Marketing and business model, primarily. Previous solar tiles were more expensive than conventional solar and so are these. The difference here is that Tesla and SolarCity are marketing it as no more expensive upfront than a conventional roof replacement by keeping ownership of the roof and having buyers sign a contract committing them to buying power from it at ever-increasing prices.
There also seems to be some efficiency and aesthetic improvements over previous tiles, but those alone wouldn't be enough to make it viable.
And from that, hopefully, scale. Solar tiles have always seemed to be a very poor relation of solar panels. The approximately 1x1.5m panel is a sort of informal global standard, and that's meant that solar tiles have been much more fragmented, much lower scale, and much more expensive.
On the other hand in theory these solar integrated building materials should greatly reduce the associated costs of installing panels - labour is usually something like half the cost of installing rooftop solar, but if you can install a solar roof without much more effort than a normal roof, those extra labour costs are massively reduced.
If you can leverage this labour cost advantage to ramp up scale of the tiles, and get them close to the per watt cost of traditional panels, this is how you can really make rooftop solar competitive. These labour costs are why rooftop solar is so much more expensive than utility solar. If you can greatly reduce them, total costs are much more tied to the cost of the tiles themselves, the inverter, and so on, where there is enormous leeway for ongoing reductions. Very exciting.
There also seems to be some efficiency and aesthetic improvements over previous tiles, but those alone wouldn't be enough to make it viable.