I'll speculate that this is a problem with all large companies: the price of failure for the team involved is huge (unemployment?), but the benefit of success is small (continued employment and maybe a bonus). If the incentives are so structured, it's no surprise that you get risk-averse behavior.
Very few teams actually get laid off just because the product fails. The only time it's happened in Google's history was Google Radio Automation, where they brought on a bunch of people with very specific domain knowledge of the radio industry and then had nowhere to put them when they decided they weren't going to get involved with radio after all. Most of the time (assuming they don't quit), the team just gets absorbed into other products.
I assume things are similar in other large tech companies.
I'll posit that the real problem with big companies is the opposite: the price of failure is mild (you get reassigned to another team and get to do more cool stuff) and the benefits of success are also mild (you might get a promotion and an iPad or something). When your actions have little effect on the outcome, your actions tend to regress to the mean. In a startup, where failure means you starve and success means you never have to work again, you have a much bigger incentive to go the extra distance.
From an innovation standpoint, it might be interesting to go with the model of high benefits of success, low cost of failure. That relative skew is what has arguably made the US the leader in this field in the macro.
Startups given the market for engineers in particular actually only have a temporal starvation penalty. It's not like they can't get another job after they fail.
Doing this in a large company might lead to even more innovative products because startups are so under-resourced and governed by promises to investors that they risk becoming myopic and less agile.
the reward for successful skunkworks at a large co is typically promotion (serious promotion to very high-paying, visible roles). That can be very rewarding for certain personality types.
I have to agree from my experience at IBM. Those rewarded tend not to be rewarded much by industry standards (years salary at best) and the entire team or even the most meritorious contributors are rarely rewarded.
I know of at least one team that came up with something major that asked for their compensation to be tied to their product success having their idea moved to another team because "they should do it for IBM purely". Of course that product failed for lack of vision and focus.
