The trade-off is already here: companies get AI superpowers, junior workers get underemployment.
The Summary
- Uber is slowing hiring to fund AI investment, with autonomous agents now producing 10% of the company's code changes
- 42% of recent college graduates remain underemployed as learning AI becomes essential for getting hired
- Uber expects employees to increase productivity by 20-100% using AI tools across legal, marketing, and development teams
- The productivity gains from AI agents aren't translating to job creation, they're replacing the need for new hires
The Signal
Uber CEO Dara Khosrowshahi just made the quiet part loud. The company is redirecting money that would have gone to new employees into AI infrastructure instead. This isn't a future scenario. It's happening in Q2 2026, at a company that just posted $13.2 billion in revenue and 25% growth in gross bookings.
The math is stark: 10% of Uber's code changes now come from autonomous agents. Humans still review before the code ships, but the work of generating that code, the entry-level programming grunt work that used to be a reliable on-ramp into tech, is increasingly done by software. Khosrowshahi frames this as giving employees "superpowers," and he's not wrong about the productivity gains. But superpowers for existing employees and opportunities for new ones are different problems.
"We're seeing uptake of these tools, whether it's our legal team or marketing team or developers."
The tools are proliferating across functions, not just engineering. Legal teams, marketing teams, developers. Every department where junior employees used to learn by doing repetitive tasks that built skill and pattern recognition. Those tasks are getting compressed or eliminated. Clara Shih, former Head of Business at Meta and founder of the New Work Foundation, points to the human cost: millions of 25-year-olds unemployed, 42% of recent grads underemployed even as companies post record earnings.
Shih argues AI needs to be profitable "not just for businesses, but for everyone." But the incentive structure points the other direction. Uber's stock jumped 6% in pre-market trading after the earnings call. Investors rewarded the efficiency story. Hire less, produce more, expand margins. The market signal is clear.
Key tensions emerging:
- Companies measure AI success by productivity per existing employee, not by jobs created
- Learning AI is becoming table stakes for new grads, but there are fewer entry-level seats at the table
- The wedge between corporate AI gains and worker AI gains is widening, not narrowing
This isn't a story about AI replacing all jobs. It's about AI changing which jobs exist and who gets access to them. The companies deploying agents most aggressively are the same ones pulling back on hiring. The graduates most affected are the ones who would have learned on the job, the ones who needed that first role to prove they could contribute. AI literacy might be necessary, but it's not sufficient when the jobs themselves are disappearing.
The Implication
If you're hiring, you're now competing with the idea that you could just give your existing team better tools instead. If you're looking for work, "I know AI" isn't enough. You need to show you can do something an agent can't, or that you can orchestrate agents better than the next candidate. The gap between Web2 employment models and Web4 productivity is getting uncomfortable fast, and neither companies nor workers have figured out the bridge.
Watch for the next shoe to drop: companies that slow hiring today will face talent crunches tomorrow when agents hit their ceiling and they realize they have no junior bench. But by then, an entire cohort will have aged out of entry-level roles.