As a conclusion (from for example Mike Barnard) it could be stated that AWE is not reliable, not efficient, so not viable.
As an opposite conclusion (from M. Loyd’s paper) it could be stated that AWE is efficient, so viable (other issues such as reliability and power versus space used being set aside).
I think we are far from making a marketable utility scale AWE. But that doesn’t mean it couldn’t happen.
The evaluation of AWE viability can be based on inherent and circumstantial factual elements with all the nuances in between because one does not know what is only inherent or only circumstantial. A software could be elaborated in order to facilitate the choices among AWE methods, then among technologies, without forgetting the characteristics of the chosen site, then the favored hypotheses. Then iterations would lead to definite better what remains inherent, what remains circumstantial, with all the possible degrees of evaluation as well as the possibility of bridges between these two categories. During investigations and testing a technical issue which could be solved according to reasonable hypotheses would go from inherent category to circumstantial category, increasing the possibility of a viable AWE.
Let us take Makani as an example of a leading AWE company. So as we can notice Makani’s failure (both crashes and no efficiency), we still cannot tell the part of what is inherent within the failure, then what hypothesis of improvement can be favored: weight in flight, higher kinetic energy downwards due to both weight and speed during large loops, Y-bridle tether connection not preventing destructive roll…
That said a significant element is what Makani or another company claimed and what (high) echo the AWE community in general accorded to their claims. Unfulfilled promises are not an element of credibility for AWE. So we should stop talking about marketing or claiming AWE as the future of wind energy (compare with what is really happening in marketed wind power) for a given year…