Y Combinator Founder Paul Graham: Don't Expect to Start a Company and Then Go or Return to University; the Two Paths Are Incompatible
Y Combinator co-founder Paul Graham has been posting repeatedly: do not expect to start a company and then go or return to university; the two paths are incompatible, and pretending they are compatible is lying to oneself. The only compatible scenario is failure in entrepreneurship; if you still want to go to school, it means you are secretly hoping for failure, and entrepreneurship is hard enough that if you wish to fail, you will definitely fail.
He breaks down the stages. More and more people are starting companies right after high school, which he judges is usually wrong, but the individuals do not listen. If one leaves after three years of university, the damage is much smaller—undergraduate programs in the UK are typically three years, and it doesn't even count as dropping out there. Project stages can run parallel to studies; once a project becomes a company, you cannot do both. University theoretically allows registration at any age, but being two or three years older than peers means the experience is no longer the same.
This set of remarks is not a sudden shift. In July 2025, he wrote: do not drop out to start a company or work for a startup; there will be other, often better, opportunities for entrepreneurship later, and those university years cannot be reclaimed. In March 2025, he told a high school student that if one truly has ambition, one can learn more in university than by joining a startup, because work in startups is severely constrained, and a lack of constraints indicates inefficiency in the company. In April 2026, he wrote: do not start a company in high school; if you succeed, you lose the window of exploration driven by curiosity; entrepreneurship can teach a lot, but what is taught is determined by the company. In May 2026, he observed from Garry Tan's office that those who drop out at 18 or skip university are more likely to fall into the category of "adequate skills, no demand, avoiding real users" because they haven't had time to accumulate "earned insights."
His own path is not symmetrical with this advice. Graham completed his undergraduate studies at Cornell and earned a PhD in computer science from Harvard. In 1995, he co-founded Viaweb, which was sold to Yahoo in 1998, and in 2005, he established Y Combinator, investing in over 3,000 companies including Airbnb, Dropbox, Stripe, and Reddit. He has long viewed the age of 20 as a breadth-first search window and entrepreneurship as a brutal depth-first search; success will close the exploration window. In April 2026, he added a responsibility clause: if one skips part or all of university to start a company, one must complete the mental training that university would have provided, which most startups default to not doing.
In terms of market mechanisms, buyers want young founders to deliver full-time immediately, through accelerators, seed funds, and the narrative of "starting a business in high school"; sellers are students treating university as an option and companies as summer projects. The event-driven narrative sells both the dropout myth and the halo of prestigious schools simultaneously. Funding flows from Y Combinator-style batches to teams that have already established companies and cannot return to the classroom; on the other end, those who want to keep a safety net will treat their companies as half-projects, reducing the probability of securing real venture capital. Beneficiaries are founders who have passed the exploration phase and can translate insights into products; those under pressure are high school entrepreneurs who write "returning to school after failure" as plan A, and early funds that attract clients through dropout stories but require founders to be online seven days a week.
Source: Public Information
ABAB AI Insight
Graham's series of posts serves as a brake on the culture Y Combinator has created. The accelerator has made "full-time now" the default, while high school entrepreneurship and Thiel-style dropout scholarships push this default even earlier; he now emphasizes that once a company is established, the fallback is not campus, but liquidation. Before selling Viaweb, he had already completed his PhD, and his advice comes from a post-factum option, not from the perspective of an 18-year-old.
The capital path is clear. Seed rounds buy all the workdays on the founder's calendar, not "side projects"; once the terms take effect, the desire to attend school will be seen by the board as giving up options. Those who want to balance both will keep their companies in project status, unable to raise the money that requires exclusive commitments; those who want to retain university will subconsciously write failure as a safe exit, thus undermining execution. Y Combinator batches are the execution end of this rule: compatibility only exists in failure samples.
The analogy is not the dormitory legends of Bill Gates or Mark Zuckerberg, but Graham's own search theory in his writings: at 20, one should explore broadly; entrepreneurship is depth-first search, and once successful, one cannot go back. Founders of Stripe and Airbnb were already past the stage of "freely changing topics" when they launched. High school entrepreneurship is more akin to moving depth search to a time when there are no topics yet. The industry position reflects an overheated labor market for founders: narratives reward younger individuals, while the distribution of experience still favors those who have studied for a few years before leaving.
Structural judgment pertains to the transfer of pricing power. What is being repriced is not the degree itself, but the exclusivity of the founder's time. The mechanism is: venture capital only pays for the calendar of those who are "already a company"; universities only issue certificates for the calendar of those who are "not yet a company"; the two calendars cannot overlap in the same week. Anyone who still writes overlapping calendars as a plan is implicitly setting failure as a goal. The price of the entrepreneurial market ultimately hinges on whether you are willing to burn that ticket back to school.
ABAB News · Cognitive Laws
Entrepreneurship with a safety net will first kill the company.
Projects can attend classes; companies can only attend classes to fail.
Trying to balance both paths is equivalent to writing failure into the business plan.