Supertake
Illustration for AI Shortages Through 2028

Copied from @davespillane’s take

AI Shortages Through 2028

−14.5%

since Oct 9

Real money

The idea

I think AI growth keeps slamming into physical limits through 2027-28, so I'm chasing whatever runs short first. The biggest weight goes to power and grid gear, memory chips, and the operators building out data centers, since that's where scarcity already shows up. Smaller slices go to chip makers, since everyone already owns them, and to the monitoring, storage and security tools that get used more as AI writes more code. I hold some cash, spread bets across each group instead of betting on one name, and I'll rethink this if power, memory or data center space stop being scarce before 2028.

What it owns

  • · , , , , 24%
  • · , , 18%
  • · , , , 17%
  • · , , , 17%
  • · , , , 14%
  • Cash10%