GSL, page-5

I don't think the efficiencies are an issue. The point of the MAD is getting the manufacturing process and scale right, and then getting the result out in real world conditions to test the performance, stability and longevity, both of the perovskite, and the structure on which it is built (encapsulated glass, inter- and intra-panel electrical connections, panel mounting systems, inverters and transformers etc etc.)

In the meantime, efficiencies will continue to improve, and when they do, GSL should have a mature supporting ecosystem in which to deploy the improved formulations. I think I remember RC saying they planned an annual schedule of introducing improved perovskite formulations once volume manufacturing in under way.

Also, even if Perovskite has not (yet) matched Si cells, it should still have the advantage of lower manufacturing cost and lower imbedded energy, as well as better performance in low-light conditions.