Too Little, Too Late?
In January 2009, Freescale launched the i.MX515, a memorably named ARM chip running at 1 GHz and designed for small computers. It's expected to show up in a lot of machines over the next few months, aiming at around a $199 price point, and a few other systems are starting to be built around the OMAP3530. Does this mean that the Pandora console, already late to market, will be overshadowed?
When pondering this question, I was reminded of something Alan Kay said when he was asked about companies taking the One Laptop Per Child designs and producing their own versions. As a nonprofit, he pointed out, it didn't matter to them if a company tried to compete with them. If a company made a better, cheaper version of their laptop, more systems would end up in the hands of children, and they'd count it as a success.
The goal of the Pandora project was never to take over the world. The first production run was only 3,000 unitsa number that barely registers as a fraction of the total market for portable computers, handheld systems, or game consoles. The first batch sold out within eight hours of starting to accept orders, however, which counts as a success by pretty much any metric.
The success of projects like Pandora has helped ARM-licensees sell their chips in market segments that traditionally would have been associated with x86 systems. At the cheap end of the market, the cost of a Windows license can add over 20% to the final sale price, and the cost differential between an Atom system and an ARM system-on-chip can be another 20%. If you're not going to include Windows, then building an ARM system can be a lot more cost effective. This year is likely to be very interesting, as buyers decide whether the ability to run Windows software (outside of emulation) is worth the increasing premium that x86-based Windows machines cost over ARM systems.