When the comments come so fast and furiously and the opionions are so extreme I'm always a bit distraught until I have summarized the situation. Usually I just do this for myself, but I have some time this afternoon, so I'll post.
1. Many posts have raised unnecessary fears about switching from Intel processors. Apple is not abandoning Intel. Moreover, Apple is fully aware that the option to run Windows has been an important selling point for many switchers; they're not going to do anything to undermine this.
2. The concept of a chip set is not very clearly defined. In terms of its broadest meaning it's just the complete set of chips inside the computer. AI's post leaves it wide open about exactly what's being suggested. Apple is not going to use the standard Montevina chip set - that in itself shouldn't be very surprising because Apple has already made use of slightly customized versions of many Intel products. Of course, the gist of the rumor is that something more significant is going on, but as stated the rumor does not claim that Apple is rejecting all elements of Montevina. It's consistent with the rumor that Apple is merely adding some unique element(s).
3. Contrary to some suggestions, Apple has a long history of chip design. The idea that Intel (or AMD or anyone else) has a level of expertise about chip design that Apple can't match is a mistake. Intel definitely has fabrication expertise that Apple can't match, and they have CPU expertise that Apple probably shouldn't compete with, but designing speciality chips is an art form in which it is still possible for small teams of talented individuals to surpass the behemoths of the industry.
4. Given Apple's unique position as a maker of both hardware and software it makes excellent sense for them to distance themselves from the competition with unique chips. This is exactly what Dell, HP, and others cannot do alone. It would be disappointing if Apple did not take advantage of this avenue for progress. The switch to Intel is over, and it's been an overwhelming success, so it's time to push ahead.
5. It seems early for the purchase of PA Semi to bear fruit, but Apple and PA Semi have had a relationship since before the Intel switch. PA Semi was a candidate to produce the Power Book processors that IBM could not. (See this link:http://www.theregister.co.uk/2006/05/19/pasemi_apple/
) Presumably Apple would not want to adopt those processors at this date, but it's not impossible that PA Semi was already working on other technologies when Apple bought them. Of course, PA Semi may not have anything to do with this rumor.
6. With the iPod and the iPhone Apple has shown tremendous competence in assembling specialty chips from different vendors to create outstanding products (with high margins!). If there are features that Apple wants that Intel isn't providing Apple knows where to buy them, and how to assemble them.
7. This is an exciting rumor because it invites us to speculate about the new features that Apple would/could build into the Mac lineup. Some of the features may be particular to the portable lines, but others may be universal, or reserved for the Pro lines. At any rate, here's the good stuff, a list of the various roles for a new chip set. For the most part these are not mutually exclusive. My money is on points d and i.
\ta. Power/heat reduction. To my mind this is actually one of the least likely motivations for a change. Unless you were to replace the Intel CPU itself, it seems unlikely that you could significantly reduce the overall power consumption of the chip set. Furthermore, Intel is not doing so poorly on these fronts that the portable lines are suffering in any particular way. Lastly, it's unclear that even getting 20% greater battery life (an unlikely prospect at best) would be a huge selling point. That said, it may be that marginal power savings are one added bonus of various changes.
\tb. A new GPU for laptops, iMac and Mini. Intel's integrated graphics processing has been underwhelming so it's not impossible that Apple would look to replace this with something better. If there's something out there they can buy for a reasonable price, they might well do this, especially if it's one element of a broader change. However, technology in this area changes rapidly and it seems unlikely that Apple would want to take responsibility for this element of their machines. So a proprietary GPU seems unlikely. But see points d and i below.
\tc. Some form of media coprocessor. Especially something that will do H.264 encoding/decoding. Since this is supposed to be a part of Montevina already, this would be an odd addition, unless Apple has something truly outstanding to show off, which isn't impossible. But see the next point.
\td. A proprietary vector processing unit like Altivec. This is probably the most exciting and the most plausible of all the suggestions. Such a unit could contribute to graphics performance and media processing but would be available to accelerate a host of other operations as well. This would appear to make excellent sense given what is known about Snow Leopard, and it might be just the ticket for ennabling Apple's "Core" technologies to shine. All kinds of developers, especially in the sciences, would be enthusiastic to use this processor - as I recall PPC had advantages over x86 for certain scientific applications, and I think Altivec was the main reason. It's not unreasonable to imagine that this could give the Xserve a significant boost in value in certain markets. Finally, assuming Apple's own developers take advantage of this processor, this could also be an enormous advantage for all of the pro apps. Who knows, it might even give Apple the impetus to introduce a Photoshop killer.
\te. Some kind of built in solid state memory. Storing some portion of the operating system in this way might increase start times, and offer other small advantages. Perhaps I don't understand this well, but I don't see what's supposed to be so exciting about this option. Yes, solid state drives in some laptops make sense and are inevitable, but it's not a very sexy change.
\tf. Built in 3G access and/or GPS. Apple now has expertise with these features. Why not add them to Macbook Pros? In fact, though I can see the advantages, I don't think this is very likely. 3G access would require a contract with a phone company and there hasn't been any rumors of Apple discussing such deals. Moreover, most 3G networks are probably not ready to invite the sort of traffic laptops generate. Which is the same reason that I don't think we'll see anything like an iphone port that allows you to use 3G on the iphone as a modem. GPS strikes me as more likely item than 3G, but I imagine that few people who travel with a laptop don't also have a phone, and GPS really makes most sense on the phone.
\tg. A special chip for interpreting multi touch gestures. Obviously there's got to be some kind of controller for the touch pad, but that's true even for the current generation, so it's unclear that a controller chip for multitouch would merit much notice. Still, Apple could surprise with something really astonishing. It's interesting to wonder how much use Apple will make of multitouch. Moreover, rumors are suggesting that the touch pad is glass, is it possible there's going to be a small touch screen monitor under the glass? What would one display on such a monitor? Would it be useful, or distracting?
\th. A voice input chip. One poster suggested that Apple might add this. It might be cool, but it seems to me that emphasis on voice input is not in the immediate future.
\ti. Support for high resolution monitors to take advantage of Resolution Independence (RI). No one else in the thread has mentioned this, but it's got to be considered among the chief possibilities. Apple has been working on RI for a couple years and encouraging developers to be prepared. RI is a technology that should put the Mac in front of the competition for a couple of years. And it will be a powerful selling point because high resolution monitors create an immediate visible difference in the computing experience. Imagine a monitor with the crisp readability of an iphone. People don't talk much about RI, but it's coming, and it's going to be awesome. Of course, high resolution monitors won't come cheap, so this is likely to be a technology that will distinguish the Macbook Pro from the Macbook. It will also be technology that will usher in the Xmac because it will be too expensive to put into an iMac, and too nice to hook up to a Mini. I'm uncertain exactly what kind of chip(s) will be required, but it's my understanding that to take advantage of high resolution monitors at least some elements of the GPU have to be tweaked along with a new display interconnect/port.
\tj. Some form of hardware authentication. It's not clear that this is either necessary or useful. Hackers would eventually find a way around it, and it would be an expense and an effort of little obvious value. However, if Apple actually adds specialty hardware, they immediately put the clone makers out of buisness.
8. Some people seem convinced that new Macbooks are going to drop in price and account for the decrease in margin that Apple is projecting. I think this is wrong. It's much more likely that the product transition they're talking about involves the iPod Touch. I would expect the new laptops to come in at nearly identical prices to the current lineup.
So this is my summary. To repeat, I think an Altivec-style coprocessor and/or the resources for Resolution Independence would be both the most likely and the coolest features. Feel free to suggest additions or corrections.