manufacture and assembly of components, and life-time & maintenance of parts should provide a 'break-even' formula for energy in vs. energy out. As long as building the machine doesn't cost more energy than it produces over X time period, you're right, the only inputs are sun and air, of which we have an practically unlimited supply.

Happy to be corrected, just commenting for the sake of discussion. This is how I understand the 'efficiency' of such experimental energy to be determined.