If microsoft was to fund 500 internal startups tomorrow, what's the chance any of them would even produce any real revenue? Out of the thousands of startups that are active today, how many of them have even the remote potential to generate 1b in revenue, or even be successful? This isnt even considering how maintaining 500 internal startups would be a micromanagement nightmare. How can you tell if someone is worth keeping? With the failure rate of startups so high, does that mean you should fire startup cells when they fail?

The resources that would be used to maintain such a division would probably be much better used focusing on one new, amazing product - backed with loads of market research and QA. You cant ship that kind of product with 5 developers and a years worth of hotpockets.