I had been working for Android OS development since it’s beginning and so I can see Android’s problem Google is trying to address by Fuchsia. Linux Kernel is absolutely great and I don’t think if “abandon Linux in Fuchsia” is the right word to say as it brings down the value of Linux Kernel. I would rather say Google has designed a brand new OS which is not based on the Linux kernel. And the reason is the design approach. Android is not a modular OS. Fuchsia is more of a modular OS straight from its bottom-most layer to bring more scalability in consumer device portfolio.

In my view, here Google’s attempt is to solve all possible bottleneck they learned with Android and needed for a modern operating system. As technology has evolved a lot on today's date. This will turn into a big chunk of texts which people will not prefer to read if I try to explain Android and Fuchsia. So rather I will go straight to the point as in your question now.

If we see target of fuchsia, it is an attempt to fit one size in all. An OS to fit in almost all size (mind that this is for only 32 bit and 64-bit machines only though I have not seen much of traces for even 32 bit but it may come with support, not a big deal if required. And current development seems target market which been already attempted by Android. I don’t think Arduino will run fuchsia ever as some people are creating fiction imagination) of the embedded devices. Devices with small memory footprint and don’t require all the capabilities of a full kernel, most of the time they just need lightweight scheduler and minimal memory manager and so on and so forth. And scale up to the device like a mobile phone, which may be the upper limit of fuchsia. Micro-kernel facilitates this approach as it is bare minimal kernel and push all the extension to libraries on an immediate higher layer. So it doesn’t take much to fit OS to fit in the different kind devices.

I have worked with a microkernel-based mobile phone OS a decade back and so I can say this is completely fair but so many researchers believe that this is first of it’s kind to play with the little kernel (from where zircon took its first step).

I don’t know where to stop writing so I would just give an analysis point here, when Google was making Android as IoT platform, they were into kind of helpless situation by the design of Android OS and couldn’t bring it down to small portfolio. Android thing was an attempt for it. So that looks a justifiable point to me to start a new design with micro-kernel. This is just with my own experience and analysis.

And at last, Google brings such innovations as surprises to keep their monotonicity to outdate other competitors though Google is not always successful which is again never a loss for them as well. These are usually parallel projects and gets silently integrated in other project.