Hyperscaler overcapacity threatens the AI trade
The bull case for the AI infrastructure trade is breaking down as hyperscalers face potential overcapacity and a shift toward monetizing excess compute.
The argument
The guest argued that the AI rally was built on the assumption of infinite demand and sustained supply shortages. However, reports of Meta planning to sell excess computing power suggest that hyperscalers may have built more capacity than they can profitably use, potentially leading to a 'secondhand' market for compute.
The thesis, stress-tested
✓ What validates it
- ✓Hyperscalers reporting lower-than-expected cloud revenue growth relative to capital expenditures
- ✓A decline in wholesale pricing for cloud-based AI compute
▸ Risks discussed
- ▸Short-term supply constraints may still persist
- ▸Hyperscalers could find highly profitable ways to monetize excess capacity
Hear it yourself
"One trigger of the pullback, both in Korea and among some AI focused cloud providers in The US, was to report that Meta planned to enter the cloud business to sell excess AI computing power. That caused Meta stock to jump off hours, but reporter Sherry Queen in Singapore told me that it raised alarm elsewhere."
00:00 / 00:22
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE