So Long to the AMD PIC

The AMD Personal Internet Communicator (PIC)

I'm sorry to see that AMD has abandoned the PIC—the Personal Internet Communicator. It's an appliance computer: just add keyboard, display, and network and you have a nice Internet terminal. My first exposure to the PIC was a year ago, in the shelter for Hurricane Katrina evacuees. AMD generously donated dozens of the devices, which we used as Internet terminals. I was impressed at the ruggedness of the devices.

I was not, however, as impressed with the ruggedness of the Windows/CE operating system. It seemed like reboots were too frequent.

Which leads me to the reasons why the PIC may have failed.

The first problem is that the PIC was a closed platform. That made it hostile to open source development.The PIC had a locked down BIOS, which prevented alternate operating systems, such as Linux, from being used. It was married to Windows/CE (abbreviated "Win/CE", pronounced "wince"). Accordingly, there was never much interest in developing interesting applications for the device.

The second problem was AMD's refusal to market within the United States. They limited sales to overseas markets, probably for fear of cannibalizing PC sales. There was some talk late last year about finally trying to enter the US market through Radio Shak. Still, it's hard to understand how the PIC could be considered competitive when it's being sold at about the same price as a low-end E-Machines PC.

Either of these alone would put a damper on the product. In combination, however, they ensured that no robust community would form around the device. Compare this to the Linksys WRT-54GL router, which has become a platform for all sorts of exciting community development. So much, in fact, that when Linksys designed a cost-reduced version of the router, they still maintained production of the community hackable version.

The PIC is a really cute device, and I could have imagined all sorts of cool things people might have done with them, if they had only been given a chance.

The final problem, and perhaps really the most fatal one, is the price/performance of the device didn't measure up. The unit cost of $250 is hard to justify for a low-low-low end headless computer. It seems to me that you'd need get it down to half that, before there was significant uptake. Compare this, for instance, to the OLPC project, which is shooting for $100 price per laptop (in quantity).

In the end, it was an awfully cute device, but there were a number of problems that kept it from living up to its potential.

(There is a Slashdot thread on this topic.)