Very informative.
There are several hardware weaknesses in this wallet that can enable the remote attacker to do anything he wants on the device.
1. GSM chip (GE866) can run python scripts, receive over-the-air commands and firmware updates.
The detailed documentation is available on the manufacture site.
For the price of 45$ anyone can buy it and practice hacking before moving to the actual wallets. Then all devices are compromised - after they have been delivered.
Actually, what prevents someone in the manufacturing and delivery chain to change the firmware and deliver "Trojan horse" to the end customer?
How do we know it has not been done already in the first shipped batch?
2. Crypto controller receives firmware updates over-the-air by GSM chip.
What prevents a remote attacker from uploading other firmware directly to the device? Or from the local cache in the GSM modem?
3. Third result in google:
http://freescale-crack.blogspot.co.il/2014/03/stm32f437-code-extraction.html Ready reverse for the specific ARM controller used in this device, just like for many other parts.
The obscurity of the code is no protection.
(There was a promise a few months ago to release the source code for the device. Is it available for review?)
The motivation is high and the attack development price is low with ready solutions, all in SW and with remote access.
These are just a few vulnerabilities from looking at the teardown pics.
The claim for no single point of failure is accurate - there are multiple points of failure.
In this case, ease-of-use focus probably compromised basic function of the hardware wallet - ultimate security of the private key.