AI spending at Microsoft and Oracle is even higher than it looks, thanks to this accounting maneuver

Dow Jones
Sep 27

MW AI spending at Microsoft and Oracle is even higher than it looks, thanks to this accounting maneuver

By Christine Ji

Big Tech companies are using finance leases to fund data-center buildouts, circumventing massive up-front cash outflows

Microsoft and Oracle have dramatically increased their use of finance leases, and Meta and Amazon are following suit.

Big Tech companies are on track to spend hundreds of billions of dollars this year to build out artificial-intelligence infrastructure - but the real amount is even higher than headline numbers suggest.

Companies like Microsoft Corp. $(MSFT)$ and Oracle Corp. $(ORCL)$ are increasingly utilizing an accounting maneuver to invest billions more into data centers without it showing up in their traditional capital-expenditure figures: finance leases.

When a company purchases an asset up front, its free cash flow - or the amount of money it has after paying for operating expenses and capex - takes a hit. Oracle, Amazon.com Inc. (AMZN), Meta Platforms Inc. (META) and Alphabet Inc. $(GOOGL)$ $(GOOG)$ have all recorded free-cash-flow declines, year over year, due to massive AI capex - meaning that they have less cash available to return to shareholders through stock buybacks or dividends.

For Big Tech companies in the midst of a heavy multiyear investment cycle, using finance leases can help them bypass the large free-cash-flow impact. A finance lease transfers ownership risks and rewards to the lessee, making the transaction economically similar to purchasing an asset with debt. No cash is used up front upon initial recognition of the finance lease, and the company makes lease payments throughout the lease term.

Among hyperscalers, Microsoft and Oracle have increasingly been utilizing finance leases to fund AI buildouts. In fiscal-year 2025, Microsoft's total finance-lease liabilities totaled $46.2 billion, a 70% increase from $27.1 billion the previous year. Oracle recorded $4 billion in finance-lease liabilities for its most recent August quarter - a significant jump from a year ago, when it had no balance.

Also read: Why the $300 billion Oracle-OpenAI deal could be fueling an AI bubble

Lease financing is an especially attractive option for Oracle. Consensus revisions for Oracle's capex have jumped 180% from a year ago, according to a recent note from Todd Castagno, global valuation, accounting and tax strategist at Morgan Stanley.

The company will need to aggressively build out AI infrastructure in the coming years, especially in light of the Oracle's reported $300 billion partnership with OpenAI. However, Oracle is already burning through more cash than it's generating, incurring $362 million of negative free cash flow in its most recent quarter. If Oracle's capex outpaces its operating cash flows, then its free cash flow could decline further. By using finance leases, Oracle can build out data centers without incurring huge up-front cash outflows.

As capex levels continue to rise in coming quarters, finance leases will become increasingly important to the future of the AI trade. Combining traditional capex with finance leases shows that capex intensity among hyperscalers is more extreme than what investors may expect. It also provides a clearer comparison between hyperscalers that buy their data centers outright versus those who lease.

According to Morgan Stanley estimates, Microsoft's fiscal-year 2026 capex-to-sales ratio increased to 38%, from 28%, when including finance leases. Oracle's capex-to-sales increased to 58%, from 41%, for the same period.

At a time when Microsoft, Oracle, Meta and Google have reached their highest capital-intensity levels since their IPOs, more hyperscalers are starting to tap into finance leases. Amazon, which had previously moved away from finance leases in favor of owning assets, has recently reversed course and begun committing to new leases. The company reported $937 million worth of property and equipment acquired through finance leases for its most recent July quarter, up dramatically from the $181 million spent in that same period a year prior.

Additionally, Amazon and Meta have been increasing their lease commitments, or contracted leases that haven't started yet. For Meta's latest July quarter, the company reported $52.6 billion of operating and financing lease commitments, primarily for data centers and networking. That marks a sharp increase from the $6.2 billion of operating lease commitments that Meta reported a year ago.

Once these leases begin and the assets come into use, these companies will begin incurring the associated expenses.

Read: This is the critical detail that could unravel the AI trade: Nobody is paying for it.

-Christine Ji

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

 

(END) Dow Jones Newswires

September 26, 2025 15:39 ET (19:39 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