Yesterday we hosted 400 top university students at the YC Summer Conference - a fun day of talks about startups.  Talking to the students afterwards, I found myself giving a lot of the same advice. So in case it's useful to others, here is my startup advice for students.

If you want to understand the evolution of SF tech culture, this is the article to read.

i grew up in Mexico, where it wasn’t always easy to find really high-quality educational resources that pushed me intellectually. i knew pretty early that i wanted to eventually go to one of the best universities i could in the US, and i spent a lot of time trying to get sharper on my own. @brilliantorg was a very meaningful part of that. i still remember being in early high school and absolutely spamming their logic courses. i loved them. they genuinely made me better at thinking, especially around math and logic, and helped build a lot of the skills that eventually got me where i wanted to go. so there is something very surreal about @wafer_ai now serving Brilliant :) their AI tutor, Koji, needs to respond incredibly quickly while students are working through math and coding problems. before Wafer, Brilliant was speculatively prefetching AI-generated responses so students wouldn’t have to wait. on our dedicated GLM-5.2 endpoint, they’re now getting ~250ms time to first token and 300+ output tok/s, 3x the throughput of their previous provider! that let them remove the prefetching layer entirely and cut inference costs by 50%. it’s hard to describe how gratifying it is to help power a product that helped me become a better thinker when i was a kid. very full circle moment. 🧵 Brilliant’s full story in thread.

I've been using @capydotai daily on all my GStack/GBrain PRs and I'm going at least 4x faster than I was when just using raw Codex/Claude Code via Conductor. On a recent PR, this is what it did that out of box Codex wouldn't have done:

Faire I am incredibly proud to have been an investor in @Faire since 2017. I worked closely with founders @maxrhodesOK, @Kolovson, @Mescortes, and Daniele Perito for several years before they started Faire. They are absolutely stellar S-tier talent. What they and the Faire team have accomplished since is extraordinary. This Wall Street Journal piece (linked below) captures the scale of that accomplishment: a $730M+ annualized revenue run rate, 45% year-over-year growth, and an expectation of reaching EBITDA breakeven in Q4. Those are remarkable numbers for a company taking on an enormous and stubborn problem. Wholesale has changed remarkably little for more than a century. Independent retailers still spend huge amounts of time discovering products, managing inventory, and absorbing costs that large chains spread across thousands of stores. Faire is changing that for hundreds of thousands of retailers and brands. The company is now on pace to facilitate more than $4.5B in merchandise purchases annually. Every order adds to Faire’s understanding of which products sell in which stores. That information improves product recommendations, gives retailers more confidence in what they buy, and connects brands with customers they would struggle to find on their own. I’m especially excited by how much opportunity remains. International markets give Faire much more room to grow. Advertising, launched in 2024, already contributes about 9% of revenue. Fulfillment could reduce customer shipping costs by roughly 80%, directly improving retailer margins. For Faire employees, I hope this article brings a deep sense of pride. You have built something truly special. For candidates, Faire offers a rare opportunity: proven demand, enormous room to grow, hard technical and operational problems, and a founding team that I would back again without hesitation. Very few teams get to build a company of this scale while improving the odds for independent retailers and brands around the world. Faire does both. I feel incredibly fortunate to have been part of the story from the beginning. I have never been more excited about what this exceptional team will build next.

YC has funded more than a dozen data and RL environment startups in the last two years that are doing >$10M+ per year (and many >$100M per year) in revenue selling to the labs. (via @garrytan on YC's Lightcone podcast)

RT @nico_laqua: Cancel culture is alive and well, but, at Corgi, we have thick skin and believe strongly in freedom of expression. For tho…

Surprised by how quickly YC itself improves. I was previously in the spring batch and just finished F26 kickoff. Even in that short time, everything felt more organized, efficient, and thoughtfully designed from onboarding to picking up the initial merch. Bookface has also become significantly richer in features and has streamlined a lot of the administrative work of running a company and fundraising . And the food so far has gotten much better too. Cool to see YC iterating as fast as its startups.

great to be here :)

When I was building YC's conversational AI, I benchmarked every LLM and inference cloud I could get my hands on. In my head-to-head benchmark, the top performer was GLM-5.2 on Wafer.

Computer use is what matters most for a great consumer agent experience and it’s what Muse really nailed. Meta has crushed companies in the past with distribution over product but this time they actually built the best product.

Look at that latency 😍 As proven by 25+ years of internet services, speed is a feature. For any app where there is a consumer (or agent!) on the other side waiting for the result, being even a bit faster results in higher conversion (and lower drop-off / churn) @wafer_ai is the choice for low latency inference 📈🚀

YC companies raise money differently than the rest of the venture market. They raise earlier, they raise faster, and the same top group of investors come back round after round. 1) EARLIER and BIGGER. The median YC seed round came in around $4.6M, with the top decile past $10M. A decade ago a $10M round was a Series A. The stage labels inflated, and YC sits at the front of that inflation. 2) FASTER. Pre-seed companies were a median of one year from founding, seed two years, Series A three. A YC company that stays on pace raises roughly every 18-24 months. That's not normal. That's a conveyor belt. 3) REPEAT INVESTORS (two tiers) One tier is YC-alumni money, and funds like Lobster Capital with strong batch relationships. The other tier, a16z, Sequoia, General Catalyst, arrives at Series A and again later, when the checks get big. So a YC company moves through a relay. - Alumni vehicles fund it out of the batch. - If it clears the bar, the brand-name firms pick it up at A. - And again and again. Each handoff is a signal to the next investor, and the same faces recur because the system is designed for them to. The YC premium everyone complains about is because you're buying into a pre-wired funding relay where the first checks come and the later checks already know to look. That's part of what you're paying up for. The machine raises earlier, faster, and from the same hands every time. Which means the machine also prices efficiently... The only thing left to compete on is judgment.

