I would agree with crazy growth done for now. Once the next gen chips are out will be interesting if some of the big companies bring back crazy growth.
Personally I'm thinking BTC price will affect more the crazy growth than the next gen chips. I don't doubt there will be growth spurts, probably 2 or 3 mid-teens adjustments are likely, but I think it will be more measured.
For giggles, let's assume next gen chips are still 28nm, just more efficient, able to hit say .35w/gh which would be a 30% gain on the best existing chips (.5w/gh). Let's assume also a June/July delivery timeframe.
I would believe that the largest adoption of those rigs would be with mining farms running on older hardware, say anything that's >6months and over .8w/gh. For those miners, they've been running the older gear longer, probably at or close to positive ROI where upgrading will benefit them in the long run because of reduced power expenditure.
The mining farms who are spinning up now on the last bits of stock from SPTech and BMTech wouldn't (well, I should say "shouldn't") be in a position to drop large chunks on buying brand new hardware when the ones they just spun up are barely 2 months old. I honestly don't think they'd be able to match the same 30% difference in the price of selling the old hardware (in other words, they would get 30% positive gain from the new chips for power consumption, but would be selling their hardware for more than a 30% loss, thus equating to a negative total ROI).
The first group, if that happens, will then start a cascade effect for smaller miners, where they can now also replace the higher wattage gear they're running with much cheaper use equipment (say those running S1's for sure, and some S3's will now pick up used S5's on the cheap). IT'll turn into a sine wave of diff adjustments, dip while old hardware is turned off and sold, then an uptick as the hardware reaches the next set of miners and gets powered on, and then lather, rinse, repeat down the chain.
But, the whole thing with this is the price of BTC. If it stays low, then even the large farms aren't turning as much of a profit, and that should result in the ability to purchase and deploy less hardware than they could if BTC were higher. If BTC goes up considerably, then these same places could deploy much larger amounts of hardware, which would drive difficulty through the roof, quickly, much like was seen from middle through late last year (June - November).
If hardware isn't being bought at a rapid pace, that leads manufacturers to not have as much revenue and on-hand capital to push forward on newer, more dense chips like 20nm and 16nm.
Then again, we can speculate till we're blue in the face... who the hell really knows what will happen...