I think the main reason is the 'instantaneous culture', when we capture a picture on our smartphone we want it to be good enough to be posted to Instagram, Facebook immediately; image enhancement over cloud adds too much latency and is dependent upon variables such as network. Leading brands have figured out taking good pictures without doing that with a combination of algorithms(image enhancement neural networks) and processing power (CPU, image processing co-processor).

But still, not all manufacturers have caught up even with the similar hardware due to their lack of expertise with the software side of the things. Hence I think there is a need gap to be filled here.

I may not be alone, Adobe recently hired Marc Levoy, who's responsible for the software behind Google's pixel camera supposedly to develop a universal camera app. I'm not sure whether they would use cloud for their computational photography, although it does make lot of business sense to Adobe.

I feel the need gap is large enough for several players, also image enhancement neural networks are fairly public knowledge nowadays and so there's not much a barrier to entry here.