YCombinator Economics

YCombinator is a company that provides seed funding for startups. It was founded by Paul Graham and others. They have Summer / Winter Founder Programs each year where they give money to college students to start their own companies. YCombinator seems to be quite successful - Kiko, ClickFacts, and the hugely popular Reddit have already been released.

But how much money are they making with this? Any self-respecting engineering student will answer this with a formula, so here's my very simple approximation:

profit = n*(s*p*a – v)

And here's what all those letters mean:

Please note that the formula above does not include any fixed costs: I left out salaries for Paul Graham & the crew, rent for 135 Garden Street, and costs for organizing Startup School. I didn't feel brash enough to give estimates for all of this. I also disregarded capital costs: Clearly, an investment that's bound for 5 years in a startup would need be valued differently than money that stays put for just a few months.

As I mentioned above, the two random variables in the equation are the percentage of startups succeeding and the average acquisition sum per startup. Time to fiddle with Excel and to get the profit plane for each YCombinator round:

Let's compare the YCombinator approach with that of traditional VC funds. Conveniently enough, the same formula as above holds, with slightly different numbers. We'll assume that a VC will invest $1 million in 10 different companies, and gets a 66% stake in return. Once again, we'll exclude fixed costs, such as office rent or the fund manager's salary. Here are the results:

I find two things particularly interesting:

• YCombinator can get very rich, while the VC can get insanely rich. In our model, the ideal – but unlikely - case where all startups are successful and are sold for $20 million on average yields about $12 million for YCombinator, while the VC makes $122 million. This is due to the fact that YCombinator's small initial investment buys them much less of a stake in a company than a VC would get.
• YCombinator is safer. The blue part of each graph is the"under water" portion, where the fund loses money. By making smaller bets, Graham and Co. start making money even at lower success rates and acquisition sums.

(The same in mathspeak: the partial derivatives dprofit/dp and dprofit/da per dollar invested are larger for the VC, but the VC's profit stays below 0 for longer since his initial investment is much larger than YCombinator's.)

Of course all of this is just an approximation. In addition, we're assuming that one of Paul Graham's startups is equally likely to succeed as a VC-backed startup. A case could be made that this is true.

Number crunching aside, I think that YCombinator has a very interesting business model. There's a huge chance that it will turn out to be profitable: Reddit is already very successful. Even if it sells for just a fraction for Yahoo's $30 million acquisition of del.icio.us, YCombinator is in the black.