May 23, 2024 · 39m · no-priors
No Priors Ep 66 | With Y Combinator President and CEO Garry Tan
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Y Combinator President and CEO Garry Tan joins Elad Gil and Sarah Guo to discuss the evolution of YC, actionable playbooks for the artificial intelligence boom, and the necessity of civic engagement in San Francisco. He shares insights on founder selection, capital discipline, startup agility against incumbents, and the enduring power of Silicon Valley networks.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 26.4% of the talking time here. How this is scored →
The hosts as informed peer 5.3 Guest teaching 2.3 Guest disagreement 0.9 The hosts pushing back 0.3
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 12:32 Dismissing the 'ChatGPT wrapper' narrative
Garry emphatically pushes back on the dismissive industry label of 'wrappers', asserting that calling AI apps wrappers is as misguided as calling all cloud software MySQL wrappers.
Hardest push from the hosts ▶ 13:15 Clarifying model IQ vs practical application requirementsSarah directly interrupts and pushes Garry to clarify whether his 85 IQ assessment refers to model limitations today or the actual threshold required for task completion.
Biggest teaching moment ▶ 8:30 The granular mechanics of domain-specific LLM workflowsGarry explains how Casetext broke down legal work into tailored time-and-motion steps to prevent hallucinations and generate actual billable utility with GPT-4.
The host holds their own ▶ 26:43 Market openness dictating founder demographics frameworkElad cleanly synthesizes Garry's observations into a broader theory: open technological frontiers empower younger founders, whereas mature/closed markets demand older domain specialists.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Early Days of YC and Silicon Valley Longevity | 6 | 2 | 0 | 0 | The exchange is highly collegial and nostalgic. Elad Gil brings his own 2007 founder perspective, noting how investors misjudged YC's early batches, while Sarah Guo chimes in with knowledge of mutual acquaintances like Chris Golda. | |
| Parallels Between Early Social Media and AI Waves | 6 | 1 | 1 | 1 | Garry compares the current AI excitement to the early social networking era. Elad matches him by detailing how early mobile photo sharing saw flash-in-the-pan apps before Instagram sustained linear growth. | |
| Pragmatic LLM Implementation and Founder Selection | 5 | 3 | 1 | 0 | Garry breaks down Jake Heller's pragmatic prompt engineering and time-and-motion workflow decomposition at Casetext. Sarah notes their shared investment in Harvey to validate the pattern, while Elad structures the hierarchy of first-hand to fourth-hand customer knowledge. | |
| AI SaaS Wrappers, Intelligence Tiers, and Cohort Growth | 5 | 3 | 2 | 1 | Garry rejects the pejorative label of AI SaaS wrappers, comparing it to dismissing cloud software as MySQL wrappers. Sarah and Elad press gently on model IQ comparisons and top-down corporate AI mandates. | |
| Foundation Models, Scrappiness, and Accountability Systems | 6 | 2 | 0 | 0 | Sarah probes whether YC companies can realistically do capital-intensive foundation model pre-training. Garry explains how Cruise broke down capital requirements into near-term milestones, and explains Paul Buchheit's group office hours as peer accountability. | |
| Startup Agility versus Incumbents and Capital Discipline | 5 | 2 | 1 | 0 | Sarah asks how YC picks startups that avoid being crushed by tech giants. Garry explains that mega tech moats make incumbents complacent and unable to move fast, while warning against raising too much venture debt and equity early on. | |
| Demographic Shifts, Open Platforms, and Deal Dynamics | 6 | 3 | 0 | 0 | Elad asks about historical founder demographic shifts. Garry explains how AI pulls in 19-21 year old dropouts, prompting Elad to synthesize a clear framework: open technological shifts favor young founders, whereas closed domain-specific markets require veteran experience. | |
| San Francisco Density and Garry's Institutional Vision | 4 | 2 | 1 | 0 | Garry delivers an impassioned pitch for San Francisco density and YC's mission to rescue ambitious builders from corporate careers in consulting and investment banking. | |
| Civic Responsibility, Public Education, and SF Governance | 5 | 3 | 2 | 1 | Garry speaks candidly on SF civic issues, particularly criticizing education ideology that removed 8th-grade algebra. Both hosts relate through their public school upbringings and discuss voter guides and local government accountability. |
Statements from this episode (21)
Garry Tan: Y Combinator has 14 group partners funding 15-25 companies each
“We have 14 group partners at YC now, and each of them fund between 15 and 25 companies each.”
