š§š Garry Tan on the Past, Present, and Future of YC
Inside YC's 20-year evolution, how to ace the YC interview, lessons from Brian Chesky, a 55x DPI Fund 1, tech optimism, and why everyone should care about local politics
Garry Tan has lived every side of the YCombinator ecosystem, which has invested in 20% of all startups worth $5 billion or more started since 2012.
Garry first went through YC as a founder, became a YC partner in 2010, started Initialized Capital in 2012 which put up a 55x DPI Fund 1 backing YC companies like Coinbase and Instacart, and then returned a decade later to lead YC as President & CEO.
This latest episode of The Peel walks through the three eras of YC. We start in the early days of Paul Graham and Jessica Livingston getting YC off the ground in Cambridge, moving to and scaling in San Francisco, to todayās push back toward in-person community and what Gary calls āfounder modeā for YC itself.
We talk about why the Bay Area matters so much for startups, whatās happening with taxes and policies in California, and why Gary has gotten more involved in local politics to keep it the best place for founders to build companies.
We also go deep on the parts of startups that people donāt talk about enough. Co-founder conflict, rage quitting, therapy, coaching, and why companies inevitably take on the personality and emotions of their founders.
We cover what YC looks for in applications, how to ace the YC interview, what the 13 week batch is really like, how Demo Day actually works, how to choose the right investors, and what Gary thinks the next phase of YC looks like, including re-batching founders after their Series A.
I also get Gary to share his personal AI workflow. We talk about meta prompting, comparing outputs across models, and the tools he uses every day to think and build faster.
Support this Episodeās Sponsors
Numeral: The end-to-end platform for sales tax and compliance.
Flex: Sign-up for Flex Elite with code TURNER, get $1,000. Apply here.
To inquire about sponsoring future episodes, click here.
š Stream on Spotify and Apple
Timestamps to jump in:
0:05 Moving from Winnipeg to California as a kid
1:35 How YC interviews work
2:55 The first batch in 2005
6:46 Why YC moved from Boston to SF
8:17 Californiaās Billionaire Tax
11:00 Tech should care about public policies
17:01 Going direct to your audience
20:28 The 2nd Era of YC
24:01 Rage quitting Palantir, learning to understand himself
32:41 Co-founder conflict kills most startups
35:15 Joining YC as a group partner
37:22 Initialized Fund 1 (55x DPI)
39:44 Why Garry went back to lead YC
42:44 YC funds 20% of all $5B+ companies
44:30 Lessons from Brian Chesky
48:01 Garryās thoughts on YC rejection
51:41 How to get into YC
58:03 What itās like inside a 13-week YC batch
1:02:23 20% of YC is hard tech
1:05:55 YC's 3rd era: founder mode, re-batching
1:07:56 Escaping the matrix
1:11:26 Garry's personal AI stack
1:20:25 Tech optimism
Referenced:
Kyle Vogt on his new startup
Follow Aaron Levie on X
Find Garry on X / Twitter and LinkedIn
Related Episodes
š Stream on YouTube, Spotify, and Apple
Transcript
Find transcripts of all prior episodes here.
Turner Novak:
Gary, welcome to the show.
Garry Tan:
Thanks for having me.
Turner Novak:
I realized we are some of the only people in tech that are from Winnipeg.
Garry Tan:
Right on.
Turner Novak:
Originally.
Garry Tan:
Yeah. I was named after Fort Garry with two Rs. My sonās name is Garrison, which is kind of funny because then itās-
Turner Novak:
Itās kind of the same thing.
Garry Tan:
Yeah. Anyway.
Turner Novak:
Nice. Then how did you end up in San Francisco From Winnipeg.
Garry Tan:
My dad moved to work in the satellite boom back in the ā80s, and so he moved to Southern California and we moved around a lot. But yeah, I mean basically California and tech drew our family and put food on the table and then it brought me into tech.
Turner Novak:
And then you stayed, obviously, in San Francisco. What was that journey like?
Garry Tan:
Yeah, we ended up moving to Fremont, which was, itās in the East Bay, and I took Bart in to take my first computer science classes at UC, Berkeley. I took Bart in the 16th admission down the street right here. And I got my first coding job that way. This was all Web 1.0. I worked at a design firm that created the first Apple eCommerce store for Steve Jobs called Adjacency. And so I mean, tech gave me everything and Iām so lucky to just be able to participate in this.
Turner Novak:
Yeah. And then today, everyone probably is familiar with you. Youāre leading YC. Really quick, for people who donāt know what YC is or maybe people who do, how do you describe it today when someone asks you, āWhat is YC?ā
Garry Tan:
I mean, itās an accelerator incubator. People apply online so you donāt have to know anyone. And we accept about 1% of the people who apply, but when you get in, you get half a million dollars in exchange for about seven-ish percent of the company. But whatās more important than the money is actually getting a YC partner like me. Iām one of 15 people who goes out there and selects companies.
Turner Novak:
Thereās 15.
Garry Tan:
We pick you out of the database, we read your application, we watch your video, we pick specific teams to meet us for ... we meet for 10 minutes and we have to decide yes or no on half a million dollars within 10 minutes.
Turner Novak:
Itās crazy.
Garry Tan:
It almost sounds like Shark Tank, but actually Shark Tank has, I think, zero billion companies, whereas weāre going on more than 100 at this point 20 years in.
Turner Novak:
Really?
Garry Tan:
Yeah.
Turner Novak:
Thatās crazy.
Garry Tan:
Yeah. So YC has started as this experiment. Paul Graham and Jessica Livingston started it really just as an experiment, like how do we give small amounts of money to people? Literally almost summer internship style. When I did YC, we only got $13,000 or so. So itās 500k now.
Turner Novak:
Thatās like last thing you get from an internship. Yeah.
Garry Tan:
Thatās right. But thatās all you needed. And that very first YC batch had Reddit with Steve Huffman and Alexis Ohanian. That first batch had Sam Altman who created Loopt, which got funded by Sequoia. Another company sold to Amazon, I believe. So that very first batch turned out to be a success and they just decided, letās keep doing it.
Turner Novak:
Was Twitch in the first batch or early?
Garry Tan:
Yeah. The founders of Twitch started Kiko, which they sold that startup on eBay.
Turner Novak:
What? Somebody listed it and someone bought it?
Garry Tan:
Yeah, I think they made ... Yeah, they listed it on eBay and bought it. And then the funny thing is when they started working on Twitch, it started as Justin.tv and they got Kyle Vogtās attention. He was an undergrad at MIT, and the other ... Justin and Emmett and Michael and the other co-founders of Twitch, they were from Yale, but they had sold a company on eBay. And so my favorite story, itās not my story, itās Kyleās story, that he found out about Justin.tv because in that very first YC batch, that team had sold a startup on eBay and that was way, way more about startups than Kyle felt like he knew. And then of course, Kyle went on to create many multi-billion dollar companies now in Cruise. And heās on his journey in robotics now with Bot Co, which is, I mean, I wish I could say anything about Bot Co like ... You have to see it. Itās insane.
Turner Novak:
Is it humanoid or itās in the home, youāre not allowed to say.
Garry Tan:
Iām under NDA but-
Turner Novak:
Under NDA. Okay.
Garry Tan:
... it is truly one of the most remarkable things Iāve seen in startups in maybe ever.
Turner Novak:
There is some information out there. People can look it up, Iāll throw whatever I can find in the description. People can dig around.
This episode is brought to you by Numeral. Numeral is the fastest, easiest way to stay compliant with US sales tax and global VAT. Itās easy to set up and they automatically handle all registrations, ongoing filings, and their API provides sales tax rates wherever you need them with all the integrations you need.
Numeral sports over 2,000 customers in both the US and globally. And they pride themselves on white glove, high-touch customer service. Plus, they guarantee their work and theyāll cover the difference if they mess anything up. Theyāre fresh off of fundraise, closing a $35 million series B from Mayfield, which theyāre going to reinvest into building an even better product.
If you want to put your sales tax on autopilot, check out Numeral at their new domain, numeral.com. Thatās N-U-M-E-R-A-L dot com for the end-to-end platform for sales tax and VAT compliance.
This episode is brought to you by Flex. Itās the AI-native private bank for business owners. I use Flex personally and I love it because they use AI to underwrite the cash flow of your business, giving you a real credit line. The best part is 60 days afloat. Double the industry standard.
Flex has all the features youād expect from a modern financial platform like unlimited cards, expense management, bill pay that syncs with your credit line and their new consumer card, Flex Elite. Flex Elite is a brand new ramp-like experience for your personal life, a credit card with points, premium perks, concierge services, personal banking, cars and expense management for your family, net worth tracking across public and private assets, and a whole lot more fully integrated with your business spend. One card for your businesses, one card for your personal life, one card for everything.
