This is most people's first reaction - they assume that if YC is getting bigger, some limited resource must be getting diluted. But that's actually not the case: there is no limited resource that isn't being increased. And it also misses an important trend that works in the opposite direction: the network effects that make YC more valuable for companies when the batch is larger.
These network effects aren't obvious to outside observers (actually some of them weren't obvious to us at first!), so here are some of them.
1) The larger the batch is, the more companies there are that are similar to you. The author of the parent comment described this as a "cohort of folks going through the same thing as you".
Suppose you're building a fintech company in India. In an earlier batch where we only had 75 companies, you might well have been the only India fintech company in the batch. Now, there are probably 4 other companies in the batch that are also India/fintech. Those companies are likely to be the ones that can be most helpful to you.
2) Related to (1), having more scale in each vertical has allowed us to specialize our program and deliver more specific advice for different kinds of companies.
We now have specialized advice, content, and events for devtools, fintech, saas, consumer, enterprise, bio/healthcare, hard tech, & hardware companies. Much like AWS's large scale allows it to offer more and more specialized products, our larger scale allows us to offer more specialized programming.
3) For a large percentage of the YC batch, other YC startups and YC founders make up an important customer segment. The more YC companies there are, the more potential customers they can get from this community.
Related to that, the YC network is also a powerful tool for getting introductions to non-YC companies. Suppose you want an introduction to a key person at Apple. Apple isn't a YC company, but the YC network of founders who used to or currently work at Apple will be very helpful for reaching the right person. And that gets better when we have more founders.
4) The YC companies effectively collectively bargain for a large variety of things: discounts on services (worth over $1M / company now), press exposure, investor attention, etc. The more companies in the pool, the stronger our bargaining position becomes.
5) The more highly successful companies YC has, the more brand recognition YC's name has. That's obviously good for YC, but it is also good for YC companies. One of the hurdles new companies have is getting other people and particularly big companies to take them seriously. The YC brand name is now pretty helpful for accomplishing that in the tech world; we're still getting to the point where it works in every other industry and that requires more scale.
6) With more YC companies, we're now able to run https://www.workatastartup.com/ to help YC companies hire. From the standpoint of someone who is looking for a job at a startup, the more (successful) startups there are on Work at a Startup, the more useful the site is.
7) Most of the advisors who work with companies (we call them "group partners") are YC alumni who started a successful company. Because of this, when we fund more successful companies, it expands the pool of potential great group partners we can hire. To scale to the next level, we need more group partners (a.k.a more expert advisors). The trick of hiring our own alumni allows this seemingly limited resource to actually scale more-or-less proportionally.
8) As YC becomes better known, more great founders become interested in applying. YC is well known in silicon valley but still has a long way to go to be well known everywhere in the world. We've consistently noticed a trend where after we fund our first company in a new country, we get a surge of applications from that country. When we have more companies in each country, we make YC more accessible and useful to new companies in that country as well.
Interestingly, some of these network effects have just become apparent to us in the past couple of years - we had to reach a certain scale to even see them. This makes me think there will probably be other network effects that kick in as we reach even larger numbers.