If it ships on time (they already had to shift the roadmap once), Cannonlake will be available in the second half of 2017. If things continue as they have, Apple will likely be releasing the A11 right about that time. Looking at past iterations, single-core performance only at the iPhone TDP level (ignoring iPads, that is), the A5 was only slightly faster than the A4 (though it added multi-core), the A6 was well over 3 times as fast as an A5, the A7 was twice as fast as an A6, the A8 was only (or "only") 15% faster than an A7, and the A9 is a little less than twice as fast as an A8.
Leaving aside the slightly earlier release of the original iPad, overall, the A9 is almost exactly 12x faster than an A4 in single-core performance in a phone, and was released a bit over 5 years later. That's a pretty good curve. (The difference would be 21x faster if you look at multi-core.)
Over the same time period, using a 45W mobile TDP and whatever was the top of the line, Intel went from a Clarksfield 840QM i7 to a Skylake 6920HQ i7. On the same single-core benchmark, the latter is a bit over 2x faster than the former (3x in multi-core).
Point being, year-over-year since its introduction A-series processor performance increases, thus far, have been huge, while Intel performance increase has leveled off considerably in the past few years. There will definitely come a point of diminishing returns, at least in terms of performance-per-watt--something Intel hit years ago--but if Apple isn't there yet there's no telling what the performance of an A11 will be. It could be a modest improvement, or it could be a literal multiple of current A9 CPUs.
If the average exponential rate of improvement over the past five years were to continue for both chip series (which of course isn't a given; 15% increase per year for each new Intel mobile generation, 64% per year for each new A-series), Cannonlake will be about 32% faster (1.3x) than the current top-of-the-line Skylake at 45W, and the A11 will be about 170% faster (2.7x) than the A9. Were that unlikely progression of technology to actually happen, that would make the iPhone 7S significantly faster than a quad-core Cannonlake laptop in single core performance, and getting close in multi-core; a hypothetical iPad Pro 3 with A11X would almost certainly be faster in both.
All that probably won't happen, but if we're going based on history alone, that's the future, and if you told someone in January of 2010 that Apple would ship more devices with its own custom designed CPUs than the entire desktop and laptop PC market, total, in 2015, no one would believe you.
In any case, it's not like Apple's chip design team is sitting around not doing anything while Intel is producing new designs.
Given Apple's way of doing things, you may well be right about that.
That said, given that Apple sold around 5 million Macs--at, relatively speaking, extremely high prices--in the last quarter, and their laptop/desktop sales have been on a consistent upward trajectory for years, they're probably not in any hurry.
I do find it amusing how many people--geeks in particular--make the extremely flawed assumption that because they want a particular capability in a piece of technology, or have some particular use case in mind that is vital to them, personally, everyone else will. Doesn't mean what you want to do isn't valid, but the geeks reading this site are not representative of the average consumer, and the reality is we're vastly outnumbered by people who have entirely different needs and use cases for a computing device.