To skip the wait list, head to flex.one and use my code TURNER to get an additional 100,000 points worth $1,000 after spending your first $10,000 with Flex Elite. Thatās flex.one and code TURNER for $1,000 on your first $10,000 of spend. Thank you, Flex. And now letās jump in.
And so this was all in Boston, that first batch, right?
Garry Tan:
Oh yeah. Thatās right. And I was the last batch. I did YC in 2008, so we were the last batch in Cambridge.
Turner Novak:
So then why move to San Francisco? How does this happen?
Garry Tan:
This was before Paul and Jessica had kids and so they really love to spend basically summers in Mountain View because of the California weather and winter in Massachusetts. Itās pretty brutal. But they liked summers in Cambridge and so we did the last summer batch in 2008. And then itās funny because itās coming full circle. We just added our first partner whoās actually based in Cambridge, Ankit Gupta. So he sold a startup. It was a great company. He sold it to Ginkgo Bioworks and then heās based basically right next to MIT.
And so weāre sort of thinking about opening an office in Cambridge and given all of what weāre seeing out there around California being more and more hostile to tech, weāre realizing Iām all in on San Francisco, Iām all in on California and weāre going to fight for it. But from an institutional standpoint, theyāre going to kill the golden goose here. And we got to think about where are founders going to be.
Turner Novak:
Yeah. Okay. So it begs an interesting question. If somebodyās maybe out of the loop and they donāt know what youāre talking about, whatās going on in San Francisco specifically or in California?
Garry Tan:
In California broadly, SEIU-UHW, which is a particular union for healthcare workers just put on the ballot or theyāre trying to get on the ballot what they call the billionaire tax. I call it the asset seizure tax. Basically, itās a one-time tax of 5% if your net worth is above a billion dollars. So that sounds kind of reasonable, but if you read the fine print, itās tailored designed to seize the assets of startup founders. So Larry-
Turner Novak:
So how can that be true?
Garry Tan:
I mean Larry and Sergey both left the state because of it in the off chance. I mean I think itās a real double-digit mid 40, 50% chance that this actually gets passed in California. So I think itās something we have to take seriously. Itās not this fringe thing and the way itās written, it actually doesnāt take your ownership. Larry and Sergey have about 3% each of Alphabet. They actually take your voting percentage as your ownership percentage.
Turner Novak:
And they still have control, donāt they? Like voting control?
Garry Tan:
Yeah. So they would get taxed 5% on 60% of the value of Alphabet.
Turner Novak:
Which is literally their entire state.
Garry Tan:
It would be half of their net worth. So I mean itās very strange because itās turned into this war. The drafters of it claim that it would not do this, and they have all of these sophist arguments around how this wonāt apply. But-
Turner Novak:
Why donāt say just change it? That seems obvious. If you look at the rules based on itās written, it will apply this way. Just change how youāre writing it.
Garry Tan:
Well, thatās one of the artifacts of the proposition system. This isnāt going through legislation. Itās not going through the Assembly or the Senate. This is not something that Gavin Newsom can veto. Itās actually going to a direct proposition on the ballot probably for November.
Turner Novak:
If it gets voted and passed, it just gets bolted onto the California Constitution?
Garry Tan:
Yeah. Thatās how it works. I mean, this is how we got Prop 13. A lot of things are broken about California, but this is one of those. SEIU-UHW might change it, but I think in the background, I think people who care about basically the state and who want tech to thrive in the state, we actually have to get organized. And thatās what Iām going to ... Iām going to keep working on that.
Turner Novak:
So I think what seems to be happening is instead of paying the tax people just leave.
Garry Tan:
Yup.
Turner Novak:
Is that a natural thing that you think could happen?
Garry Tan:
Thatās happening right now. I think about a trillion dollars in personal net worth has left the state, which is about half of the billionaires in California. And the result-
Turner Novak:
Youāre getting half the effect you thought you were getting when you put this bill into place.
Garry Tan:
Yeah. And when you look at taxes, 76% of taxes are paid by the top 10%. Iām not saying that thatās wrong, but what do you do when more than a trillion dollars of net worth leaves the state? It means that middle-class people, everyone else has to pay a lot more. And so itās hilarious. Wall Street Journal had a quote from a wealth manager whoās a mere millionaire, which is crazy to me. And he says, āWell, I donāt care. It doesnāt affect me.ā But the reality is Iām also not a billionaire. The thing is, everyone will end up paying taxes or services that are really important will get cut.
And so when you have bad policy, we just have to work against it. Weāve got to vote the right way. And so I donāt know, coming up in tech, I never paid attention to this stuff and then now, Iām starting to realize, no, we have to pay attention. Weāre sitting here in San Francisco right now. I think weāre in the middle of a really great resurgence, but the job is not done. The same forces, similar sets of people who put out this asset seizure tax, they are also trying to 10x the gross receipts tax here in San Francisco.
Turner Novak:
And that is where you donāt necessarily pay taxes on revenue or income. Itās based on transaction volume of your service or something like that?
Garry Tan:
Thatās right. Yeah. And so this is exactly the tax that actually drove both Stripe and Square, a number of fintechs out of San Francisco. And we have something like a 30 or 40% vacancy rate on the right over here looking out the window. Itās like you will see a ton of empty space and itās been sitting empty for years. How are we actually going to fix that if nobody wants ... The second you get product market fit, are you going to stay in San Francisco? Are you going to pay literally a 1000x the tax that you would in Mountain View or Sunnyvale? Most smart people would probably find a way to say, āyou know what, I donāt have to be here.ā
And these things matter, right? San Francisco matters a lot. The reason why Iām fighting for it is that I want startups to continue to be in San Francisco and the Bay Area. If they stay, thatās the one thing that you can choose about your startup that can change your outcome so drastically. Itās two and a half times more likely if people choose to stay in the Bay Area that their company ends up being worth a billion dollars. New York is about 2x and everywhere else in the country is basically it. Thatās the baseline.
Turner Novak:
So just moving to New York or San Francisco increases your chances all else equal by 2 and 2.5?
Garry Tan:
Yeah. And I mean, I think there are infinity types of effects here. Youāre around people who ... You have access to capital. Thatās number one. Youāre around people who are very, very ambitious. And so your ambition rises when all your friendsā ambition is really, really high. I get access to way smarter people who have actually done ... I think the thing that scares me is that if the policies donāt support tech, itās not merely going to move. Itās actually going to hurt America. Itās going to hurt our ability to actually stay on the forefront of things and people scoff at that. But San Francisco is where we made the self-driving car.
This is where Uber came up. This is where Airbnb started. This is where GPT-1 was formulated by Alec Radford. This is ... The dawn of the AI revolution like Anthropic and OpenAI started in the Mission District. And the third crazy thing thatās happening that blew my mind, it was going to pass until we harnessed people on the internet in December, the sitting supervisor in San Francisco for the Mission District where they came up with self-driving cars and large language models wanted to ban AI labs, ban laboratories and research and development entirely in her district. She wouldāve made it so that you had to go to a special council, probably headed by her and her cronies, to even be allowed to open an R&D lab even for cancer research.
So to me, thatās a ban on R&D in the place where some of the best work in the world happened right here in our city. How do we get here?
Turner Novak:
Yeah. Well, even if you go back like semiconductor, think of Intel being basically birthed and grown in the Bay Area, the internet, all these different companies, different software companies, Instacart, how they probably get their groceries delivered. DoorDash. Iām sure theyāve stayed at Airbnb. Theyāve used Google products. The iPhone Apple was birthed and born here in the Bay Area. And if none of those exist, you donāt have a way to reach your constituents as a politician. You donāt have a way to go on vacation. You donāt have a way to access the internet. All that goes away.
Garry Tan:
Yeah. A lot of the internet was built right here. I mean, a lot of it right here in California. So I think people donāt understand itās the golden goose, but we should enjoy the eggs. We shouldnāt eat the goose.
Turner Novak:
Fair. So if I want to get involved in local politics, whether itās San Francisco or anywhere, what would you recommend doing?
Garry Tan:
Oh, follow me on X. And I guess what Iām trying to do is do a lot more research media. I mean this idea, this understanding of the actual proposition. I didnāt necessarily want to break it on my X. I sent it to a number of reporters and it just astonished me. People, they wouldnāt publish it. They didnāt think it was noteworthy.
Turner Novak:
I mean, based on what you just described, Iām like, holy shit, this is pretty big deal.
Garry Tan:
I mean, this is where I wonder, it doesnāt feel like the media environment is designed to support this idea that tech should exist at all. I think that itās so unpopular with certain sets of people in society that tech could be a positive force that anything that might even resemble an argument to not do an asset seizure, for instance, the most mainstream business publications are starting to refuse to publish that, which is just surprising.
