The hidden cost of AI spending at big tech

This article summarizes an episode of WSJ Podcasts’s video series featuring CAE Consulting principal Kevin Koharki and Financial analyst Jonathan Weil.

Image credit: Made by Tech in Asia with the help of AI

Behind the glittering facade of infinite tech wealth lies a precarious foundation built on borrowed money, constant share dilution, and optimistic forecasts.

Purdue University accounting professor and CAE Consulting principal Kevin Koharki believes investors should take a closer look at how stock-based compensation, buybacks, and capital expenditures affect free cash flow.

Factoring in employee stock reveals the true cost of AI

Executives present stock compensation as a free benefit rather than a business expense to conceal the true cost of their payroll.

To measure how much money a business truly retains, Koharki developed a stricter method that treats employee shares as a direct expense whenever a company offsets newly created equity through a precise sequence of deductions:

• Start with operating cash flow: Use the baseline figure representing the everyday money a business generates.
• Subtract equipment spending: Deduct money spent on physical assets, including the computer chips and buildings required to run new technology.
• Deduct tax payments: Account for the cash employers must send to the government when workers receive shares.
• Estimate buyback costs: Factor in share repurchases used to control dilution, a purpose confirmed by public disclosures for recent transactions at Meta.
• Remove lease payments: Deduct payments made on equipment leases to prevent companies from treating financed computers as free money.

This calculation lowers the estimated available cash for these companies, as Meta’s expected 2025 cash falls from US$46 billion to US$5 billion, while Alphabet’s drops from US$73 billion to US$24 billion.

Share repurchases drain the bank account just like regular payroll
“If at the end of the day I’m going to buy back the stock and pay cash,” Koharki argues, “then that to me is an operating activity.”

Substituting shares for a regular salary produces the same financial drain on the business once those shares are repurchased, revealing a much smaller pool of retained earnings once equipment purchases and employee share costs are subtracted.

Technology giants must find new ways to pay for their projects

This revised math transforms an abstract accounting debate into a direct financial warning about how tech giants fund future growth.

Investors have historically viewed cloud and software businesses as highly profitable operations that required relatively little physical infrastructure.

The AI boom is changing that equation as spending on data centers, chips, and related assets accelerates.

“Your CapEx is going up,” Koharki notes, “let’s say, 30, 40, 50% a year.”

Hidden obligations force regular investors to shoulder the burden

Current market prices demand a financial miracle

Hardware suppliers look profitable because buyers delay reporting expenses