Oracle: The AI Layoff Confession

Confession. Everything about a modern corporation is built to prevent one. The lawyers, the crisis teams, the soft passive voice of every quarterly call, a whole machine tuned so no executive ever says anything expensive in public. On 22 June, Oracle said it regardless.

Not from a stage, where words get walked back by lunch, but in a federal filing.

"The adoption and deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce," the 10-K states. With that single sentence, Oracle became the first Big Tech name to put AI on the record as the reason its people lost their jobs.

The sentence has a body count. Oracle shed 21,000 people in a single year, down from 162,000 to 141,000, roughly one worker in eight.

The bill for sending them home ran to US$1.84bn, nearly five times what it paid the year before. Nobody spends two billion dollars to make people leave on a whim. Someone at the top had decided they were no longer the plan.

Larry Ellison, the company's Chairman and Chief Technology Officer, stands at the company's own developer conference and tells the room his programmers have stopped writing code.

 "The code that Oracle is writing, Oracle isn't writing," he says. "Our AI models are writing."

The men who claim it isn't happening

The strangest objections come from the people standing to gain most from the boom. NVIDIA CEO Jensen Huang, whose company sells the hardware of the entire AI gold rush, calls the story lazy to its face.

"The narrative that connects AI to job loss for many of the CEOs that are doing it, it is just too lazy," he says, asking how a technology that "became productive and useful only six months ago" could already be clearing out floors of offices.

OpenAI CEO Sam Altman, whose company makes the most famous of those models, goes a step further. Almost every firm cutting staff now points at AI, he says, "whether or not it really is about AI".

What were those decisions, then? One plausible answer wears a central banker's suit. When money was nearly free in the pandemic years, the giants hired as though the boom would run forever, fattening payrolls by some estimates as much as a quarter to three quarters.

Then Fed Chair Jerome Powell made borrowing expensive, the easy money dried up and the bill came due. Powell cut rates repeatedly before handing the chair to Kevin Warsh in May, and still the layoffs have not slowed, proof the cheap-money hangover was only ever half the story.

The cheapest money in the building

So whose hand is on the knife then? Trace the spending and it answers itself. Oracle has poured more than US$70bn into the data centres that keep AI running, borrowed tens of billions to do it, and will not stop losing cash until 2030 at the earliest.

The build-out across the sector now runs past US$700bn. Nobody puts up buildings like that for free, and the fastest money in any company is the wages it stops paying.

Larry leaves no doubt about the aim. Oracle "will build more cloud infrastructure data centres than all its competitors combined", he says in a June earnings call.

Gartner surveyed 350 executives running AI inside their companies this spring. The ones cutting the most jobs posted almost exactly the same returns as the ones cutting the fewest. The productivity miracle, the entire justification for firing anyone, barely shows up in the numbers at all.

A precedent that travels

Oracle is only conspicuous for saying out loud what the rest do quietly. AI has now been the single largest stated reason for job cuts in America three months running, a record 38,579 of them in May alone, according to the outplacement firm Challenger, Gray & Christmas. Add up the year and the country is shedding more than a thousand jobs every working day.

The cuts do not land evenly either. Stanford's economists pull the payroll data and find that employment for software developers aged 22 to 25 has fallen almost 20% since 2022, while their older colleagues in the same roles actually gained. Whatever is doing the cutting is remaking the bottom of the org chart first.

What has been pioneered is not the AI layoff. It is a company admitting to one in writing, and precedents set in securities filings tend to travel.

Expect more boards to reach for the same language, and expect sharper questions about the gap between the story a company tells its investors and the one it tells its staff. The cause may stay unknowable, some blend of the machine, the money and the mood. The disclosure will not.