How do we get here? But on the other hand, on the bright side, you do this all day and Iām trying to do this all day. We get to go direct. What weāre doing right now is not some sort ...
Turner Novak:
Iām not gatekeeping this. Whatever youāre saying, weāre putting it on the internet.
Garry Tan:
Weāre just putting it out there. This is what Iāve learned. And I think thatās the counterbalance, I guess mean. We used to have to go through a media and there was a mediation of how reality works, whereas now we can just go direct. Iām not going to get everything all the time. And I own up to that. I try to be as truthful as I possibly can be because I have to wake up and look at myself in the mirror and say, āAm I doing the right thing?ā And so maybe thatās our covenant as creators on the internet now.
Itās like, well, Iām not going to get it right all the time, but at least my guarantee to the people who follow me is like, Iām trying to do it, man. I have a particular worldview. Iām not without bias, but itās a bias thatās informed by ... I meet with founders all day. I meet with tech people all day, and Iām meeting legislators. Iām meeting policy people and this is what Iām seeing.
Turner Novak:
Plus, if you just suddenly start saying things that arenāt true or arenāt well researched and informed and people are just like, āGarry doesnāt know what heās talking about. Iām not going to pay attention to him anymore.ā You lose your credibility if youāre not continuing to fulfill that promise to your audience or to the people that trust you.
So on the other side, there could be audience capture. You might be totally biased of only communicating things that are favorable to tech in a way. So I guess thereās that argument too, which I mean then, itās good to have the counterbalance, I guess, of all these different people with different opinions and biases.
Garry Tan:
Yeah, Iām not the only person that people should follow, and I would never tell people use me as the only source of information. That would be bad. So I just think of that meme, hey.
Turner Novak:
The captain, Captain. Phillips
Garry Tan:
Weāre the media now.
Turner Novak:
Yeah, weāre the media now. You mentioned something interesting with these eras of YC. You mightāve actually said this before we started recording, but you said thereās been these different eras of YC, and so maybe the first era was, maybe it was Boston, maybe it was still the first era when Jessica and Paul moved to San Francisco.
What was the second act or second era of YC?
Garry Tan:
I mean, basically, it was Paul and Jessica and Trevor Blackwell and Robert Morris in Cambridge initially. And then they started coming to the Bay Area. And then actually Harj Taggar, whoās the managing partner with me at YC now, was one of the first outside partners. He was a YC alum. He actually co-founded a company with Patrick Collison that went through YC the first time called Octomatic. And I got to work with him, Paul Buchheit, who created Gmail, Jeff Ralston who created Yahoo! Mail. We were a set of people starting in 2011 who became partners. And so ...
Turner Novak:
This was like the second era?
Garry Tan:
Yup. So PG and Jessica were still directly involved, and it was way more than a full-time job, I would say. And they had very young children as well.
Turner Novak:
Was it like a weāre going to scale YC, weāre going to try to help more founders, weāre going to put more capital to work, weāre going to try to make a bigger impact on the world?
Garry Tan:
Oh yeah, definitely. I mean, I think that first era was fascinating to see because I saw here was this place where I was a software engineer. I was working at Palantir. I designed the logo at Palantir, actually. Employee number 10. And then we were trying to hire away some YC alums who had just gone through the program.
Turner Novak:
So thatās how you first came across YC?
Garry Tan:
Thatās how I found out about YC.
Turner Novak:
Oh, interesting.
Garry Tan:
So Palantir, we were only 10 people and we were just really into hiring the smartest engineers. And so itās funny because 2006, 2007, that was right when YC started. And so it was probably ... It wasnāt until maybe 2011 or so. That was when Dropbox had become a billion-dollar startup, but a lot of people said, āOh, one, youāre lucky.ā And then in 2010, 2011, that was when Airbnb became a billion-dollar company and then, āOne, youāre lucky. Two, youāre good.ā Suddenly, everyone in the Valley realized YC was actually really a place to pay attention to.
Turner Novak:
So Airbnb put YC on the map just in the sense of, holy shit, theyāve done this twice. There must be something there.
Garry Tan:
Yeah. And then thatās when YC went from giving people $12,000, $15,000 to $120,000, $175,000. And yeah, it was just interesting to see the difference. I knew back in 2007, 2008, this was where really smart engineers wanted to go start their company. It was like, āI wanted to be like Steve Huffman. I wanted to be Drew Houston. I wanted to be like James Lindenbaum,ā who created Heroku, which we used and really one of the best cloud software startups of that era.
And so, it was already in the minds of the creators and the founders themselves that hereās the super technical guild of all the smartest people making the things that we make. And then it took ...
Turner Novak:
Then other people started paying attention?
Garry Tan:
Yeah. It took three or four years for the investors to catch up, but they caught up by 2011.
Turner Novak:
Thatās classic though. The investors always catch up later when the proof pointās there.
Garry Tan:
Which makes sense.
Turner Novak:
Yeah. And then I think thereās a famous ... I actually mightāve tweeted this as a joke. But you were officially a photographer at YC at one point. That was your first role. You were just volunteering, taking pictures?
Garry Tan:
Yeah. Totally. I mean, we were applying to YC. I was actually trying to become an editorial hip-hop photographer at the time.
Turner Novak:
Really?
Garry Tan:
Yeah. Thatās right.
Turner Novak:
Okay. So, this like a portfolio work of build your brand.
Garry Tan:
A little bit. It was a quarter-life crisis of mine where I had rage-quit my job at Palantir.
Turner Novak:
Oh, really?
Garry Tan:
And I mean, I was just younger and I spent a lot of time on my YouTube channel talking about my psychological and business journey. And honestly, I managed to reprogram myself in a way and in ways that helped me get to where Iām at today. But I put a lot of credit to Cameron Yarbrough. He actually started a exec coaching startup called Torch, torch.io. And he was one of the people who taught me like, āOh, yeah. Everyone goes through whateverās going on for them. You have your 10,000 hours of human training from your childhood and your teachers and all the things that you saw. But on the one hand, some people treat that as destiny.ā Well, thatās what I got, and thatās it. And what I learned from Cameron was that these are things that you can actually control. You can change. Are you a West World fan?
Turner Novak:
I was, the HBO show?
Garry Tan:
Yeah, yeah. With the robots, and you could freeze all motor function and then you could go into a debug mode. Thatās what therapy and coaching can do, and thatās something that helped me a lot. I basically stopped rage-quitting after being able to spot like, āOh, well, this is a pattern.ā I rage-quit my job working for Joe Lonsdale at Palantir. I rage-quit actually my startup in 2010. And after that, I realized, āOh, no, no. I have this cycle where I will self-abandon a little bit too much.ā
For me, the central core thing that I had to learn about myself was that I had this mode where we might be co-founders, I disagree with you on this. But rather than just saying it, I might eat it. I just donāt even say it, because I want us to have a good relationship. I mistook this idea that you always hear the advice like, āOh, you better have a good co-founding relationship.ā And I mistook that for like, āOh, you should never fight,ā which is not true at all.
Great co-founders fight absolutely all the time and about everything, but mainly because they care so much and they do it in a way that preserves the relationship. There is never a conflict or a fight that is so much more important than the relationship itself. And that could be a co-founder in marriage thatās in all good close friendships. This is the repeating thing. And so, I learned that actually I shouldnāt, and then my pattern would be, I would keep it to myself. Iād say like, āOh, I can deal with this.ā
Some of this is that my mental model for human beings is thereās a horse and a rider, and I have a very strong rider. My prefrontal cortex and my language center can control the horse, I guess. Itās just like you would call that very strong will, I guess. And this is a very subtle, weird thing where I would use that to squelch all of these other things that I knew was correct, whether itās like, āOh, the homepage should say this.ā Or āActually, the startup, we started like this, but we should pivot to that.ā These are all things that you should be able to talk to your co-founder about. But it was on my end. I was a terrible co-founder who to my co-founder at Posterous back in 2008, that was my YC startup.
And so, yeah, these are things that you can control and you can change. And for me, I would run into a point where I had chosen to self-abandon enough that to use the horse and rider analogy, the horse would just say, āF this.ā I couldnāt eat. I couldnāt sleep. It became a somatic, actually. And this happened a few times, and it took quite a bit of therapy and coaching to realize, āOh, yeah. This is a cycle. This is something that happens and I can actually do something about it. I can be aware of it.ā
So, today, in my work with my therapist and my coach, often Iām trying to think through like, āOkay. Freeze all motor function, whatās going on here?ā And then if Iām very lucky or very, very astute, and usually with the help of others, Iām not able to do this myself, I can place an if statement in there. And itās like, āWhen my face gets hot, Iām going to stop.ā And I might have a default reaction like shutting it down or like I said self-abandoning. The part that would happen would be I bottled it up so much, then Iād blow up.
