A Cautionary Tale for Meta, Microsoft and Other AI 'Hyperscalers' -- WSJ

Dow Jones
Oct 23

By Spencer Jakab

This is an online version of Spencer's Markets A.M. newsletter. Get investing insights in your inbox each weekday by signing up here-it's free.

What's inside the box? Huge, nondescript buildings that consume as much power as cities are all the buzz these days, but only some of them.

Data centers costing billions of dollars apiece are vital to the AI boom, and you're probably helping to finance them indirectly through bond funds or by owning the shares of "hyperscalers" Microsoft, Meta, Alphabet and Amazon.

It's also possible to make a direct investment in data centers via real-estate investment trusts, but shares of the two big ones are down this year. That's because the server banks these trusts own mostly handle more prosaic computing tasks like handling flight reservations, storing your selfies or making my Markets A.M. newsletter possible.

Investors' lack of enthusiasm about what not long ago seemed like owning a piece of the future might hold clues about the eventual returns on today's AI boom. Digital Realty, the world's fourth-biggest REIT of any kind, will unveil quarterly results today. Larger Equinix reports next week.

Both landlords have expanded rapidly, but adding more data centers isn't the same as adding value for investors. Digital Realty, for example, will have nearly $6 billion in revenue this year, up from $1.8 billion a decade ago when it joined the S&P 500 index. It has high occupancy and prime locations for future development.

But it consumes oodles of cash to maintain and grow facilities so it constantly issues shares and borrows money. Its funds from operations per share are unchanged over a decade after inflation.

Why not just reinvest more of that rent money? REITs are pass-through entities that pay no corporate tax but have to distribute about 90% of income. Even so, Digital Realty's 2.8% dividend yield is paltry. Equinix's is just 2.3%. Their five year average returns on invested capital were just 2.4% and 2.5%, respectively, according to FactSet.

Both have some AI exposure, but pure-play AI data centers are in a different universe in terms of buzz. That doesn't mean their financial future will be rosier. They, too, will have to constantly upgrade chips and servers and renew leases. Being a landlord is expensive.

At least the public REITs already earn a positive return. Consulting firm Bain estimates that AI revenue will have to reach a whopping $2 trillion in five years to justify current AI infrastructure investments-more than five times the revenue for all subscription software today.

Today's building boom matters to all investors because it has sucked the air out of the room. Goldman Sachs says that the hyperscalers alone account for 27% of the S&P 500's capital expenditures and that their outlays will grow more than 20% next year.

If their returns are anything like data-center REITs today then that'll weigh on the market for years.

This item is part of a Wall Street Journal live coverage event. The full stream can be found by searching P/WSJL (WSJ Live Coverage).

(END) Dow Jones Newswires

October 23, 2025 07:42 ET (11:42 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.

At the request of the copyright holder, you need to log in to view this content

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10