As a baseline, the OMAP3430 consumes about the same amount of power while active as the OMAP2420, but it has a number of powersaving advantages. First, it's quite a bit faster (2-3x faster depending on the task) which means for the same task it's going to be spending much less time active and processing and a lot more time idle than the OMAP2420 (this is called Race to Idle). As idle usage consumes orders of magnitude less power than even 0.1% usage, this translates to noticeable and immediate power savings even if all other factors are equal (which they're not). Second, idle power consumption has been improved significantly. The OMAP2420 uses nearly no power while idle (fanoush managed to get about 30 days of battery life at idle) but the OMAP3430 is even better, as it uses basically no power at idle. Although the difference seems small on a larger scale, the relative difference is large and with as much time as the device will spend in idle this means big power savings. Finally, the software, too, has seen large improvements in powersaving. Both at the library and application level and at the kernel level. Fremantle on the OMAP3 is more intelligent about hardware power management and more careful in how in consumers power in the userspace than Diablo on OMAP2. Unfortunately, most of these advantages are offset by the inclusion of additional hardware (accelerometers, cameras, LED flashes, GPU—although the smaller screen represents a major reduction in power consumption) and, in particular, the cellular radio. So in the end, the N900, I suspect, is going to have a much lower power consumption at the extreme low end of usage and a slightly higher one running full blast. The end result for the average user is probably going to be marginally worse battery life, on average, than OMAP2 devices with a potential for slightly better battery life depending on usage.