I went into a one-on-one with one of my first designers Iād ever hired, and I just blew up at him and he was like, āGarry, Iāve never been more disrespected in my entire life. How did this happen? Whereās this coming from?ā
Turner Novak:
Itās probably just years of you just not voicing the thing, and then you hit a tipping point and hereās 18 things, and Iām laying on you all at once.
Garry Tan:
Exactly. And thatās definitely not healthy. And so, the cool thing though is I think it happens all the time. I think that itās actually one of the more fundamental things that I really want founders to understand, because I just gave my map, and it took me many, many years to map that out, to freeze all motor functions regularly every single week and figure out what was going on. Everyone else has clearly other things. My thing is not universal. Mineās specific to me. Everyone has their own set of things. But like it or not, do you know how people talk about startups and startups themselves take on the personality of their founders?
Turner Novak:
Yeah. Iāve heard that.
Garry Tan:
And so, this is a direct outgrowth of that. If people donāt handle these things and get a chance to figure this stuff out, it just shows up in your org in ways that are ... Itāll just happen someplace. And youāll be like, āWell, where did that come from?ā And then if you really pull on the thread, itās like, āOh,ā because thatās me. Thatās how I ... Who you hire, who you fire, how you deal with almost any conflict, that is actually how the organization deals with conflict.
So, on the one hand, most people look at this internal work. Itās like, āOh, itās like some Westcoast, do some ayahuasca and talk about our feelings type bullshit. That doesnāt matter.ā And Iām like, āNo, no, no. Maybe psychedelics are not for everyone. Be very careful about that. Thatās my take.ā And on the flip side like, āNo, no, no. How you do anything is how you do everything.ā And so, I certainly recognize this repeating pattern in myself. And I mean, I think we see it play out across thousands of companies, and we canāt be there all the time. Being a YC partner for a company, we try to be there.
But we also see these repeating things and we have to take control of ourselves. If you can be a better person, actually, youāll show up better, youāll hire better people, youāll manage better. Everyone will be more functioning and functional, not afraid, more truthful, more focused on the outcome, and then that will produce a better product and actually winning in the marketplace. So, itās like itās not this in the sky, letās talk about our feelings thing. Itās actually like itās fundamental to whether or not a startup works or not.
Turner Novak:
Well, so speaking of that question, I feel like YC has published data before on these are the most common reasons that a startup fails. I think number one is āyou run out of money.ā And then maybe number two is co-founder disagreement or conflict. Is that still true?
Garry Tan:
I mean, maybe this is a funny aspect of YC. YC companies tend not to die, because they ran out of money, because they have some money. Usually, they shut down because they get tired, they get demoralized. I mean, itās real. Honestly, often co-founder issues. And so, a great co-founder conflict, you might not agree on everything, but youāre able to get to some outcome. And then the outcome and decision leads to success. And then good things beget good things.
Turner Novak:
So, even if you have disagreements, you might disagree on everything at first, or youāre able to come to a common understanding or decision and just do it and just move past it and maybe not move past it in a bad sense of bearing, but talking about it and figuring out like, āThis is what the decision means to us or why we should do it.ā
Garry Tan:
Absolutely. And then the nuance is honestly, I like the analogy of an idea maze. Youāre at the beginning, youāre actually at a maze and youāre trying to get to product market fit to, you want to create something that people want. Make something people want is the T-shirt that we give everyone.
Turner Novak:
Oh, you actually get a T-shirt.
Garry Tan:
You literally get the ... If you had to sum YC up in one sentence, one mantra, itās āMake something people want.ā You can just think on that. We should just make people say, āMake something people want,ā as a Hail Mary or something. Itās like, āHere we give you absolution.ā Just say, āMake something people want 1,000 times and meditate on it and your startup will succeed.ā
Turner Novak:
But then you start thinking of make something people want. What do people actually want? I feel like some people, you just forget. When I started this podcast, it was like, āWhat would I want to hear in a podcast?ā Itās not just whatās the checklist of what you should do when you make a podcast? It was more of like, āWhat would it be fun to talk to Garry? But what would I actually want to hear if I had him come on?ā
Garry Tan:
Awesome.
Turner Novak:
So, I think about that with everything. Itās like make something people want, I guess.
Garry Tan:
Yeah. You start with the verb like āmake.ā You got to make something. And then whatās funny is make something is too short, because itās missing the interaction with other people.
Turner Novak:
Oh, yeah. So, then I guess weāre continuing the phases of YC. So, you joined at some point, it was after you went to journey.
Garry Tan:
Yeah. 2011.
Turner Novak:
So, you had raged quit, and then you were suddenly like ...
Garry Tan:
Yeah. I raged quit my startup. And then Harj said, āWell, youāre a great designer, and this is the era of consumer startup still.ā So, this is it. This is the moment when people said, āOh, well you need a designer with your co-founding team.ā And so, I actually came on having no intention of ever becoming an investor, ever becoming a GP or starting a VC fund.
Turner Novak:
Oh, really?
Garry Tan:
I was just like, āIām really burnt-out and I need to recuperate.ā And Harj and PG and Jessica said, āWhy donāt you just come and help a bunch of YC companies design their homepages?ā And then I just did so many office hours that my pet theory is maybe they saw how much work I was doing, and they were promoting other people who were great into investing partner, and they said, āWell, youāre doing so much work. We need to promote you too.ā So, Iām like, āI snuck right in there.ā
Turner Novak:
So, unclear if youāre actually good, but you were working hard.
Garry Tan:
This was my imposter syndrome at work. But it worked out. I mean, I learned a lot. What was really cool about it, I really loved that moment, because I was the same age and I was a practitioner right at that moment. Itās very interesting to be 12 years, 13, 14 years into the other side of that. I think this is probably the life of many an investor. You start out as a true peer to a lot of the people youāre backing. And itās very funny.
But on the other hand, I feel like I can be right back there. This whole weekend, I was just sitting there in Claude Code making a new website and doing new interaction design. Iām like, āOh, my God. Iām back.ā
Turner Novak:
Back in the game.
Garry Tan:
Thatās right.
Turner Novak:
The investor truly rolls up their sleeves and theyāre in the trenches.
Garry Tan:
Thatās right.
Turner Novak:
Theyāre trying out all the new tools.
Garry Tan:
Yeah. Yeah.
Turner Novak:
So, then there was a period you started initialized, then you were still at YC. How did that go?
Garry Tan:
Yeah. I mean this was back when people could do be a YC partner and do personal angel investing. And then Harj and Alexis and I were the young partners who had not made any money yet. Palantir was still private. And Posterous had sold to or was about to sell to Twitter, but Twitter was also private. And so, Jessica said, āHey, you could probably be an investor byā ... And she introduced us to some LPs, basically. And so, we raised Initialized Fund 1, which is a $7 million fund, and we ended up returning more than 55X DPI on that one.
Turner Novak:
Not too bad.
Garry Tan:
Yeah. So, I mean, it was mostly Coinbase and Instacart. But I think thatās probably the thing in an investorās worldview that you read about it, you hear about it, and then until you see it, you donāt believe it. We funded 100 companies. And if you added up the DPI on all 98 companies, it was like 1X. And then Instacart was 1X, but we couldnāt take any pro rata, because Sequoia did it. And then Coinbase, we took the pro rata in multiple rounds, ended up turning the fund 53X.
Turner Novak:
Oh, wow.
Garry Tan:
So, when you see the power law, it is astonishing and unbelievable, honestly.
Turner Novak:
Yeah. And then, Initialized seemed like it was going really well. So, you stepped back from YC on a daily basis and were running Initialized full-time?
Garry Tan:
Oh, yeah. Yeah, yeah. I mean, basically, this was the Sam Altman era. And at that moment, I had helped raise the continuity fund, brought in the first LPs for it, and then I said, āWell, YC is in pretty good shape and itās got institutional backing. Itās going to keep going.ā And then for me, it was like itās more fun to be a pirate than to join the Navy. But here I am, Iām an admiral again.
Turner Novak:
So, then how did that come about where you had started your own fund? It seemed like it was going really well. I think you had probably raised your third or fourth fund, maybe I should have looked this up again before. But it seemed like, āOh, youāre set with Initialized.ā Why would you go back to YC? How did that happen?
Garry Tan:
Yeah. I mean I guess this is a special thing. YC is not just a boot camp or itās not even just a school. Itās a way of life. Yeah. I guess more seriously, I mean, all of our wins, like 80%, 90% of the wins were either YC or YC alumni.
Turner Novak:
From Initialized?
