AI Apocalypse? Maybe, But We Are Counting The Wrong Things!

The return on AI depends on a multi-year redesign of how work, roles and technology fit together. Current accounting leaves much of that organizational investment off the balance sheet.

Investors spent this week asking who can finance the AI buildout. It is a fair question, and it is only half of the one that matters. The other half is what the spending must be paired with to earn a return. The answer is a system for turning a company’s messy work into cognitive capital, and much of the organizational investment is recorded as a current expense.

The financing question is real. On Wednesday a hotter-than-expected inflation reading helped send the Nasdaq Composite down 1.88%. Broadcom’s chief executive, Hock Tan, had warned earlier in the week that financing could be a bigger obstacle to AI expansion than shortages of chips or electricity. CoreWeave, which reported about $35.1 billion in debt and $16.5 billion in lease liabilities at June 30, became the focus of fears that the data-center boom is running on credit. All of this is about the capital ledger: debt, rates and whether revenue arrives in time.

The easy return and the hard one

Companies can reach first for the easy return. They take a process that already exists and automate part of it. The result can show up in headcount, which is easy to count. The Federal Reserve’s Beige Book, released Oct. 14, reported that some firms in five of its 12 districts had cut staffing through attrition, delayed hiring or expected to add fewer jobs after raising productivity with AI. The Fed also noted that most districts saw little or no net change in overall employment. Challenger, Gray & Christmas data show AI cited in more than 120,000 announced U.S. job cuts through September, about 21% of the total.

What the hard return requires

Two companies that are not technology firms

Two large companies outside the technology sector illustrate elements of this approach, and how hard it is to see.