This appears to be a based on a paper from last year. http://science.sciencemag.org/content/351/6268/74

While good and interesting work, there's a lot left to do. They're using cysteine as a reductant, so not all of the energy is coming from light. Furthermore while the system uses 90% of the cysteine and 80% of blue light, it is only 2% efficient with broad spectrum (sun-like) lighting. Even that well only under dim lighting.

Edit: Here's a group that's developed a system that's 5 times as efficient and only uses CO2 and water as feedstock rather than cysteine and cadmium. http://science.sciencemag.org/content/352/6290/1210