Garry Tan:
For Initialized. I mean, super big ones that weāre still super psyched about. I mean, ended up being the first investor in Rippling, first seed investors at Demo Day for Flock Safety. I mean, we have a ton of incredible companies that are all today worth a lot of money. And then I looked down and realized, āWell, this is all a part of this larger ecosystem.ā And mid to end of 2022, Sam had gone off to work on OpenAI. Jeff Ralston, who Iād worked with for many years, took over as a caretaker president. And then he said a bunch of people from that era, especially PG and Jessica said, āWe really want you to come back.ā And Brian Chesky was a big part of that, of Airbnb. He was on the board and still is on the board at YC. And he said, āGarry, itās time for you to come back.ā
Turner Novak:
Were you expecting that?
Garry Tan:
Not at all. I mean, I never thought that would happen. On the other hand, itās like a dream job to me. I mean, YC gave me my start as a founder, gave me my start as an investor, was probably 50% to 80% of my friends are all from this world. I mean, basically, going from being on the branch to tending the root of the tree of prosperity, some of it was like, āIf it wasnāt going to go the right way, I might as well retire.ā Itās one of those only people of a certain age will understand my reference of ... Itās like the Hair Club for Men in many, many different ways. Iām not just the president. Iām also a client.
So, I both went through YC and I was also one of the investors at Demo Day. And I managed to turn in multiple 10X plus legend VC funds. And so, Iām like, āNo, no, no. I understand this ecosystem really, really well.ā
Turner Novak:
Yeah. Because almost been every constituent that you serve in the ecosystem youāve been on the other end of it.
Garry Tan:
Yeah. And I mean, Iām sure that most of my VC friends would agree with this at this point. But that was a concerted effort the last two or three years to basically realize that YC is a managed marketplace and thatās a good thing. Basically, itās a managed marketplace for all the smartest people starting the next companies. Weāre about 20% of all the names, all the companies worth $5 billion and up, started from 2012 and later.
Turner Novak:
I did participate in YC.
Garry Tan:
Yeah. Yeah. And frankly, itās a network effects biz. So, it stayed about 20% for going on 15, 18 years, I would say. And why canāt that be 30? Why canāt that be 40? So, you can mark 2026 on the calendar. And if Iām successful in 5 or 10 years, Iād like that to be 30, I like that to be 50. Why not? This is and should be a network effects business where itās not about the money, itās about having a community of all these people who you can trust and theyāre there to support you. Thatās actively ... I think of it as Disney Imagineering, Disneyland is this incredible place.
Turner Novak:
Yeah.
Garry Tan:
And you did. And if you go down into the bowels of Disneyland, youāll realize, āOh, thereās an incredible amount of care and thought about every little detail. Every part of the experience is designed.ā And I really credit Brian Chesky for that. I mean, heās the original designer, founder. Having him on a board is absolutely insane. Because heās like mile-a-minute ideas about what we need to do to create. Itās no surprise at all that he created Airbnb, because heās that good, thinking about human experiences.
Turner Novak:
So, any examples of that? What have you learned from Brian Chesky in terms of designing human experiences or ways to make YC a better product?
Garry Tan:
I mean, a lot of it is just focus. Iām sure people ... I mean, if people havenāt seen it, they should definitely go and watch. I mean, you need to read Paul Grahamās essay about Founder Mode. And then Brianās done a bunch of things. Actually, I liked the one he did with Jessica Livingston on The Social Radars podcast about Founder Mode. And basically, he had to remake that company from scratch during COVID. And he basically realized that you have to ... Leadership is his presence, not absence.
The classic advice that people give to startup founders who reach product market fit is hire the smartest people you possibly can find and give them the keys to the kingdom. And his advice flies in the face of it and heās like, āOh, I did that and it created a company that I did not recognize.ā My directs would ... I mean, thereās just so many different things that he realized. And then I think as founders, especially post-product market fit, you just donāt listen to basically your own beating heart around what you need to do for the company.
And so, what he said, Iām not the founder of YC, but I feel like he gave me between him and Paul Graham and Jessica Livingston, this incredible sense that like, āNo, no, no. Iām not the founder, but we are going to run it like Founder Mode.ā I need to take a really, really radical ownership about what is that experience like, āWhatās Demo Day like? What is the application process like?ā We brought back in-person interviews. We got an office right here in San Francisco. We told all the founders, āMove to the Dogpatch. Move to the center of where all the stuff is being built and be surrounded by all the other people who do that.ā
When I came back, it was a remote program. Basically, the batch is not remote at all. You cannot do YC remotely. But it wasnāt really working as a remote program.
Turner Novak:
Really? So, when you talk about that 20% number, did it start to dip or did you feel like you were losing that 20% and it wasnāt going up anymore?
Garry Tan:
I mean, thatās what it felt like, honestly. And I mean some of it is we did a lot of really amazing things like YouTube started becoming a really big factor for YC. But YouTube is not enough. This is part of the reason why weāre going to open an office in Cambridge. We need to be on the ground with the worldās best founders. And frankly, some of the best ones are right there at MIT and Harvard.
Turner Novak:
So, youāre still trying to scale it. Because it sounds like before the answer to scaling the YC experience was online application and interviews, and maybe itās a little bit more decentralized. I think I saw a chart of international YC companies, it was going up quite a bit where you guys funded more teams around the world. But now, it sounds like itās still trying to hold that in-person. What makes it special? Weāre creating this network. But maybe trying to scale that, maybe try to recreate what exists currently in the Dogpatch and Cambridge.
Garry Tan:
Oh, yeah. I mean, the thing that keeps me up at night, some of it is we meant 99% of people who apply to YC end up getting somewhat of a mortal ego wound by being rejected.
Turner Novak:
I was going to ask you about that. I see a lot of, I donāt know, itās like people who are upset that they didnāt get in.
Garry Tan:
I mean, it keeps me up at night because itās not a good thing, and this is something we talk about inside YC. Itās something like a third of the batch, maybe even ... depending on the batch, sometimes half of the batch will have been rejected prior to getting in. And that numberās been going up. People sometimes have to ... Itās quite common for people to apply two to four times before they get in. So, I think weāre always worried about over rotating. Basically, being too small can be bad, being too big can be bad. Weāre always trying to figure out what that number is.
But I think weāre pretty clearly right now there are way more really, really capable founders who we are rejecting and that keeps me up at night. We should not have to ask people to be rejected like time after time. I will say though, if you can survive the mortal ego wound of being rejected once, twice, or three times, youāre actually made to be a founder. If you cannot be rejected even once and you mortally hate YC after being rejected once, maybe try a different profession. Because being a founder really does require an absolutely insane amount of resilience.
And one rejection, the get ready buddy is going to be 1,000 more in every possible way. And a really great founder just says, āYou know what? They got this wrong, but Iām going to prove myself right.ā And to be frank, that is what we love. I particularly love funding founders who, āWe mightāve gotten it wrong.ā Thereās actually an example of this in fall batch, actually. Thereās a company that theyāre very active on Twitter right now and Diana picked them up for the winter batch. But both Nico and I had interviewed and we should have accepted for fall.
If youāre watching, Iām so sorry, I made a mistake. And Iāve said that on Twitter already. Yeah. I mean, this stuff keeps me up at night. I donāt like to hand out, honestly, how many is it? We get about 80,000 applications, so it really is something like 78,500 mortal wounds.
Turner Novak:
This is per year?
Garry Tan:
Yeah.
Turner Novak:
You accept probably like 200, 250, 300 per batch?
Garry Tan:
Oh, no. Yeah. I mean, I think weāre doing about between 700 and 800 companies a year.
Turner Novak:
Itās about 1%?
Garry Tan:
Yeah. I mean itās a lot of companies. We reject a lot of companies. And on the one hand like, āHey, please apply again.ā Thereās that meme I try to post thatās like, āHey, getting rejected from YC is the first step to being accepted to YC.ā
Turner Novak:
Yeah. Nice.
Garry Tan:
Yeah. Itās the Adventure Time meme.
Turner Novak:
So, what would you recommend? If somebodyās listening to this and they have or havenāt applied to YC yet, what are some of the things that maybe theyāre obvious, maybe theyāre not obvious, but what are you looking for, someone applies to YC?
Garry Tan:
I mean, I love builder. So, just being able to see ... I mean, honestly, what Claude Code can do today, whatās funny is boilerplate, what just comes out, kind of sucks. And so, one of the things Iām even realizing from the last weekend of going deep on it is that you can create, there is pretty much no excuse why people canāt make beautiful, super well-made stuff. Iām worried, maybe this is the Boomer 2010 startup founder talking in me right now. But I think that there is a craft to creating online and startup experiences for people. I mean, if youāre building software anyway. Thereās just something there that I think itās easier than ever to create something that is all the way sanded down.
