Cost. We can make single junction GaAs solar cells with efficiency close to 30% (damn close to the theoretical limit for a single junction cell of ~33%)and we can make GaAs based multijunction cells that have efficiencies above 40%. It's just that those cells are so expensive that they really are only useful for niche applications (e.g. space).

Meanwhile silicon and CdTe solar cells continue to decrease in price and increase in efficiency year after year. At this point the current technology is cheap enough to be profitably used to supply power in many regions of the world. Any one who says we need a break through in solar to make it economic is wrong. At this point the solar panels themselves are less than half the cost of a solar installation.

Nearly all press releases talking about efficiencies over 50% are just talking about unpractical theoretical limits that would require cooling the solar cells to freezing temperatures or other schemes (i.e. talking about the carnot limit instead of the shockleyy-quiesser limit).