I posted my comparisons on some of the perceived advantages of going with nVidia chipsets in another forum and I thought I might as well posted them here. I'm not saying that going with nVidia chipsets is unlikely or bad, it's just that it isn't some huge saving grace. And I'd thought I'd get this out before whatever comes tomorrow.
http://www.xbitlabs.com/news/mobile/print/20070907170824.html
Size is not a big factor. Intel has the GS45 available for small-form factor applications. A regular GM45 is 34x34mm while a GS45 is 27x25mm. The GS45 can operate in high-powered mode for the MB and MBP, offering the exact same functionality as the GM45, only being physically smaller. The GS45 can also operate in low-powered mode for tighter thermal requirements such as the MBA. Similarly, the ICH9M southbridge is also available in small-form factor form. The total surface-area of a regular GM45 and ICH9M is 3342mm² while the small form factor version is 1415mm² with no loss in functionality.
http://www.nvidia.com/object/hybridsli_notebook.html
Being able to switch between integrated and discrete graphics is a feature of the GM45 and Montevina. Intel's implementation has the advantage of being GPU agnostic working with nVidia or ATI GPUs, while nVidia's chipset forces you to use nVidia GPUs.
Hybrid SLI for performance boost is irrelevent since it only works for low-end GPUs 9500M G and below. A 9500M GS or above can't operate in SLI with the IGP so it's no help to the MBP. It'll only help the MB if the MB also had a low-end GPU in addition to an IGP which is unlikely.
This technically was really pioneered by Intel in their previous Santa Rosa chipset and is called Turbo Memory or Robson Technology while in development. I believe Intel has also implemented it on their desktop chipsets now too. Montevina of course supports it.
nVidia's IGP is really the only advantage they have over Intel chipsets.
http://www.notebookjournal.de/praxis/79/3
The GMA X4500MHD already performs somewhere between a discrete 8400M G and discrete 8400M GS. The 8400M GS has since been renamed a 9300M G.
http://www.vr-zone.com/articles/Intel_GM47_Mobile_Chipset_Delivers_2X_Graphics_Performance/5592.html
Intel also has a GM47 on their roadmaps which has the GMA X4500MHD overclocked from 533MHz to 640MHz. This'll probably come close to 8400M GS/9300M G performance if Intel gets their drivers together.
So nVidia's IGP solution really needs to be 9400M class to solidly differentiate itself from the GMA X4500MHD and Intel's possible overclocked response. Hopefully, nVidia will deliver and Apple helps by using dedicated VRAM.
The other issue is power consumption. nVidia's chipsets generally aren't known to be the lowest power solutions compared to ATI or Intel. Intel's Montevina has also moved to a 65nm process compared to 90nm in Santa Rosa so power consumption is down despite the faster IGP.
Really, the advantages of going nVidia over Intel aren't quite as clear cut as some of the rumours or nVidia's marketing would make it seem.