"By supporting the startups in return for some stock and a commitment to use Microsoft development software, product platforms, and operating systems..."

MS on the desktop is okay, Visual Studio is in a class of its own, but using their server-side stuff? And paying for it? What's he smoking?

My guess is that Graham believes Microsoft couldn't/wouldn't provide proper incubation for an inexperienced group. They'd immediately try to impose a "big company" mentality. As a corporation, he is probably right.

Microsoft could fund an investment group that essentially used the company's influence to help startups externally. If that same group shielded a startup from Microsoft's corporate influence internally, I think they could be a powerful incubator, not to mention a phenomenal asset. Unfortunately this will never happen.

I think the YC model wouldn't work for a company like M$ because there is way too much drama. Founders get into fights, startups make horrible paperwork mistakes and ideas and markets change all the time. There is way too much counseling and looking after required to really do a YC style investment program with a large number of young founders.

M$ would do better to give money to a company like YCombinator and just encourage all the startups to use microsoft products.

"...and just encourage all the startups to use microsoft products."

I can't see that happening any time soon.

or Microsoft should invest in YC startups.

My relatively uninformed hunch is that, per hour of time expended, YC is not nearly as profitable as traditional venture capital. YC probably gets a much better nominal return-on-investment, but a more modest return-on-effort. YC has to provide mentoring in addition to capital, which means that there's an inherent bottleneck in the amount of business it can do.

I suspect Paul Graham is running Y Combinator not because he wants to maximize his profits, but because he enjoys mentoring bright young entrepreneurs. So although his return-on-effort is lower than it would be in traditional VC, this isn't problematic, because the utility of said effort is positive, and YC is presumably more than profitable enough.

How we'll do compared to VC funds is a hard question to answer because of the huge variation in their performance. The median VC fund actually loses money. I hope we do better than that.

It's true that we're not doing YC mainly for money, though. But we're not doing it just for fun either. The main thing that drives us, I think, is that it seems an interesting hack.