Originally Posted by Fake_William_Shatner
I can't see how it would take 5 years and 1,000 engineers. I assume the technology is already known. In order for the analyst to know what the requirements are, they would have to know Apple's intent.
For instance, Apple could use some IBM self-programming chips that change to morph into various Basebands -- allowing one phone to work in many different locations. The same tech has been used to create self-adapting antennas because one design does not work best for every frequency. The antenna breaks and adds connections in a pattern that is best designed for a specific signal -- so one antenna can be optimized for different carriers. I'm not absolutely sure this is in use -- but I was reading about the tech about 10 years ago.
My thought on Apple's goal here;
There's probably a lot of legacy cruft in the Baseband chips and a lot of functions that could be optimized that work for most cell phone systems -- reducing energy use and improving performance. Then Apple would have "images" of baseband configs that are connected or disconnected (via software PROM -- as is their style).
I'm not a Basband pro, but I suspect that most of the tech is FRAND now or not too expensive to license -- or else Apple wouldn't be trying to do this.
I'm betting Apple did their homework and the analyst has done less homework -- as most tech analysts in the media are about as accurate as monkeys with darts.
Yeah, I can't imagine 5 years and 1000 engineers too. Simplistic math: 1000 eng x $200000 per eng x 5 year = $1b investment. Maybe if they are starting zero, but they obviously are not, and they are hiring people who have done it before, and there is IP that can be licensed. So turnaround time could be 3 years with a lot less engineers, and something on the order of $200m instead.
And how does one quantify the difficulty level of design and implementing a CPU, a GPU, a SoC, and a baseband chip anyways? If you have the right people, it could be remarkably easy and cost effective. If you have the wrong people, it'll fail.
Apple does have a lot of economic and design reasons to do it, and it's inevitable that they'd do it when we say they designed their own SoC, and especially obvious when they designed their own CPU. The GPU is probably next, presuming that design a better one than the contemporary ImgTec one. After that, the radio chipset.
Currently, Apple has a Qualcomm chipset residing on their iOS device PCBs. It's not small. It takes up about as much space as the SoC and the storage, and it costs money. If they can implement one on the SoC, it saves them room on the PCB, saves them on the cost of the device from component costs to engineering/integration work, and they have a chance to make it better than an off-the-shelf one. So, lots of good reasons for them to do this.
Lastly, I think they are already doing something like this for their wristband project. If there ever was a device that needed an integrated and super low power WiFi and Bluetooth chipset/baseband into an SoC, it'll be this wristband thing. They've likely been working on it for a 2 or 3 years now, and it is the thing that Bob Mansfield has been heading up for awhile.