EDITION OF OCTOBER 26, 2026
AI profitability: the limits of Apollo’s assessment
According to a note by Apollo’s economist reported by TechCrunch, margins and revenue per employee are stagnating within the scope studied. This observation raises questions about AI’s financial benefits, but the available material does not allow its effects to be isolated.
Legal disclosure: this content was produced by an artificial intelligence system. Editorial transparency
An Apollo note questions whether AI is already delivering discernible financial benefits for companies. The reported assessment concerns aggregate indicators and warrants careful reading: it raises the question of investment returns without, on its own, establishing the effectiveness or failure of each use.
Are investments in artificial intelligence already showing up in companies’ financial results? A note by Torsten Slok, chief economist at Apollo Global Management, offers a skeptical answer. According to TechCrunch’s account, profit margins and revenue per employee appear to be stagnating in the group of companies he examines. Slok sees no sign of an improvement in productivity attributable to AI.
His scope is the S&P 493: the S&P 500 excluding Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla. This definition is essential to understanding the assessment. The analysis excludes these large technology companies; the available material does not provide a detailed assessment of each sector or each remaining company. The economist’s very categorical conclusion about the absence of financial benefits must therefore remain attributed to his analysis.
Slok also offers an explanation: access to increasingly inexpensive intelligence would not, in itself, be enough to accelerate revenue growth or increase margins. He emphasizes the judgment required for business decisions, along with responsibility, trust, culture and unwritten rules. These are the arguments of the note’s author, rather than an independent demonstration of the limits of every use of AI.
The distinction between observation and causality is crucial here. Aggregate indicators that appear stable cannot, on their own, determine what would have happened without these investments. The account specifies neither the measurement period nor the amounts invested by the companies observed, and it provides no comparison between companies that use AI more extensively and others. It therefore does not allow a return on this spending to be calculated or changes in these indicators to be attributed solely to AI adoption.
A reasonable interpretation is more limited: within the scope presented by Slok, the expected financial benefits are not visible through the indicators cited. Assessing a promise of profitability would require linking a specific use to its costs and results. The note raises this question forcefully; the information reported is insufficient to resolve it for companies as a whole.