Surprisingly my medium sized employer of ~1000 employees has had surprisingly good experiences with internal startups, during the last three years there's been three failures, one small success and one major success.

The startups were very cross-functional and very isolated, and the only external pressure was the time until the product becomes profitable. Being isolated from the main organization and a small size allowed the startups to try new things fast, but knowing when to stop allowed even the failures to be considered as successful experiments for scrapping the very different products early enough.

Before these the company had tried to expand its business by bying another smaller company, which was just a huge disaster, wasting years of work and tens of millions for trying to keep the one new business alive just because for too long it was regarded as too big to fail. The good thing from this was that the old CEO "found new challenges" and the new one started driving the startup model with an emphasis on knowing when to stop.