Y Combinator: From 2005 Startup Funding to $500B Companies

March eleventh, two thousand five. Two people are walking home from dinner in Harvard Square, arguing about what's wrong with venture capital.

That's a fun date night.

She's Jessica Livingston, working at an investment bank, waiting on a VC firm that's taking forever to decide whether to hire her. He's Paul Graham, who sold his startup to Yahoo and keeps meaning to do angel investing but never has.

Keeps meaning to. That's very relatable.

Seven years he'd had the money sitting there. And somewhere between Harvard Square and the house, the idea gels. By the time they turn onto Walker Street, they've decided: start their own investment firm, and Jessica quits the bank to run it.

So how much money are we talking? Because that's the part that decides whether this is real.

Graham put in a hundred thousand. He recruited Robert Morris and Trevor Blackwell, who each put in fifty. Two hundred thousand dollars, total.

Two hundred K. That is... a nice car.

And it wasn't even called Y Combinator. The first name was Cambridge Seed.

Which sounds like a garden centre.

They dropped it within days, because they realised this could be national and they didn't want a name tied to one city.

Okay, but what was actually broken? Startups got funded before two thousand five.

Badly. Jessica's description is that your only option was finding a rich relative or friend to hand you some cash. After that, the next stop was VCs who wanted to put millions in. Nothing in between.

So there's a gap where you just need enough to eat for six months.

And the paperwork was chaos. Graham's line is that often neither the investor, nor the founders, nor the lawyer knew what the documents should look like. He points out Facebook started life as a Florida LLC.

Florida? Facebook?

That's how random it was. YC's actual product, at the start, was just standardised seed funding. Same terms, every time.

Hang on, that's it? That's the whole innovation?

No, and this is the part they didn't plan. Jessica says it straight: neither of them knew anything about angel investing. So they decided to learn fast by funding a bunch of startups all at once, over one summer, in Cambridge.

So the batch model — the thing everyone copies — was just them cramming?

It was a shortcut for their own inexperience. Graham later said funding startups synchronously was the most important idea they had, and they didn't realise its significance at the time.

That's genuinely my favourite kind of accident.

They called it the Summer Founders Program. Paul wrote a website, put it up, and Jessica's memory is: gosh, we hope people apply.

So how many did they take that first summer? Twenty? Thirty?

Nine companies.

Nine. Out of which anything survived?

One of the people in that inaugural batch was Sam Altman. Ten years later he founded OpenAI, which started as YC Research. OpenAI is now a five hundred billion dollar company.

From a nine-company summer camp in Cambridge.

The next few years were tiny, though. Eighteen companies in two thousand six. Thirty-two in two thousand seven.

Those are class sizes, not portfolios.

Then two thousand eight: forty-three companies, and the total money they raised after Demo Day that year was about fifty million dollars. Which, by YC standards, is nothing.

That's the financial crisis hitting, presumably. Rough year to be selling a dream.

Rough year to be a startup full stop. Airbnb was broke in two thousand eight, before YC. They fixed it by selling eight hundred boxes of cereal.

Cereal. Actual breakfast cereal.

Forty thousand dollars of revenue from it. Then they did YC in the winter oh-nine batch and went public in twenty twenty at over a hundred billion.

Okay, I want the arithmetic on the deal now. What does YC actually get?

Today they invest five hundred thousand dollars. A hundred and twenty-five of that buys a fixed seven percent, and the other three hundred and seventy-five goes in on an uncapped SAFE that prices later.

Seven percent for a hundred and twenty-five grand still sounds steep to me.

Graham thought about that. The guy who funded his own startup, Viaweb, was a family friend called Julian Weber. Ten thousand dollars, plus setting up the company and teaching them business — for ten percent.

And Graham resented that?

For about one second. He thought, what a good deal Julian got. Then realised that without Julian, Viaweb never makes it. Good deal for both sides. That's why he knew there was room for YC.

So where does two hundred thousand dollars and nine startups end up?

As of twenty twenty-six, YC has invested in over five thousand six hundred companies, with a combined valuation around six hundred billion dollars. Stripe, Coinbase, DoorDash, Dropbox, Instacart, Twitch.

And they still run it the same way? Three months, dinners, Demo Day?

Four batches a year now, in San Francisco. Founders get sorted into small groups, eat together weekly, and the alumni talks are strictly off the record — because the real story of a startup's early days is messier than the one they tell later.

I'd pay for those recordings.

The usable bit, I think, is the bit they stumbled into. They didn't know how to invest, so they did nine at once to learn faster. The shortcut became the product.

Do the thing badly nine times instead of carefully once. I can work with that.