In a memo to employees viewed by CNBC Wednesday (June 3), Uber CEO Dara Khosrowshahi said the “changes are necessary to maximize the effectiveness of the People team and the enormous potential ahead of us.”
The People operation serves as Uber’s recruitment human resources arm. A spokesperson for the company told CNBC that the cuts applied to under 1% of Uber’s 34,000-person workforce.
The report also cited a note to the affected teams from Uber President Jill Hazelbaker, who said the layoffs are meant to foster a “more connected, modern, operationally excellent organization.”
Some parts of the company have become “complex and fragmented, with overlapping responsibilities, unclear ownership, and teams operating too far from the businesses and partners they support,” added Hazelbaker, a recent addition to Uber.
With these cuts, Uber joins a growing list of companies laying off workers recently. Many of these companies have cited the benefit of embracing artificial intelligence (AI) automation. Uber is not one of these, though it is wrestling with rising AI costs.
Advertisement: Scroll to Continue
A report this week from Bloomberg News said the company has begun limiting workers employees to $1,500 in monthly token spending per AI coding tool.
The limits apply only to agentic coding software, and is on a per-tool basis, meaning employees can spend $1,500 on each different tool. The report said Uber gives employees a dashboard that lets them track their use and ask permission to go over the limits.
“We think this is all a pretty straightforward way to responsibly encourage agentic AI adoption and experimentation at scale across the company,” an Uber spokesperson told Bloomberg.
The change came after Uber reportedly exhausted its annual AI budget in a matter of months. Other companies said to have imposed employee AI limits include Walmart and Microsoft.
As covered here earlier this year, companies launching production-scale AI deployments have begun to find that enterprise AI is tethered to billing models that scale just as elastically as earlier ones, though with less transparency.
“The commercial infrastructure underpinning traditional software-as-a-service (SaaS) doesn’t translate cleanly to systems that meter value by computation rather than by user,” PYMNTS wrote in February.
“In the case of AI, billing models charge by token, per application programming interface (API) call, per generated image, per inference cycle, per autonomous workflow, or, in some cases, for all of them simultaneously.”
