For the pumping mode there is a Tu Delft dataset : tests on “Hydra” (14 m²) with 10 m/s wind speed result in an average power of 5.75 kW, and an average power generation phase of 13.8 kW (table 8, page 44).
On said table 8 dataset 1, 2, 3, 4, 5 concern “Mutiny” (25 m²). In 5, wind speed is 8 m/s, average power is 5.19 kW, average power generation phase is 14.1 kW. All these values concern mechanical power.
To compare with an equivalent wind turbine comprising a rotor of 6 m diameter, Cp of 0.45, 7600 W at 10 m/s wind speed: a blade of 3 m weighs 9 kg (http://atv.tm.free.fr/EN_blades.htm), while the weight of Hydra or Mutiny with the pod is 20-25 kg.
In a side: regular power, reliability, durability, no land use. In the other side, land use 1 km², space use 0.5 km³, erratic power, hard control, no reliability, but a little less weight.
So AWES for utility-scale is not for tomorrow, if it comes some day. Note that the paper dates from 2012, and no significantly better results came after. So the “newborn baby” is getting old without growing much. Perhaps some niche market such as some individual uses could be found.