There are about a half dozen threads on this topic.
here are a couple of links.
http://www.raspberrypi.org/phpBB3/viewt ... 24&t=13908
http://www.raspberrypi.org/phpBB3/viewt ... f=24&t=495
It just is not possible to keep a multi-core, Wireless, 2GB memory with a SATA under $100. You are describing a board that will cost more than a Panda Board ES (about $180) which is dual core, 1GB, WiFi and Bluetooth with no SATA.
The cubieboard, if it pans out, will be a real contender against the Raspberry Pi at $49 (512MB $45).
http://www.raspberrypi.org/phpBB3/viewt ... 62&t=16505
The Raspberry Pi has a specific group as its target. Those needing an inexpensive device to learn programming. It has become an everything to everybody board. That is because the price is low enough that they do not fear destroying it.
The next RasPi version will have the same goals. I do not see a WiFi version on the horizon. A multi-core version might happen but, it will be an incidental occurence, the processor package has features they want, in the right price range and just incidentally is multi-core.
The features I think they will look for in the next generation are; ( )= reason they might do it.
*ARM SoC with v7 not v6, possibly faster, most likely will end up with a 1GHz at best. (software support)
*SoC supports ethernet natively. Most likely 10/100Mb not Gigabit. (expense)(LAN9512 not necessary)
*SoC is designed to be a USB host. It will still be USB 2.0. They may look at OTG support. (current USB problems)(design of BCM2835)
*SoC supports inexpensive, newer memory packages, just like it does now. (hopefully will move up to DDR3)
*They will try to steer away from any hub chips (LAN9512) mounted on the board. (price savings in manufacture)
You could end up with a 900MHz, 512MB, one USB, one Ethernet board with extra GPIO. Essentially an upgraded A Model with B model features. No LAN9512 chip would save money and board area for other uses such as additional GPIO. It would have a chance of being supported by UBUNTU and ANDROID Operating Systems due to newer instruction sets. At least this is what I would shoot for in a new version. I agree that dropping to a single model would be easier. But would you be able to hit the $25 price point that is their main aim, for the education goals they set themselves.
I personally believe that the majority of purchases for education will be the B model. Schools will want hte Ethernet. Wireless is just not as essential as most people believe it is. Its cost would drive the new RasPi way out of the $25 price range. It would also add more testing and certification to the mix.
A likely prospect is the BCM11311, which I believe is a SoC for tablets. It has many of these features already. Currently I believe its cost is too high to meet the price point that the foundation needs. That may change! It will be a Broadcom chip at its core.
It is a big endeavor to create a new version. I think it will be a while before they thionk it is necessary to even try. I believe their experiences, in creating and producing a board for the public, will be seen in the next design whenever that will be.
If people were only using the RasPi B model, and soon the A model, for what it was intended, there would be no real significant issues left to address at this point. It programs just fine. It is all the other stuff that it is used for that, that is bringing out issues outside the iriginal design criteria.