It all depends on the MHz. Silverthorne's targeted TDP is 0.55 Watts. If that is at 600 MHz clock rate, it'll be within 600 MHz ARM implementations. I wouldn't expect ARM processors of equivalent performance per Hz to be much lower power consumption than Intel's attempt, and as noted, Intel has a process advantage to provide parity if they are off by a factor of 2 to their competitors. If that's a 1 GHz clock rate, I don't think ARM will be competing very well as Intel's process advantage would be too much.
If those numbers are real and include the entire chip set then they are indeed very good. One has to remember that this is a high integration shot at a SOC. The chip will be including a memory interface and a GPU - plus whatever else. So at the moment I'm taking this too mean that the total power is pretty damn good especially if it is maxing out at two watts. In any event Intels power numbers are always a bit sneaky as you really want to know what the power profile is when driving a farily intensive process(es). Say watching a movie while handling Wifi in back ground.
Either way if the chip set is half as good as is being indicated, then Intel will have made significant advancements for the i86 world. The problem is ARM isn't standing still, you should be able to implement a multiprocessor ARM solution and still end up under Intels power levels and probably end up with a better more responsive system. In any event Intel gets the slot simply because of the instruction set and being good enough.
Having a competitive SoC on the other hand may prove to be trouble for Intel. They're going to have 2 chip solution (Moorestown). That may be one chip too many. They got a really nice Apple-ish concept device though:
The only thing that would be attractive to Apple is the x86 instruction set or a subset of that and Intel's low power processes. These features can't be underestimated though. Frankly I have to agree that a two chip solution will be a hard sell, especially in something along the lines of an iPod. However considering what they have integrated into the main chip, I have to wonder if the second support chip will always be required.
The problem is the manufactures just don't understand people perception of value for the money. At least not until ASUS came out with the Eee PC. The simply answer is that no body would be willing to spend the money asked for the devices. Especially in the corporate world and its mind set of screw the work and buy the cheapest thing that can get the job done.
Edit: Well, Anandtech won't show the images
As a side note I don't like the idea that Intel will get the slot simply because it is good enough and has the i86 component locked up. Alternative hardware seems too be dropping like flies. I expect Power PC to be dead in a couple of years, mainly due ot IBM dropping the ball there. Microchip is taking a shot at MIPs and then your have ARM. I fully expect that in a coule of years we will have a two horse race with Intel and ARM the only survivors. Intel will take just about anything that requires flexibility in user level software leaving the embedded world to ARM.