The story I always think about is Steve Jobs would famously talk about how a cabinet-maker can recognize other cabinet-makers as being really great at their job. Not by looking at the front. Everyone can look at the front. Most everyone thinks about the fit and finish of whatās in the front. A great cabinet-maker will finish the back, and then thatās how ... Game really does recognize game.
And when I look at the partners that we have, we have people who have sold their companies for hundreds of millions of dollars that theyāre people who have ... Theyāre really great practitioners of getting to product market fit themselves. Like Tom Blomfield created Monzo, which is worth billions of dollars. Itās one of your absolutely top fintech companies in the world. And so, we have 15 people, all of whom I can say without a doubt, especially after doing funding, even two or three years of companies, two or three years of YC time as a partner is equivalent to 20 years in venture, I would say. Because think about ...
Turner Novak:
Is it the reps? Just how many?
Garry Tan:
The reps are unbelievable. If you are reading personally 1,000 to 2,000 applications every single year, youāre working very directly, writing ... I mean, frankly, even as a seed investor, thatās pretty wild. Writing 50 seed checks in a given year, sometimes 75.
Turner Novak:
Yeah. I do about six to eight in a year, not quite as much.
Garry Tan:
So, if you get those reps, the organization learns. And actually, thatās the reason why it works. I donāt think that ... I can try to pull out the billion-dollar idea for founders, and thatās hit or miss. That just requires inspiration. I think that people like Paul Graham or Vinod Khosla, for instance, par excellence, incredible. I would say that all of the YC partners are top 2% to 5% in the world being able to do that. But on top of that, because we get so many reps and weāve seen so many things, itās actually about stopping you from blowing yourself up on the 10,000 landmines.
Itās like, āOh, thatās going to lead to co-founder conflict. Oh, be careful about taking money from that investor. Oh, you got single channel concentration.ā Thereās literally 10,000 things. The good thing is because weāve seen everything, weāll never freak out. I mean, thatās actually really good investors are like that. Theyāre soft advisors who never freak out. Theyāre like, āOh, this is happening. Interesting.ā
Turner Novak:
Your company is probably going to die in a month if you donāt make this change. But thereās three people that have actually done this before and hereās what they did.
Garry Tan:
Yeah. And then, look, I can tell you this, but also talk to the person where it happened. And thatās where it takes a village to create these things. And so, yeah, if I look at why is it that YC can do this? Well, whoever is in that batch is the top 1% of whatās going on. And then if you put them in a room, they just, by default, create this community of default trusts. And then the partners themselves, you get 1 out of the 15 thatās like, āHey, weāre here for you for the life of the company and beyond. Weāre actually there to fund the founders.ā
Turner Novak:
Yeah. Talk a little bit more with that. Is there a common reason maybe between that top 5% and 1%? You guys send an email thatās like, āHey, youāre in the top. I think itās either 5% or 10%.ā Is there a common reason that maybe people donāt quite get to that accepted, but theyāre ranked pretty highly? Is there something ... Maybe the first two times they didnāt get in, but then the third time they get in and theyāre really successful. Are there things that you feel like theyāre like fringe almost make it into YC types of things?
Garry Tan:
Oh, man. I mean, the hard part is, I couldnāt tell you. Every single company is such a different unique snowflake.
Turner Novak:
Thereās a lot of subjectivity probably, too, where you just have a gut feeling on this one, but not this one?
Garry Tan:
Yeah. I mean, we often try to re-interview, because actually more and more common that if one partner says like, āWell, thereās something here. I canāt do it.ā I mean, sometimes people just get full up. We definitely, we donāt want partners ...
Turner Novak:
You each pick your ...
Garry Tan:
Yeah. Each partner picks their own cohort. And basically, sometimes our eyes are bigger than our mouths and bellies, I guess. I donāt know. So, we try to pass it to other people who might want it. I think a lot of it is just about ... Yeah. It is that snap judgment in 10 minutes and all we can really do is give you more shots to have it click for someone. I mean, itās a considered thing. YC done wrong, it feels like random winning of the lottery. And YC done right is like youāre meeting someone and itās actually pretty high-pressure for the partners, because like it or not, weāre going to get text messages and emails and weāre signing up for the long haul, work. So, itās considered.
Turner Novak:
You talked a little bit about you get into YC, thereās this batch. How does it go just for someone whoās not familiar from the outside and then what would you recommend to get the most out of YC?
Garry Tan:
Yeah.
Turner Novak:
If I just got in, and Iām listening to this, Iām super excited. Maybe you tell me this when I get in. But what would you say to someone to get the most out of YC?
Garry Tan:
Yeah. I mean, we just extended it to a 13-week program. I mean, this is what I mean by itās Disneyland and we have to imagineer this stuff. We pay attention to, we talk to people and itās like, āHow did it go? What can we do better?ā We went to four batches and realized, āOh, yeah. Actually, the batch got a little bit shorter and people need a little bit more time.ā
Turner Novak:
So, you added, you went from two to four and you shortened it when you went to four?
Garry Tan:
Yeah, yeah. We went down to maybe 11-ish weeks, 12-ish weeks. We realized, āOh, yeah. No. We need to add 2 more weeks and get this to 13.ā And so, the first 9 to 10-ish weeks, I would say, weāre just trying to find product market fit. Letās get in market. Letās nail our messaging. Letās figure out who weāre building it for and letās build it. Letās get a set of real customers. Sometimes enterprise, sometimes itās ... And DevTools is very funny, because DevTools is actually consumer startup, but your total addressable market is 20 million people. So, now, with Codegen, maybe thatās 200.
And so, yeah, AI infrastructureās growing really, really fast and thatās just ongoing, always changing. I think YC is totally insane for infra and cloud startups in that respect, because you basically looked to your left, looked to your right and looked like at 800 companies that were funded the prior year, and all of those people will reply to your email and weāll at least give you fair shot. Itās not a guarantee theyāre going to sign on. But man, the first three to five customers matters a lot, and that sets the tone for the whole company.
And then, we actually try to keep the amount of time spent on fundraising to maybe two weeks, basically. The funny thing about fundraising is I think the mistake that people make as founders is considering that thatās the finish line. And it was like, āNo, no, no. Thatās not the finish line. Thatās the starting gun, guys. Have a party and then actually get back to work the next day.ā Because safes are, theyāre not actually debt, but you should treat it like that. Someone gave you a million dollars, they gave you $5 million, but guess what? You signed up to say that, āHey, weāre going to return a big multiple of that based on the things that weāre going to do from here.ā That million dollar is going to be worth 100 million. Itās going to be worth a billion.
Turner Novak:
And you mentioned earlier, you alluded to things or investors you should take money from or shouldnāt take money from. If Iām doing YC, Iām approaching Demo Day. Maybe explain a little bit how Demo Day usually goes for someone who has only heard of this but never been there. And then how do you usually recommend somebody picks what investors they should or shouldnāt work with?
Garry Tan:
Yeah. Totally. I mean, I guess the reality is people start basically towards the end. We want people mainly working on the startups, because thatās what actually matters. And then there is a starting gun inside YC or going to talk to investors. So, if I were an investor, I want to talk to them that probably Saturday. We do the sendoff, youāre done with fundraising prep, youāre ready to talk to investors. And then usually that Saturday. Itās like, āAll right. Off to the races. Let me start meeting people.ā
And so, itās pretty important, the number one thing that a startup founder can do going into YC is make really, really valuable things and get real customers that pay and retain. And then fundraising is easy because everyone can see it.
Turner Novak:
Yeah. Because itās like you are trying to give someone an attractive asset to invest in and thatās what theyāre doing. They want to make money. They want to give you money. And then in 10 years you give them back 100 or 1,000 times more.
Garry Tan:
Totally.
Turner Novak:
Thatās because you made a valuable thing thatās worth a lot of money.
Garry Tan:
Yeah. I mean the cool thing is about 20% of YC now is hard tech. Defense tech is up more than 2X in the last batch. Defense tech is interesting because itās both. You want commercial validation. Icarus was one of the top companies from the last batch and they literally got ... they booked more than a million and a half dollars from the Department of War during the batch.
Turner Novak:
Oh, interesting.
Garry Tan:
So, thatās why they were one of the top companies. They, during the batch, was able to go from here to here. And I think that thatās the number one thing. Objects in motion, stay in motion.
Turner Novak:
Stay in motion. Yeah.
Garry Tan:
Objects at rest, stay at rest. And so, YC is the moment that you need to learn how to run fast. And this is something that we hear from all the billion-dollar companies even they look back and say, āOh, if they feel like their 1,000-person company is going slow, they try to tap that energy that they got when the first week, the first eight weeks of YC, when they were running fast.ā Airbnb famously talks about they, on Post-it notes in the bathroom, put a little graph of like, āEvery single week, we need to grow our revenue on the site, managed marketplace 10% every week.ā And they nailed it. And thatās why they raised from Sequoia prior to Demo Day.
