Oracle plans fresh August layoffs as its AI spending spree bites

The database giant is preparing to cut more staff before its new fiscal quarter begins, even as it borrows tens of billions to build the AI data centres it says are reshaping the company.


Oracle plans fresh August layoffs as its AI spending spree bites

Oracle is planning another round of job cuts this month, according to Business Insider, with some teams reportedly facing double-digit-percentage reductions.

Managers have been told to identify the employees affected, and the goal is to trim payroll before 1 September, the start of the company’s fiscal second quarter. Oracle declined to comment.

It is the latest twist in a story we has been tracking for months: a company shedding people while spending furiously on machines.

The pattern speaks directly to what we have called AI’s real bottleneck, which is no longer chips but capital and consent. Oracle, it seems, is testing how much of both it can stretch.

The fresh cuts land on top of a brutal year. In the fiscal year that ended on 31 May 2026, Oracle eliminated roughly 21,000 jobs, a drop of about 13%, taking headcount from around 162,000 to about 141,000.

The company booked $1.8bn in restructuring charges over the period, with as much as $2.1bn expected before the accounting is done.

What makes the cuts jar is the reason Oracle has offered for them. In its own filings, the firm partly attributed the reductions to the “adoption and deployment of AI technologies.”

In other words, the same technology being sold as a growth engine is also, conveniently, the justification for sending thousands of people home.

Follow the money and the logic becomes clearer, if not more comfortable. Oracle’s capital expenditure hit about $55.7bn in FY2026, up from $21.2bn a year earlier, as it raced to build the data centres underpinning its cloud ambitions. That is an extraordinary sum for a company of Oracle’s size to commit in a single year.

Paying for it has meant reaching deep into the debt markets. Oracle ran an operating cash shortfall of roughly $23.7bn, raised about $43bn in debt over the year, with a further $40bn or so expected, and topped up with $5bn in equity.

The build-out, in short, is being financed on borrowed money rather than out of pocket, a structure that leaves the company exposed if the AI demand it is betting on arrives more slowly than promised.

The underlying business is hardly limping. Revenue grew 17%, and Oracle Cloud Infrastructure, its answer to Amazon and Microsoft, surged 77% as demand for AI computing power kept climbing. On paper, this is a firm winning the very race it is spending so heavily to enter.

Investors are not convinced, though. Oracle’s shares are down about 26% so far this year, a sharp verdict on a strategy that swaps steady profitability for a wager on future AI demand.

The market appears to be asking whether the returns will ever justify the borrowing, and for now it is unimpressed. A quarter wiped off the share price is not the reaction of a market that believes the story.

Strip away the jargon and the arithmetic is stark. Oracle is cutting human costs to help fund the machines, taking on mountains of debt to do it, and watching its own valuation slide while it waits for the bet to pay off. It is a clean illustration of the AI-capex squeeze now running through much of the industry.

Oracle is far from alone in making that trade. Across the sector, firms have been converting payroll into AI capital expenditure, treating salaries as a cost to be redirected toward silicon and server halls. What is unusual about Oracle is how nakedly the numbers lay the choice bare.

For the thousands facing this month’s cuts, the macroeconomic framing offers cold comfort. These are real jobs, in real teams, being removed to hit a deadline set by the calendar and the balance sheet.

Whether the gamble on AI eventually vindicates Oracle’s spending, the people paying for it now will not be around to find out.

Get the TNW newsletter

Get the most important tech news in your inbox each week.