This is plausible, although I would be surprised if they didn't manage to get a special rate from the consortium. At the very least I would expect which ever group at Sony given sway over Blu-ray to have to front the bill for it, at least in their internal accounting. If they use an internel currency (pretty much identical in design and function of the rouble in the Soviet Union... seriously) for interdepartmental resource allocation like IBM and Microsoft -- I only mention these two because I am familiar with their schemes -- it could easily be that SCE's blueray costs are being subsidised. This wouldn't show up in Sony's fillings, but it would be considered reality at Sony. This would also explain why SCE seems to have come off pretty unscathed in Sony's company wide belt-tightening, since the cost of the PS3 to Sony as a whole wouldn't be the same as the cost to SCE.
(The most recent data I had was a $10 fee per player, but you may be right as the PS3 did esure the victory of the format, and theres also the internal currency thing to promote inter-department cooperation which may not exist explicitly but probably something similar exists at least implicitly in their internal organisation.)
Rambus would be my first guess for IP related PS3 costs being higher than the 360s as they are notorius in the industry.
(Yes I agree, Rambus can be quite naughty when it comes to charging high fees for the use of their patents. Also they have about zero love from the other memory manufacturers considering the RD-Ram/DDR patent debacle.)
True enough, but how much that is going to matter is going to be a function of how many new fab processes we are going to see before the systems end of life. The 8Gb usb flash drives on froogle seem to start at $10. Wholesale bulk prices for just the NAND chips should be at least a little bit cheaper. Not sure if 8gb would be enough to support the bare minimum on the PS3, but the point is we are a 1 or 2 shrinks away from the point at which adding minimal, functional flash storage is almost a non-cost. I expect the 360 entry level models, possibly even coinciding with the release of NATAL, to eventually go this route also.
(The problem with using flash is that cannot get away with using anything less than 30GB internal flash and they cannot just install NAND flash without a controller. If you go by typical performance the flash inside the PS3 has to conform to both the performance level of the mechanical HDD which necessitates more expensive, faster flash and the expense of a flash controller and they have to install enough of it to manage the compulsary installs which means ~25-30GB would be the minimum level. For comparisons sake you should look to SSD prices and work backwards rather than flash prices and work forwards because I don't see how you can spot the flash controller price naked.
Microsoft on the other hand can install relatively small amounts of reasonably slow flash without the controller and get away with it as all games are designed to work off a much slower transfer speed which is the average or maximum speed of the optical drive which is only 14MB/s maximum if I recall correctly. If they really got keen they could likely beef the flash storage space up past 20GB if they wanted to. The conflict of interest they face there however is eating into their profitable HDD revenues.)
The royalties may be expensive, but I highly doubt Sony forgot to guarentee their freedom to migrate to SoC as the generation progresses. Their is also no way Sony doesn't have access to the Verilog/VHDL code, although there should be no doubt that everyone who walks within 10 feet of it must sign 10,000 NDAs in blood.For a more complicated PCB as they have dual 128 bit buses so they will likely need a larger board and or more layers long term than the Xbox 360.
(You should know this better than me, however the transfer properties of the two faster ram types are different from the XD ram that the PS3 currently uses. I don't think the explicit DMA model of the Cell would tolerate ram which had a higher latency and both DDR3 and GDDR5 aren't designed for strict low latency.
The other issue is that Nvidia and IBM are now likely competitors in the HPC space. I wonder how willingly the villian Nvidia will let the hero IBM examine its designs and the opposite is true with the design of the Cell. If the Cell and G300 aren't direct competitors already theres probably a design in the pipeline for an upcoming generation which would make them competitors.)
I disagree here, I think the opposite is likely in the limit for the Blu-ray player in the PS3 and the dvd player in the 360. The later must have a motor capable of spining the disk at a much higher velocity than the former. The new diode in the blu-ray drive is a solid-state device. At least in recent history, Solid-state devices tend to fall in costs on the timescale of months, whereas advances and cost reductions of mechanical devices tend to happen over decades.
(Hmm, high speed DVD motors will be around for a while as I don't see DVD support being dropped from computers for a long long time, but on the other hand I would say that the Blu Ray drive would likely still cost a little more but that will be negligable due to needing both a blue and red diode)
I think the IP Sony sold to toshiba was for the next gen cell and the actual fabs which they built back in the halcyon days of 90nm when anyone could have a fab in their garage :P. Even if they didn't, I doubt they wouldn't have negotiated a clause to use it with the PS3.
There are only three groups with 45nm fabs: Intel, TSMC, and a consortium of everyone else led by IBM which drefrays the costs of going 45nm all it's members. TSMC had to drop high-k metal gate tech from their initial plans without which 45nm is of marginal value and I'm not sure if Chipzilla even lets 3rd parties use their 45nm fabs as their fab lead is main advantage of their chips. So that leaves only IBM's superfriends group as a potential fab for both Sony and MS.
(TSMCs 40NM process is a world of hurt. Check Newegg for the HD 4770 which was technically released months ago! The recent reports I hear are still pegging yields at ~30% which is rediculous considering AMD is using the same fine grain redundancy they used with the RV770 and managed to get exceedingly good yields from the 55nm half node process. Maybe TSMC needs to become an IBM superfriend too! 
To be honest, im half expecting a switch to Globalfoundries for Microsofts Xbox 360 chips this year)
lol, that's right squilliam go right for the ego, you'll have me back in redmond's camp in no time :P (seriously though, thanks for the compliment)
(hehe, you have the huge advantage in that you both work in the industry and you can ask someone and they will actually answer your questions for things you don't know. I have to piece together hearsay and opinion! )
Tease.
