That way it doesn't matter how the solar panels are performing at any given moment,
But it does matter what the average output of the solar cells are over time, versus the average amount of charging of cars they are doing. TFA says that the station will provide slightly more energy in a year than would be required to charge cars, but they don't specify how many, or how often, and I didn't see where they talked about how big the charging station was.
What if it takes 10 m^2 or panels to run it, and can only support one car of average usage? Then it doesn't scale up as easily. You'd need a football field sized Supercharger to maintain ~50 cars.
In 2009, there were 254m passenger vehicles in the US. In 2007 there were 164k gas stations in the US. That's an average of 1548 passenger vehicles per gas station. This doesn't count commercial vehicles and semi trucks. My assumptions about the charging station up above are surely conservative, but it's clear that on-site solar generation is not going to scale up the same way as conventional gasoline stations.
Now, distributed power generation certainly is the way to go, and so it would make sense to proceed with this strategy, but I don't think it will continue to a net contributor to the grid as volume ramps up.