Mismatch

The AI world is starting to look a lot like the real estate market in ghost cities in China. An incredible push in data center infrastructure, to the tune of $600Bn (see chart below), built for an expectation of demand that may not be there, at least in the short term. The usage for model trainings could be there, but the revenues associated with the use of those models, may not be present yet. And there lies the problem, the mismatch between the debt used to build those data centers (CapEx), the cash burn needed to operate them (OpEX), and the actual usage and rent price when there is too much supply of data center space. The infrastructure should be absorbed over time, the same way it happened with the internet in the 2000’s, but the short term mismatch can become a problem, particularly if financed with debt. Oracle has committed $200Bn to build and rent data centers space to Open AI, and the demand, backed by revenues may not be there, as OpenAI continues to operate at a big loss. That’s why Oracle CDS is going vertical. Microsoft, Amazon, Meta and Google have higher cash flow to support these endeavors, and even they will suffer. Oracle does not have that cash flow. In 2024, it’s FCF was $11Bn, last year it sunk to -$395M, and it will continue to bleed cash during 2026. That’s why they’re raising capital. It will be interesting to see the March quarterly earnings report.

Want to know more? You can register for free at Fund@mental.

#iamfundamental #soyfundamental #wealthmanagement #familyoffice #financialadvisor #financialplanning #policymistake #ratecut #stagflation