We've always approached that question the way hackers approach scaling software: we don't know where the limits will be; they'll probably turn out to be higher

We've always approached that question the way hackers approach scaling software: we don't know where the limits will be; they'll probably turn out to be higher than we'd expect; the only way to find them will be to repeatedly expand, and hit and remove bottlenecks. So while we often hypothesize about what YC would look like if we funded e.g. 10x more startups, we know better than to plan to. It's a question to be decided empirically.