The competition was judged by NREL, IEEE and Google. I tend to believe them.
But what I'd like to see is how it was done, and whether they came up with something that might improve the typical consumer inverter. I mean, their device uses 400VDC input. I don't have much use for that.
It looks like "wide band gap" semiconductors were a big part of the solution. May not be relevant to a typical 12VDC inverter?
Or we could wire homes and offices with a 12V DC line alongside the traditional AC wiring. Each home could have one central AC-DC converter, and you would no longer need any wall warts or onboard power supplies. Think how many AC-to-DC converters you've had to buy in your life, in appliances, TVs, and on and on, plus all those wall warts.
This is a grid-tied inverter, not to be confused with the standalone inverters that might be connected to a car, or a photovoltaic system with batteries.. Efficiencies of 95% are not uncommon. What is impressive is the size. The inverter on our house is 10 years old and has efficiency in this ballpark, but is about the size of a carryon suitcase. But it's passively cooled - no fan, just a huge finned heatsink. Even in 2006, there were actively cooled inverters no bigger than a loaf of bread. The thing is, they have a fan.
Right, no schematic, but if I had to guess at topology, I'd say it was not revolutionary. Conceptually, what they have to do is charge up a capacitor bank from the high-voltage DC of the solar, then dump the charge onto the AC line at the right moment in time. One might think that just shorting the capacitor to the line would mess up the purity of the AC, but the grid wiring is of such low impedance that they get the required low harmonic distortion figure.
Again, I'm impressed with their accomplishment, but am reminded that market constraints were removed. I doubt this is weatherproofed like the NEMA enclosure of the commercial product. Their proto might not even be safe for indoor use. Will their tiny fan last for 15 years? Will the product work at the reasonable temperature of 40 C? I notice that Google requested operation at 30 C, but they tested at 29. I'm thinking there must have been a reason for testing at 1 degree below spec. I would expect that the proto would not comply with other grid-tie requirements such as UL 1741, and also probably has no built-in telemetry, which is practically a requirement for a product nowadays.