We can just agree to disagree....
I don't know that we're actually disagreeing.

My point is that a manufacturer is going to release a technology/product at the time when they predict it will yield the greatest profit. Likewise, they're not going to 'waste' money on R&D with no expectation of an ROI - and as a general rule, an early ROI is better than a delayed ROI, unless it's predicted that there's substantially more profit to be gained from a delayed release.
In the example of the 120 MP APS-H sensor, I actually doubt they produced incrementally increasing versions - that may not have been cost effective. If it were my R&D money, I'd have had a team do some computer modeling to suggest the maximum feasible density, then attempted to produce three versions - that predicted max, and something lower and something higher. Three shots on goal, simultaneous, rather than racheting it up with repeated costs for each increment. But honestly, I can't speak to any sort inside knowledge of how they did it.
Certainly, in some industries technology is 'held back' - you mentioned defense, so compare consumer GPS systems with military GPS systems. The latter is always ahead of the former (although the companies are still making a profit from their technology, albeit from a different sector.
Back to releasing a product when it will generate the maximum product, I do think that's the key driver - and I definitely acknowledge that that is not always as soon as the technology is available and producible. In the specific case of boceprevir and telaprevir, and the general case of drug development, the maximum profit results from the earliest possible release. If a cure for HIV were developed (and some are being actively worked on), it would be released as soon as possible (and I do speak from direct knowledge in this case - I happen to hold a senior position in a large pharma company).