Turner Novak:
But didnāt, no one want to invest in Airbnb, originally, right?
Garry Tan:
I mean, they got rejected from all the legends. So, yeah, the funniest thing was I think Brian would always obscure who the people were. But now itās such a legend story. I think I saw Mike Maples post like, āOh, yeah, that was me.ā And Mike has nothing to prove. Heās one of the biggest legends in all of venture. So, itās just a part of lore now. I mean, and thatās what gives me heartburn sometimes. Like, āMan, if we make a mistake and we say no to a founder who ends up creating Airbnb, Iām sure weāre going to do it. The reality is Iām sure weāre going to do it and Iām going to feel bad about it and Iām going to take that shame and I take that feeling and Iām going to turn it back into productive energy to make sure that YC fixes that.ā
How did that happen? Weāre going to fix that. And then the next founder, weāre going to fund them. Weāre going to get it right. And I mean, I guess thatās what Brian Chesky taught me. I need to exercise Founder Mode on this tree of prosperity. And then if we do that, then actually, āWell, yeah, actually YC will fund 30% to 50% of all the companies that matter.ā
Turner Novak:
Yeah.
Garry Tan:
And then if weāre doing it, then weāre actually even better equipped to help the next generation. Because everything that we do, we have to take our own advice. Thatās actually a common refrain we use all the time now. Itās like, āWell, if weāre a YC startup and we could do option A or option B. Option A, spend a whole bunch of money and hire a bunch of people. Or option B, letās do it quick and dirty and see what happens and take feedback and then speed of iteration. Letās fix it next time.ā We got to do the startup way 10 times out of 10. I feel like thatās my real job is like, āNo, no, no. We have to do it this way.ā
YC at 20 years by default, almost always post-product market fit, you fall into manager mode and itās like, āNope. Weāre back to founder mode.ā
Turner Novak:
So, I guess if weāre continuing ... This has just been an hour and a half conversation with the acts of YC. Maybe weāre at the end or the beginning of act three or four or five, I donāt know where weāre at, but what does the next act look like for YC?
Garry Tan:
I think done right, more and more ... We actually just straight up want more companies to succeed. And so, what does that look like? Iām going to experiment with re-batching. I think that ...
Turner Novak:
Re-batch, what does that mean?
Garry Tan:
Post series A, after you get your series A, like we should give you another batch. Iām going to start with the companies that I funded and then if that works well ... I mean, weāre just going to experiment. Weāre going to try things. And then if they work, weāre going to double down and build software around it.
Turner Novak:
It reminds me of Standard Capital. They do the YC batch style for this first series A, which no one has really done. Yeah.
Garry Tan:
Thatās right. Yeah. Dalton and PB are insane. And Paul Buchheit, just super extreme legend investor.
Turner Novak:
Yeah. Yeah. I have a portfolio company that I think it will have been announced by the time this podcast comes out. But they work with them and it was surprising. I was like, āOh, thatās cool. I didnāt know you guys would be up for that.ā But it actually is a pretty interesting value proposition to the founders.
Garry Tan:
Amazing.
Turner Novak:
So, yeah, Iām excited about that. Actually, one thing also, I heard that the YC has this internal software called Bookface. I heard that you actually made it.
Garry Tan:
Oh, yeah, yeah, yeah.
Turner Novak:
Is that true? Okay.
Garry Tan:
Thatās right.
Turner Novak:
Whatās the story with that?
Garry Tan:
Well, basically YC started becoming a thing and then people started pretending to be YC companies.
Turner Novak:
What?
Garry Tan:
And then there was no way to verify it. We had a Google group or something. I was like, āOh, okay. Let me just ... Vibe coding wasnāt a thing back then. I just actually real coded it in Rails.ā And initially, it was actually just a Facebook and this was a little bit of an inside joke from a skit from the office.
Turner Novak:
In the office with the Jim, the Bookfacing. Okay.
Garry Tan:
Yeah. Jim wearing Bookface, yeah, yeah. See of a certain generation. See, I like you. We still have ... The 22-year-olds do not watch any of the stuff that we came up with. So, I recently had a showing of the Matrix for all the 22-year-old founders.
Turner Novak:
Okay. Iām barely on. I need to go back and re-watch it, re-watch all of them. I feel like it comes up and Iām like, āI watched them once when I was a kid, but I donāt remember.ā
Garry Tan:
Totally. Itās surprisingly relevant to startup founders.
Turner Novak:
The Matrix is.
Garry Tan:
Oh, definitely.
Turner Novak:
So, a startup founder is like Neo?
Garry Tan:
Yeah. For sure. Yeah. Why not?
Turner Novak:
Interesting.
Garry Tan:
Yeah. Well, no. I mean, heās stuck in this office and at the whim of the man you have ...
Turner Novak:
Thatās fair. Yeah.
Garry Tan:
Youāre in this cubicle and you have no control and no power. And then heās a hacker though, and he goes to these cool raves and then he meets these other secret hacker types that understand the nature of the universe. I mean, thatās how I felt when I was the software engineer and then went to become a founder. I was like, āThereās nothing that felt right about me working as a Level 59 PM at Microsoft.ā And then I started to glimpse the Matrix when I worked at Palantir and working for Peter Thiel and Joe Lonsdale and all my ... I was fraternity brothers with Stephen Cohen and Joe Lonsdale and I could see it, but I was still in the Matrix.
Turner Novak:
So, on the cusp of getting out?
Garry Tan:
Yeah. And then finally, YC was what helped me take the red pill and make the leap. And it was like, āOh, no. I can be not jacked into the Matrix. I can be in the real, real.ā
Turner Novak:
Yeah.
Garry Tan:
And thatās what it feels like, though. Being a startup founder allows you to be directly present and in touch with the market, the users, the customers. Customers can go anywhere. Customers donāt have to use your stuff. And so, being in the real allows you to create products and things that actually get good outcomes. So, I donāt know. To take it full circle, thatās why this billionaire tax, they call it the billionaire tax, I call it the asset seizure tax. Gavin Newsom himself is just running around saying, āGuys, we canāt do this. Why are we doing this?ā There are 49 other states, people can just go. And thatās what theyāre doing. Larry and Sergey have left more than a trillion dollars of people have left the state.
Turner Novak:
Once they leave, they go through all that hassle of settling across the country and they be like, āYou know what? Floridaās actually pretty nice or Michigan or Montana or any of the other 49 states, do I need to go back?ā
Garry Tan:
Yeah. So, I mean, if people are watching and theyāre startup founders ... There would be startup founders and theyāre in their ... I too was working as a cog in the Microsoft machine when I was 22 years old. And there was something about like, āGo home and watch the Matrix.ā
Turner Novak:
Yeah.
Garry Tan:
And then realize starting a company is taking the red pill and you might just turn out to be Neo.
Turner Novak:
One question I want to ask you, do you have a personal AI stack? How you using AI? You talked about using Claude a little bit, just how do you use it today? What are you taking the most advantage of? If there was Garryās top three hacks of using AI, what are you using today?
Garry Tan:
The number one thing, I started learned this from Jared Friedman and Pete Koomen, my partners at YC, but I call it meta prompting. So, you can take almost anything that you might do all the time and you just drop it into a context window and then say, āHereās a bunch of inputs and outputs. Maybe you have a bunch of notes.ā And then for me, for instance, I have a YouTube outlines where itās like, āOh, Turner had a really funny tweet.ā I would put that in the outline and then I might find two or three other things that flowed together. And then I would write YouTube script for me to go on the teleprompter and recite so I could get it recorded.
But you can take a bunch of these inputs and outputs and then just drop it into a context window and tell it, āWrite me a prompt that can act as an agent that does take this input and output over here.ā And you can even introspect, youād be like, āWhat are things you notice about things that I did to convert this from the input to the output?ā You do this for almost any knowledge work, and then you can just start using it.
And initially, itās going to suck, because itās just not that smart yet. But whatās funny is now, I also use it to Iterate my writing. And so, for something like a YouTube script, you really want it to sound like your voice. And so, you can be very direct about, āWell, I would never say that.ā Or āDonāt say it like this.ā Or, āOh, you used the long word when ... Use the short word, man.ā You just speak to it conversationally. And then, simultaneously, you start with the prompt, then you have an output. But then you can use that new output, but right after you get that new output and youāre happy with what happened, youād be like, āGive me the next version of the prompt.ā
Turner Novak:
Evolve my ...
