Wrote Programmer, the wise:
1. Apple released Jaguar last August and the WWDC preview hints at Panther being released in roughly the same timeframe (Aug-Sept).
You're building the premise that Panther and the 970 are inextricably or conveniently linked. I agree that Panther appears to be a late summer, early fall release, but I don't think the 970 MUST be tied to Panther. Just my opinion.
2. IBM's original 970 announcement said 2nd half 2003 (i.e. July or later) for full production and rumours are that they schedule is going well. Accounting for the time to get the chips from the IBM fab into the Apple factory and through the distribution network, shipping machines in the Aug-Sept time frame seems reasonable.
Rumors (which are worth what is paid for them: zilch) are that the 970 is well ahead of schedule, that the IBM Fab is working out very well, and that 970s are in production.
Getting the motherboards will take longer than getting the processors, given the "info" we've heard. Rumor is the IBM is "in production" of the 970, yet the Motherboard contracts have just been awarded (Asia). getting the relatively light, small processors from NY to a California plant is a two-day FedEx ride, once ready. The assembly stages can all be worked out with the preview 970s or duds that have the same physical characteristics. Getting the motherboards made and shipped to California might take slightly longer. Educated guesses are that the processors will take about a month to fab - I have no idea, but I'll use that timeframe. The motherboards will likely take slightly longer since that contract was just awarded a couple of weeks ago. Assume 7 weeks from now, just to pick a number from the ether. That means assembly begins around the first of June. We've heard nothing about a new portable motherboard.
3. Supporting a new processor which has major differences (new scheduler in the compiler, 128-byte cache lines, new supervisor mode details, new virtual memory hardware, etc) requires at least a minor OS revision.
The very point of OS X is portability. The overlying OS is ignorant of the hardware to a great degree. The kernel handles that. I don't think the changes would be so widespread that they would have motivation to wait for Panther.
4. The 970 isn't going to be the only new feature of these machines -- they are likely to have AGP 8x and possibly other goodies not yet seen. This will also require OS support.
This is stuff they could have implemented in-house months ago and have as part of the new motherboards.
5. The more OS revisions you create, the more you have to support. If you are going to release a major OS update within a month anyhow, why put out an "extra" minor one to support new hardware that you could easily delay shipment of by up to a month (if at all).
6. We don't know if Panther will include 64-bit support. My position on this is that the OS internals are probably already 64-bit ready so why not include it at that level? The GUI libraries may not be ready for 64-bit, but that's not nearly as important as being ready to support 64-bit server applications.
7. Steve Jobs likes to make a big splash. If he has the option to do a double barreled introduction with a new super-fast OS running on a new super-fast Mac, then he probably will. These new toys may be in a demo-able state in June and WWDC has been delayed exactly so that SJ can do his thing... even if they don't actually ship until September. They'll probably start taking large numbers of pre-orders.
Apple cannot afford to delay shipping one day. The PowerMac sales are awful and Apple would lose millions and millions and millions of dollars if they withheld the release of faster computers at this point.
They wanted sales so badly that they intro'd the 17" PowerBook nearly 3 months before it shipped. I don't think they would wait for Panther if they didn't HAVE to.
If they can put out a 10.2.5 or .6 update that enables support for the 970, they will drastically increase PowerMac sales the day they start taking orders. If they hold off until Panther, they lose potential sales - sales they cannot afford to lose, I think.
Of course Panther may show off the 970 better than 10.2.x, but that's another selling point to get folks to upgrade to Panther, not a reason to delay ready products.
8. The aluminum PowerBooks were just introduced, and there is 15" unit waiting to join them as soon as the existing stock is gone. Having moved to an all new lineup I think it is very unlikely that Apple will immediately move to a completely new high-powered processor with its high powered companion chip, from a new fab at the start of its production run. These parts are going to start out relatively expensive and it is most likely that Apple's PowerMac (plus whatever IBM's plans are) will consume all of the available supply. A super-fast PowerMac will not canabilize PowerBook sales in an appreciable fashion, and so what if they do? The PowerMac has a high margin as well and people that want a portable simply aren't going to buy a desktop machine. This fall the 7457 will likely arrive and push the clock rates and battery life of the PowerBooks up beyond what can be achieved with the 970 on the 0.13 process. Once the PowerBook goes 970 it can't go back, so Apple won't take it that way until there is a clear advantage.
The 7457 is a bird in the bush. The 970 MAY be one in the hand. I too doubt that the 970 will go portable first, but it would be a helluva surprise.
Basically, I see the financial motivations being paramount here. Apple has GOT to boost high-end sales to sustain profitability, and a few hundred thousand June/July/August PowerMac sales will do wonders for Apple's third and fourth financial quarters. If that means that Panther is a paid upgrade for some of those buyers, well, Apple's done that before, right?