Elad Gil: Non-zero-sum behavior early in Silicon Valley compounds over decades
“People forget how much longevity there is sometimes in Silicon Valley, and they forget how you act early really impacts almost how people perceive you 10 years later, 15 years later, et cetera, and so it's, if you're very non-zero-sum early, it really matters …”
Tan: Posterous had a chance to be Instagram without realizing it
“Posterous was funny because we had a chance to be Instagram, but we didn't know it at the time.”
Tan: YC backs technical founders because business mechanics can be taught
“Which actually, I think that's even, you know, going back to the YC, the cult of YC, I'm starting to realize that is actually one of the core things that we've always believed that extremely technical people
those are the people who probably should be, like, …”
Gil: Talking to customers is first-hand information, investors are third-hand
“I would almost view it as like talking to customers is first hand, talking to their founders is second hand, talking to investors is third hand. And if you ever go and talk to an investor, an LP or whatever, it's like fourth hand.”
Tan: Calling AI Startups ChatGPT Wrappers Is Like Calling SaaS MySQL Wrappers
“ChatGPT wrapper thing is actually just wrong. It's like saying that all of SaaS and cloud are basically, you know, MySQL wrappers. It doesn't make any sense.”
Tan: Last YC Batch Grew Aggregate ARR from $6M to $30M
“I think last batch the average ARR of a YC company was six million dollars at the beginning. And at the end, it was, ah, north of 30 mil, and this was over three months. So, to get that kind of growth in a very short amount of time, like, I actually don't thin…”
Tan: Two-Thirds of YC AI Companies Are Applications, One-Third Are Infrastructure
“Yeah, it's about, I mean, two thirds is still like very practical implementation of like ideally frontier models. And then maybe one third increasingly are some sort of infra.”
Tan: YC funds startups building foundation models in bio and robotics
“So we absolutely have companies that are building foundational models. You know, Diffuse Bio is doing over on the bio side. You know there, you know, there's a company working on, like, robotics foundational models.”
Tan: YC group office hours use peer pressure to drive founder accountability
“I think Paul Buchheit, like creator of Gmail, he was one of the first people to make a group office hours, which is sort of using social pressure. It's almost like Alcoholics Anonymous as like, but for founders where it's like you really don't want to You, I m…”
Tan: Mega tech giants are incapable of crushing agile startups
“Actually the mega tech companies have such unassailable moats, and they make so much money that they're actually somewhat incapable of, like, stomping anymore.”
Tan: Raising too much capital prevents founders from making necessary cuts
“Too much money equals potentially just not doing the right things, and then the worst thing is, like, not even being able to cut to the point where you can get back on track.”
Tan: Y Combinator Founder Average Age Is Trending Down Toward 25
“I think that that's actually the average age of YC at this point. It is trending younger. Like, I wouldn't be surprised if this year was this current batch for summer was close to 25 or 26.”
Tan: YC sees surge in 19–21-year-olds dropping out for AI
“So we've seen far more, you know, 1920, 21 year olds dropping out of college to start AI companies now, mainly because if they don't, like, they'll just miss the boom.”
Tan: Median YC startup post-batch fundraising has reached $1.5M–$2M
“The median YC startup, I think In 2015, 20 14, did not raise, like, that much. It probably still was, like, below a million dollars, and then these days, it's, like, close to a million and a half to two.”
Tan: Only SF Directly Connects AI Model Builders With Edge Deployers
“Only here can you know, can the tool makers and the foundational model builders and the researchers literally sit down and break bread with the people way out on the edge, like, trying to practically Deploy these things to every industry and every type of know…”
Tan: Private enterprise cannot fix public school education
“Private enterprise literally cannot help people have a great public school education.”
Tan: Without middle school algebra, I wouldn't have gotten into Stanford
“Like if I couldn't study algebra in middle school, if I couldn't study calculus by senior year, I couldn't have gotten into Stanford. I couldn't have studied computer science, and I never would have been able to participate in, like, what is the craziest thing…”
Tan: SF declined because people feared advocating for basic policies
“I think that SF got very, very bad because you couldn't say basic things like, I believe in algebra for middle schoolers, right?”
Tan: Politically motivated education researchers control math curriculum debate
“There are researchers who are highly politically and ideologically motivated who have Control of the apparatus that is like, you know, sort of quote unquote science, I guess. And then you have real scientists, real mathematicians, real computer scientists at B…”