It's definitely a combination of issues. Intel's 10nm was supposed to launch back in 2016 - it was extremely aggressive - keep in mind that it's 10nm is roughly equivalent to the 7nm that TSMC is shipping currently. The problem is that between 2014-15 when Intel was developing 10nm and now, a bunch of new techniques have been developed in order to make that shrink easier, notably SAQP at that size - but Intel had already invested so much into their technology and their engineers were basically saying "don't worry it's right around the corner". The problem is they kept saying that year after year. Intel's solution was to go back and reconfigure it's 10nm process with some of the newer stuff and that's what allowed them to ship 10nm products in 2018 - but like you said, they can't seem to get the switching performance out of their architecture on 10nm and thus they never really shipped anything on it except for a few mobile parts - where the power consumption does matter. They aren't behind - they are roughly the same with Intel having an advantage in some characteristics and TSMC in others. I don't think this will change with Intel's 7nm vs TSMC's 5nm. The bigger takeaway is that Intel completely lost it's process advantage. For year it was a major component to Intel's success and now it's almost an anchor for them. AMD is able to freely shop around it's fabrication to whoever can deliver the best process but Intel is stuck with what it's team can come up with.