A few more details:

Code will run on M3/M4 arm, like OP said.

Code has an optional compiler that uses integers instead of python objects and achieves NEAR C SPEED.

Uses GC with response time of 4 ms.

Basic compilation mode is optimized for memory consumption.

You can choose which compilation mode per function using decorators.

I believe this can run with chips costing around $3 or $4 at the low end, due to memory requirements.

It has an easy to use C api.

It can run on the netduino+(which is arduino shield compatible).

There's a REPL from the PC.

I pledged at a level that gets me one board. So far I don't think they have shipped. The author has documented that the first shippers will be hand soldered, which makes good sense, and I'm looking forward to mine.

Now, just to come up with a use for it. ;-)

Heh.. same here also a backer , yet have no direct use case at the moment. Though I also own two Raspberry Pi's and a Beaglebone black that are fun to play with, I guess I just like the fact that this is just Python without the overhead of a complete Operating system.

I'd describe it as a microcontroller breakout board with a built-in Python interpreter. I've also got Pi's, and make extensive use of lower level MCU breakout boards (e.g., 16 bit, programmed in C) for prototyping. It's easier to throw a micro at a problem than to coax a PC to work as a real time computer. And there usually has to be some kind of electronic interface anyway, so why not a micro?

The Python board will do a couple of things. First, by studying the design, I'll get a gentle introduction to a micro that's much more powerful than anything I've ever dealt with. I'll see what kind of code is needed to manage such a beast. Second, I'll learn the tradeoffs between a higher level language and C for embedded projects.