The people building the agent economy don't have degrees yet, and they're betting Stanford will still be there when the music stops.
The Summary
- Stanford economics professor Nicholas Bloom says at least five of his students have paused their degrees to join early-stage AI startups — a higher rate than any prior year he's taught.
- The trend mirrors the dot-com boom of the early 2000s, when undergrads flooded internet startups, treating education as optionality rather than prerequisite.
- Students see AI as a rare window where undergrads can work at the cutting edge, with valuations high enough to justify the pause.
The Signal
Five students out of a couple hundred doesn't sound like a mass exodus. But Nicholas Bloom teaches economics at Stanford, where he's watched cohorts come and go for years. He says this year is different. Not since the early 2000s has he seen undergrads treating their degree as something you can put on ice while you chase the next platform shift.
Stanford doesn't track why students take leaves of absence, so there's no hard data on how many are pausing for AI startups versus burnout, travel, or family reasons. But Bloom's observation matters because it's specific: these aren't seniors hedging post-grad plans. These are students mid-degree who see a window closing.
"AI is so new that undergrads can potentially be at the cutting edge of this, and the valuations are so high that it's worth it for them to pause even for an early-stage idea."
The comparison to the dot-com era is instructive. Back then, you could learn HTML over a weekend and build something investors would fund. The barrier to entry was knowledge, and young people had time to acquire it faster than incumbents could adapt. AI in 2026 has a similar shape: the tools are democratized, the playbook is unwritten, and being 22 with no legacy code to maintain is an advantage, not a liability.
What's different this time is the speed and the stakes. In 2000, you built a website. In 2026, you're building agents that negotiate, transact, and operate semi-autonomously. The companies forming now aren't just platforms. They're infrastructure for a new layer of economic activity. Students aren't dropping out to join Google. They're dropping out to build the companies Google will try to acquire in three years.
Key dynamics making this rational for students:
- AI startup valuations are compressed upward by enterprise urgency and VC fear of missing Web4.
- Undergrads can contribute real value to early-stage teams building LLM wrappers, agentic workflows, or vertical AI tools.
- A failed startup at 21 is a learning experience. A failed startup at 35 with a mortgage is a crisis.
Bloom notes students view this as a "reversible bet", not a permanent detour. Stanford will readmit them. Their friends will still be there. But the AI boom might not be. That asymmetry makes the decision easier than it should be.
The Implication
If Stanford undergrads are treating their degrees as pause-able, it's because the market is signaling that execution beats credentials right now. For students at non-elite schools, the calculus is harder. You don't have the same safety net. But the opportunity is the same: AI is still new enough that a smart 20-year-old with taste and hustle can build something that matters.
For everyone else, watch what happens when this cohort hits the job market in three years. Either they'll have built the next wave of agent infrastructure and won't need traditional employment, or they'll have learned how to ship product under uncertainty, which is worth more than any degree. Either way, the people hiring in 2029 will be looking for builders, not graduates.