The hardware itself is shaping up to be very stable so far. A new board revision is planned for the near future and has a couple bug fixes, numerous clean ups to remove the extra parts I placed for testing, addition of the final wifi module and changes to final connectors and headers (smaller JTAG, different expansion connector, AV connector). Product launch is likely to be delayed for the reason explained below and this gives us even more time to make sure absolutely everything is perfectly functional.
Anyone who has been following the Beagleboard progress will know that the USB phy chip has issues and is likely going to need a new rev in silicon. Any TI customer using the OMAP chip for its built in host capabilities is waiting for a solution because the OMAP uses a non-standard clocking scheme to drive the phy that only one or two companies is trying to support at this time. As usual these parts are covered by NDA for a while so I can't say too much more than is public with the Beagle.
The bad news is more of a delay, the good news is that early adopters (first 3000 and even the next batch of 3000) should feel very confident that their Pandoras work just as well as systems made a year from now since we have more time to test them. Craig has also just pointed out that if it is defective from manufacture, it will be fixed under warranty.