I work in a related field. Non-invasive blood glucose has been a holy grail for decades. It has bankrupted multiple start-ups. It would be exciting to see something that finally works, but my optimism is tempered.
They also know how to turn very complicated things into miniaturized production products.
If they could provide enough proof that it works the way they say it does, I bet they could find a really good suitor there. And you know the various other health/smart watch companies would love to get one up on Apple, outside of the obvious direct benefit it would provide their customers.
The challenge is, there's no a priori proof that a method can't work, because the information you need to make that assertion has to come from the same kind of research as trying to make it work. So far the start-ups have all failed in the same way, which is that a signal that looks promising in a test tube can't be reliably distinguished from the myriad sources of variation in the living system.
Note that I'm not in any way discouraging the work, just offering some historical context for the problem space.
I'll take a closer look at the paper in Nature.
The paper is way overhyped. They've just built a meta material near field antenna. Nothing new I can see.
There are other papers available via Googling that glucose can be measured at around 4.2GHz. Seems hard to do it reliably, though.