Garry Tan:
Yeah. Based on what we talked about in this conversation. Give me a better prompt that incorporates all the things we talked about. And you can do this with literally everything. And in theory, honestly, thereās so much about what people do day-to-day. You could use it for tweets, you use it for editing podcasts, you use it for pretty much everything. I just have a folder now of prompts that I use all the time and Iām trying to iterate my YouTube prompt one is I think on V-27 or something.
Turner Novak:
Oh, interesting. Okay. I have not used it a whole lot yet for content creation. Iāve done a lot of just throw into ChatGPT and it has memory. It gets context, but I havenāt done prompt evolution. I guess I need to master that.
Garry Tan:
Yeah. Mastering is great. I mean, the other thing thatās great is sometimes when itās not quite working, Iāll use GPT 5.2, Pro, Iāll use whatever the max thinking, Iāll even use Grok and then Iāll take ... For whatever reason, Claude seems to be the best at evaluating the outputs. Iāll take all of the outputs from all of the different max models and then Iāll put it into Claude and Iāll ask, well, given ... hereās my prompt, hereās the output from four LLMs, including yourself, rate each response and tell me what the pros and cons of each approach are. And then itāll spit it out. And then you can agree or disagree with all.
I usually like to say, āGive it to me in numbered form.ā So, I can say, āI agree with one, I disagree with two. Three is this, but the nuance is blah.ā And then after that, you can actually just say, āOkay. Well, given all of this, synthesize.ā
Turner Novak:
Yeah. Thatās interesting. I like it a lot for rapid ideas. Itās just like you get 100 ideas and one of them is good. You can give it feedback on, āI really liked this one.ā Or because it just comes up with stuff that maybe you would or wouldnāt have came up with.
Garry Tan:
Totally.
Turner Novak:
It just speeds up the process a little bit.
Garry Tan:
Yeah. And then I use Perplexity for anything that has to search the web. So, I think that the Claude and Gemini and these things are pretty good. But I still think my experience, is Perplexity still draws on and synthesizes way more sources. So, I really like that max mode.
Turner Novak:
I like that about Perplexity where ... Because one of my holdups or one of the things I really like about Google is it shows you all these sources. You can see where things come from and if you just like the default ChatGPT response, you donāt know where the stuff is coming from and maybe itās the correctness or bias or whatever. But on Perplexity, it shows you all the sources and you can jump in. So, I do like that. So, I still use Perplexity a little bit, too. I like Perplexity for ... I feel like it does the best when I say, āIām having Garry Tan on my podcast, what should I ask him?ā
Garry Tan:
Totally.
Turner Novak:
I throw it in all of them. But I feel like Perplexity always gives me the best on average.
Garry Tan:
Yeah. Itās incredible times we live in right now.
Turner Novak:
Yeah. One last question for you. Do you have a favorite CEO, founder or business that youāve gotten inspiration from over the years? Who are you drawing from the most today? And maybe itās just Brian, because it sounds like heās had a pretty big influence.
Garry Tan:
I love Brian. Getting to work with him on the board is insane. And then, actually, watching Aaron Levie from Box has been really, really wild. I only befriended him maybe a year and a half ago. And he is very, very funny. You should ask him for magic tricks if you ever meet him.
Turner Novak:
So, Iāve had him on the podcast. Actually, Iāve never talking about magic.
Garry Tan:
Oh, my God. Yeah. But I mean, I think thereās something really special about Aaron and that heās stuck with Box all of these years. And then the way heās thinking about how AI is going to change SaaS, I think is just ... I mean, I really look to him for just like, āHowās this all going to change and work?ā
Turner Novak:
Yeah.
Garry Tan:
Really, I love his tweets. So, Iāve learned a lot from him.
Turner Novak:
Yeah. Itās so fascinating where you would think all of these scaled incumbent software companies are so well positioned to just capture everything thatās going to happen in AI. And maybe itās sort of. That happened in some cases. But thereās also a lot of cases where LLMs have been around for three years and the product hasnāt changed at all, aside from the website saying that it has AI or they throw in a chat box.
Garry Tan:
Oh, my God. Whatās funny is Iām pretty sure you could just make a long, short fund on whether or not the CEO is a hacker and has used Claude Code. And that would be ... It would outperform the S&P.
Turner Novak:
Probably. Yeah. Well, actually right after this, Iām recording with Chetan and Benchmark.
Garry Tan:
That is the best.
Turner Novak:
I donāt know if thatās going to come out after or before this one. I have to figure out the schedule. But thatās one of the things weāre going to talk about.
Garry Tan:
Fantastic.
Turner Novak:
Is this universe of all these software companies, and the TAM is trillion dollars in software spend, especially when you consider how itās going to change with AI. And it just seems like a lot of the big players have not done anything yet. And will they win anything or will it mostly be startups?
Garry Tan:
Yeah. I think people are underestimating to what degree software is a very, very small part of the majority of peopleās lives in the world. And the reason why is that there are only 10 to 20 million people in the world who are actually good at code. And then weāre about to enter this other moment where actually billions of people can actually just very directly create their own stuff.
And then the hope is, I mean, my hope would be actually a lot of things get better, faster, cheaper. We need more markets and we need more ability to enter, basically, capitalism. We need actually, not fewer markets, but more markets, actually. And I think that thatās whatās going to happen. Thereās a tiny, tiny percentage of GDP is touched by software at all. And then the result is all kinds of just loss.
Just this person was there and they could have ... I mean, Uber or Airbnb or any of these marketplaces are good examples of there was not going to be a market, and then the market was created out of thin air and then a lot more of it is done. And then whatās the result? The result is, I can get around. The nature of cities has changed really fundamentally.
Turner Novak:
Yeah.
Garry Tan:
Millions of people, perhaps maybe tens to hundreds of millions of people who maybe wouldāve been hurt by drunk drivers. That doesnāt have to happen anymore.
Turner Novak:
And now, sober drivers also.
Garry Tan:
Thatās right.
Turner Novak:
With Waymo.
Garry Tan:
Thatās right.
Turner Novak:
A car objectively, I will not injure a human and will just follow all the rules. And I was scratching my nose or I got distracted for a second, and just objectively the software decides and knows, āOh, this door opened and I sensed it within a split millimeter of a second and got out of the way.ā
Garry Tan:
Yeah. Yeah. I mean, thereās basically extreme tech pessimism. But to me, itās like tech is here to make things better. And then, I mean, granted, I have a bias. Yes, I am surrounded by the top 1% of people who earnestly are trying to build technology in service of humanity.
Turner Novak:
Yeah.
Garry Tan:
And if that happens, all of these things can happen. And so, I mean, even with AI, the dominant narrative is somehow all the jobs are going to go away. But I really like what Ryan Petersen from Flexport says. He says, āLook like humans just want stuff, man. Youāre never going to get to the end of the list of wants that people have in the world and technology in the hands of people who can harness it and put it into the market. Thatās the mechanism that actually literally improves the life and well-being of every person.ā
People today, I think middle-class people today live lives that the kings and emperors of 200, or even 100 years ago or even 80 years ago, they couldnāt even dream of having the life that we have just day-to-day, our access to healthcare, our access to transportation, our access to education, they couldnāt even imagine that ...
Turner Novak:
You can fly around the world.
Garry Tan:
Yeah.
Turner Novak:
Just you can go to bed and wake up on the other side of the planet. Thatās insane.
Garry Tan:
And then soon itāll be the moon and Mars.
Turner Novak:
Yeah. Actually, probably one of my most insane stats, have you ever heard the one about firewood in the US?
Garry Tan:
Whatās that?
Turner Novak:
Firewood used to be like 25% of US GDP.
Garry Tan:
Oh, my God.
Turner Novak:
Just the sale of firewood.
Garry Tan:
Thatās insane.
Turner Novak:
Which is thatās not even a line item anymore. Itās probably a couple, maybe 50 million bucks that we spend a year on firewood. That used to be the majority of the economy was people buying and selling firewood.
Garry Tan:
Iāve got to look that up. Iāll make a YouTube video and Iāll credit you.
Turner Novak:
Okay. Yeah. No. Itās just super fascinating.
Garry Tan:
Weāll do a collab.
Turner Novak:
Weāll do a special. Iāll try to find the tweet and Iāll send it to you. But itās just crazy and itās just like shows, itās like this chart just kept going down and itās like, 0.02%, I donāt know, 50 years ago, whenever this chart was.
Garry Tan:
Yeah. Go back a couple 100 years and itās subsistence farming. And instead, we get to sit here, hang out as friends, just like talking about ideas. This is so cool.
Turner Novak:
Yeah. This is a lot of fun. Thanks for doing it.
Garry Tan:
Thanks for having me.
Stream the full episode on YouTube, Spotify, or Apple.
Find transcripts of all other episodes here.