On what made YC great c. 2009: - yc greatness is best understood by comparing it to trad ecosystem at that time - the Pioneer Way yc office was surrounded by trees, equipment suppliers, small scale logistics companies, and of course startups. People were loading and unloading trucks around the corner. By contrast trad VC offices were tucked away in country club grounds on Sand Hill Rd - YC parking lot must have had cars normal ppl buy because I never noticed one. On Sand Hill you couldn't help but notice Porsches, Maseratis and so on - when you entered YC you'd walk through a white empty vestibule they never quite knew what to do with, then enter the cheery "orange room" filled with tables and benches for founder dinners. Sand Hill offices were somber forgettable places set up by professional decorators who would never have to work there themselves - at YC you were greeted by pg in his cargo shorts and polo, and Jessica always brimming with energy - on Sand Hill you were greeted by an immaculately dressed receptionist who looked like she'd been sent over by central casting. (It was always a woman; I've been rejected dozens of times by venture capitalists on Sand Hill, and I don't think I was greeted by a man once.) She'd offer coffee. You'd decline. She'd walk you to a conference room. The walk was usually long and silent; the only sound you'd hear was the loud clicking of her high heels against the floor. Then you'd settle in around an oversized mahogany table and wait. The wait was never so long that you could call them on it, but somehow always long enough that it felt disrespectful - in fairness you sometimes had to wait at YC too. This was because Paul was walking around somewhere with a founder, enthralled by an idea they were both brainstorming. And in a few minutes he'd be equally enthralled by an idea he was about to be brainstorming with you - while waiting for pg you'd look around. On one side there was tlb's Anybots lab, stocked with all kinds of electronics and robot parts. On the other side there were rice cookers (or were they pressure cookers?) pg and co used to cook founders dinners. A Sand Hill partner might take you to a Michelin star restaurant, but nobody on Sand Hill was ever going to cook for you - there was a structural element. YC partners were writing small checks and treated it as an experiment. They could be very tough on you if you screwed up, but it always felt like they stood next to you, helping you peer into the future to make the thing work - but partnership in a Sand Hill firm was not an experiment. It was a highly desirable career path, with all the implications that come with that. Almost everyone who worked there was afraid to make a mistake or to look dumb in front of their partners. And so on Sand Hill founders felt like they were looked down upon from Mt Olympus. (What we were too green to realize is that sitting on Mt Olympus is a very precarious thing) - pg & tlb were hackers. Not in their distant past-- now. Tlb was building robots. Pg was building HN on top of Arc on top of Scheme. It was very difficult to imagine anyone on Sand Hill doing that - anyone could see the bones of yc-- they were the OGs of building in public. Pg's essays and HN comments would document the choices, the problems, the decisions. Since he worked out many ideas right there on the page, he'd implicitly document his way of thinking and the attitude of self-belief, curiosity, and optimism that's so indispensable while working in the fog - eventually trad firms would clone elements of YC. Seed funds, batches, blogs, and so on. It worked, but only to a point. Sand Hill ppl would often ask-- what’s the secret? We’re doing all the same stuff, why are you founders so loyal to YC but not to us? Of course everything was out in the open for anyone to see, there wasn't anything insightful to add /fin

Current YC batch is one of the best of all time. Also one of the most undervalued.

YC S26 batch is THE strongest batch ever! Let’s go @ycombinator 🔥🔥🔥

S26 is an exceptional @ycombinator batch; everyone that meets the companies says so. Honored to have worked closely with 36 of them this batch. Some will become generational companies, no doubt 🫡

Ive been using this for a couple weeks and its the rare instance of a new product from a bigco that’s obviously built by people who actually care. The browser/credentials/payments details are all really well done and the speed + accuracy of the intelligence is 👌

yc takes two technical co-founders and quietly turns one of them into a salesperson. happened to me. during the batch, i remember thinking there's no way i'm going to do sales. now it might be my favorite part of the job. if i had a kid going to college right now, i'd suggest they study computer science, then go into sales anyway. being technical and being able to sell is a rare combination, and it compounds like crazy.

Introducing the YC Early Access Network. 4x a year, senior technology leaders get early looks at YC's enterprise AI companies and help shape their roadmaps. Invite-only, by application. Fall '26 kicks off Oct 22 in SF. Apply now: events.ycombinator.com/yc-ea…

The @ycombinator S26 might have been the strongest batch ever. It's hard to believe YC could get much better, but I think with this batch, they really really did! Below are my investments in the S26 Batch. Take a peek 👀 👇🏻