Battery electrochemistry still has headroom, but solar is already approaching the Shockley–Queisser ceiling, which limits efficiency at about 30% for single junction and 50% for multiple junction devices. That's the absolute conversion limit, similar to the Carnot limit for thermal engines.

So if you can buy today cells that are 18-20% efficient, you should not expect dramatic improvements in the next decades, only very gradual. The cost will continue to drop, so given the huge areas available for solar, the potential is still huge.