Company Cuts Staff to Pay for AI, Then Explains Why AI Did Not Cut Staff

WASHINGTON — There is a word that has entered the business vocabulary this year, and it is a good word. It is “efficiency.” Every company in America is now, in one form or another, “improving its efficiency.” It is the word you use when you are about to do a thing that is, by every other available word, a layoff, but that by this word is a streamlining, a right-sizing, a recalibration of the workforce toward the future.

The future, it turns out, has a price tag, and the price tag has a name: AI.

The story, as it has been reported this week, is that Oracle is preparing another round of layoffs, and the reason is not, as the company’s own investors might expect, that the work is shrinking. The reason is that the company is spending so much on AI infrastructure that it needs to find the money somewhere else. The somewhere else is the people. Per reporting on the round, Oracle has already cut tens of thousands of jobs this year, and the new cuts are, in the company’s own logic, needed to fund the AI spending — the machines that Oracle has, for months, been telling you are not the reason any of this is happening.

This is the paradox, and it is the story.

Here is how the arithmetic works, and it is worth following it, because the arithmetic is the joke. The company buys a great deal of AI. The AI is expensive. The AI is also, the company says, going to make the company more efficient. Being more efficient means, in the old and honest language of business, needing fewer people to do the same work. So the company has fewer people. It is more efficient. It needs to pay for the AI that made it so. It cuts the people. It is now even more efficient. It needs to pay for more AI to be more efficient still.

A financial analyst, speaking on the condition that we not say what he actually said, described the situation as “a very good quarter.” When pressed on what, exactly, was good, he said, “The margins. The margins are the good part.”

Now, it would be easy to call this a contradiction, and for the last few decades, that is exactly what a contradiction was. A contradiction was a thing that could not both be true. But the business vocabulary has moved on, and a contradiction is now called a “tension,” and a “tension” is something you “manage,” and a thing you manage is a thing you are, in a very real sense, winning.

The CEO of a rival company, when asked about the practice, was characteristically measured. “I don’t think it’s as simple as people say,” he said. “There’s a lot of nuance in what ‘AI’ means. There’s a lot of nuance in what ‘a job’ means. There’s a lot of nuance in what ‘paying for it’ means.” He did not say which of the three nuances he meant. He did not say that he meant any of them.

It is worth noting, for the record, that the technology that made the cuts possible is itself a thing that was sold to the public, in very specific terms, as a thing that would create work, not a thing that would be paid for by the absence of work. That pitch is still on the company’s website. It is right there, under “Careers.” It says, in a font that is very much a font, that the company is “building the future of work.” It does not say what that future is. It does not say that the future is a server room and one small human worker in a polo shirt, standing in the aisle, looking up.

At press time, the company said it was “excited about the efficiency gains” and “committed to our people.” When asked to reconcile those two sentences, the spokesperson said, “They’re both true.” We believe the spokesperson. We believe that, in a way that is very much a way, both are true.