On the Y Combinator Summer 2021 batch

At Rebel Fund, our job is to invest in the top 3–4% startups in each new Y Combinator (YC) batch. To accomplish this, we build what we believe to be the most comprehensive database that exists of YC startups — including over 60,000 data points on ~400 startups in just the recent YC Summer 2021 (S21) batch. This database provides deep insights into the batch and helps us zero in on the most promising teams and ideas to invest in.

I’ve written before about our proprietary Rebel Theorem machine learning algorithm that analyzes these tens of thousands of datapoints to score and rank each YC company in each batch according to its likelihood of achieving a great outcome for early investors. The goal of this post is to share some statistics and insights on the YC S21 batch that we feel comfortable publishing.

Geography

So where are these YC startups located?

Even though YC is based in Silicon Valley, it attracts startups from all around the world. While a slight majority hail from North America, there is a significant international contingency in YC batches these days, including startups from Europe, Asia, Latin America, and Africa.

When it comes to international startups, YC has a strong leaning towards emerging markets, which becomes clear when we break down the startups by sub-location. For example, the Asian startups are mostly from southern and southeast Asia (e.g., India, Indonesia, and the Phillippines) and Latin America and Africa are mostly emerging markets as well.

Given the rise of globalization and remote/distributed companies, it’s important to disaggregate the location of a startup from the market that they serve. When we break down the startups by market location, we see an interesting trend: Developed market startups in Europe and North America tend to serve the US / World markets, whereas the emerging market startups tend to focus locally. Anecdotally, we’ve noticed many of these emerging market startups are taking business models that have already been proven in the developed world, and localizing them to their respective region.

Industry

The break down of YC startups by industry always fascinates me, since YC sits at the bleeding edge of innovation so acts as a great leading indicator of where the technology world is heading.

My popular On $300B of Y Combinator Startup Success post speaks to how YC startups have been migrating towards B2B, Fintech, and other “old economy” sectors ripe for disruption like Healthcare & Industrials, and this latest batch is no exception.

Since the industry categories above are quite broad, we’ve broken down the startups further by sub-industry. This shows just how diverse YC startups are, and leads to a few interesting observations:

• Engineering, Product and Design is the single largest category, which shows how common YC startups developing tools for other startups are
• Marketplace is the second largest category, probably due to the strong network effects and defensibility in this category
• Despite all the crypto excitement these days, Cryptocurrency startups only accounted for 1% of the S21 batch

The last point may be surprising to many, especially with the mega-successful Coinbase being a YC company. I think it’s due to a lower supply of crypto applicants to the program since they already have ample access to capital, plus YC’s historic aversion to investing into sectors that feel overhyped. That said, a lot of crypto applications (NTFs, DeFi, trading, etc) are reaching a critical mass of adoption right now that I think could change this dynamic in the near future.

Age

There is a common misconception that startups need lots of history and revenue traction to get accepted into YC. This isn’t true — the vast majority of S21 startups are 2 years old or less, and many of those pre-revenue. While there are also some startups in the batch in the 3 to 5 year old range, they’re rarely older.

Founders Education & Work Experience

At Rebel, most of the data we collect on YC startups is related to the founders themselves. Every investment we make is ultimately a “team bet” so we track the work history, educational history, and prior successes of every YC founder in meticulous detail. Here I’ll share some insights on the batch as a whole, starting with founders’ educational background.

YC founders are a well-educated bunch, and 54% of S21 batch startups have at least one founder with an advanced degree. The most common degree type is a Masters (non-MBA) in a technical field, reflecting YC’s long-held bias towards technical founding teams. Next are MBAs, followed by PhDs — which are also typically in technical fields. There are a few MDs in the batch as well, often building something in health or biotech.

Next we’ll look at how many years of total work experience S21 founders have.

Despite the stereotype of Silicon Valley founders being a bunch of young college-dropout wonderkids, the overwhelming majority of YC founders today have 5–12 years of work experience, with a median of 9 years experience. A small but significant subgroup have 20 years of experience or more.

So how old are these founders? We don’t know exactly, but estimate each founder’s age by adding 18 to how many years ago they started their first university degree program.

When we do so, it’s clear that YC founders cover a huge age range, with teenagers all the way to 60+ year olds in the batch. However, the vast majority of YC founders fall into the 24–34 year old age range with a median age of 30 years — probably a bit older than many would expect — but consistent with the median 9 years of work experience above.

When we developed our Rebel Theorem machine-learning algorithm to predict YC startup outcomes, we found that the single greatest success predictor is whether the founders had a successful startup in the past. YC must know this as well, as they have a clear bias towards repeat founders, while also still reserving plenty of spots for first-timers.

As you can see in the graph below, there were over 270 founders in the S21 batch who had started 2 or more companies in the past, representing close to half the founders that we tracked! Note the “1” category sometimes reflects their YC startup and sometimes a prior startup, depending on when we gathered the data, so many of them are repeat founders as well.

If we look at the founders’ cumulative years of work experience building companies as cofounders, we can see that most of them have several years under their belt, and a decent sized minority have 10 years of cofounder experience or more.

Few realize just how mature most YC founders are today. It’s safe to say the “typical” S21 founder is a 30+ year old experienced repeat entrepreneur with years of experience honing his or her skills.

Founders Top School and Employers

Silicon Valley has long demonstrated a bias towards “pedigreed” founders, and with good reason — there’s data to suggest that founders that graduated from certain top universities and/or worked at certain top employers are far more likely to have successful startups than their peers.

In the fantastic book Super Founders by Ali Tamaseb, the author analyzes 30,000 data points on hundreds of startups to explore the best predictors of billion-dollar outcomes. Amongst other things, he found that graduating from certain universities and working at certain employers is indeed predictive of a founders’s startup success. When we developed our proprietary Rebel Theorem algorithm, we discovered the same.

Since Mr. Tamaseb has published his findings but the inner workings of Rebel Theorem remain confidential, here we’ve analyzed the S21 batch startups according to his top school and employer rankings (though ours aren’t too dissimilar). As you’ll see in the charts below, YC may have independently reached similar rankings.

The top universities attended by S21 founders align incredibly well with those Mr. Tamaseb found to be most predictive, with Stanford, Harvard, MIT, and UC-Berkeley leading the charge. In fact, if you treat the Super Founders rankings as a baseline, you can even see where YC introduces its own biases, with West Coast schools a bit favored over East Coast, and with IIT in India being disproportionately represented (I suspect due to YC’s efforts to recruit founders in India)

When we compare the breakdown of S21 founders’ former employers to Mr. Tamaseb’s top rankings, we can see that YC still demonstrates similar tastes, though not quite as similar as with top universities. Both have a strong bias towards ex-Googlers, but YC’s #2 is Tamaseb’s #5 (Microsoft) and YC’s #3 is Tamaseb’s #6 (Facebook). We can also see that YC stands in stark disagreement with Tamaseb on Yahoo! and HP alum, but that could simply be due to Tamaseb working off an older data set.

In any event, our top university and employer findings suggest that YC is selecting founders very consistently with how data science suggests it should — whether via YC’s own algorithm or finely tuned YC partner instincts.

The charts above illustrate just a few of the dozens of data points that Rebel collects and analyzes on every YC startup in the batch to help inform our investment decisions. We hope you’ve enjoyed nerding out on YC startup and founder data as much as we do!