the AI bubble is all the rage right now, but what if we’re not in for a pop, but a slow deflate due to the inability to power new datacenters?

Support my work and get this video as a paid member with no ads for only $5.

Share

A recent Wall Street Journal Article [Paywall] announced that there was a “yearslong” wait for gas turbine engines. Companies like Boom Supersonic are manufacturing turbine power plants. Another company called ProEnergy sells conversion kit solutions to turn old jet engines into gas-turbines.

Add to all this, Microsoft recently announced that it intends to reopen the shuttered Three Mile Island Nuclear Power Plant to create power for its datacenters.

So why all of this sudden interest in turbines?

With a three to six month wait for powerful AI chips, but a five to seven year wait for electrical grid connection, it’s clear that the limiting factor for datacenter construction isn’t chips, it’s power.

This leads me to believe that we may certainly be in an AI bubble, but rather than a sharp “pop” we may be in for a slow deflation as datacenter construction bumps up against the limiting factor of electricity.

Also note that some have suggested that the engines from surplus Spirit Airlines planes were jokingly floated as a possible solution, these engines are far too valuable as replacement parts to be used in datacenter construction.

Share

User's avatar

Continue reading this post for free, courtesy of Ryan McBeth.