The Question AI's Biggest Believers Can't Answer
Last week, the biggest names in tech gathered in San Francisco for Morgan Stanley's annual TMT Conference. Record earnings. Soaring stock prices. An AI arms race is minting new fortunes at historic speed. The mood was, by most measures, triumphant.
And yet one question kept coming up — in hallways, in investor panels, over drinks — that no amount of compute spend could resolve.
What will our kids do?
Morgan Stanley analyst Adam Jonas flagged it as the single most common question he fielded throughout the conference. For all the optimism about large language models and agentic AI, a quiet anxiety about the next generation of workers ran through a room full of some of the most powerful business leaders on earth.
Sam Altman said the quiet part out loud
OpenAI's CEO told conference attendees he can envision one or five people running an entire company — and said that transition is now measured in years, not decades. Days earlier, at a summit in India, he had been even blunter: "The world is not prepared. We are going to have extremely capable models soon. It's going to be a faster takeoff than I originally thought."
Jensen Huang distilled the moment to three words: "Compute equals revenue." NVIDIA's CEO described demand for computing power as "higher than incredibly high," with Amazon Web Services ramping "like mad" and major AI labs needing millions of net new GPUs.
Meanwhile, OpenAI's latest model release posted record scores across AI evaluations — the kind of capability leap that Morgan Stanley's analysts say the market still hasn't fully priced in.
The layoffs CEOs actually talked about
What made this year's conference different wasn't the optimism — that's a constant. It was the candor.
Multiple executives described, in clinical detail, the AI-driven efficiencies behind their recent workforce reductions. A Morgan Stanley survey of roughly 1,000 executives across five countries found an average net workforce reduction of 4% over the past twelve months, directly attributable to AI adoption. And that covers only the sectors where AI is currently most advanced.
The economics are now showing up in aggregate data, not just case studies. Harvard economist Jason Furman and Stanford's Erik Brynjolfsson — previously cautious about drawing broad conclusions from productivity research — now agree that AI gains are registering in the macro numbers. University of Chicago economist Alex Imas, whose work Morgan Stanley highlighted at the conference, described himself as "amazed and alarmed." As a researcher, he said, he can do things he's never been able to do. As a parent, he's "super worried about what sort of jobs" his kids will have.
That tension — extraordinary capability, uncertain human consequence — was the subtext of the entire conference.
Who wins, and who doesn't
Morgan Stanley's analysts were direct about the distributional picture. High-income consumers, whose portfolios are swelling with AI-driven gains, are expected to increase spending. Middle- and upper-middle-income consumers, whose jobs are most exposed to automation, are expected to pull back.
Assets that AI can't replicate — luxury experiences, rare earths, proprietary data, authentic human connection — are expected to hold or climb in value. Everything else faces deflationary pressure as AI replicates human work at a fraction of the cost. The bank said it has been "continually surprised at how quickly, and violently," this dynamic has become a driver of stock performance across sectors.
"2026 is gonna be insane"
The most striking quote of the conference didn't come from a keynote. It came from a retirement announcement. Jimmy Ba, cofounder of xAI, said upon stepping down, "Recursive self-improvement loops likely do live in the next 12 months. It's time to recalibrate my gradient in the big picture. 2026 is gonna be insane and likely the busiest and most consequential year for the future of our species."
Several executives at major AI labs echoed the sentiment, warning that near-term progress would "surprise, and potentially shock, investors." Morgan Stanley's own analysts said they expect a non-linear jump in model capabilities to become evident between April and June of this year.
The machines are getting smarter faster than almost anyone predicted. The people building them know it, the investors funding them are starting to feel it, and the question of what the next generation will actually do for work, income, and identity remains, for now, genuinely open.