In an interview with Technology Review he complained (or should we say boasted?) about the high amount of ideas that makes it into real products. According to Rattner about 50 percent of the ideas Intel scientists propose end up in retail products and he thinks that way too high.
Rattner believes Intel's researchers need to take more risk and he would like to see success quotas of 10-15 percent. This will give researchers more room for experiments and may lead to more innovations.