Not sure TI really had choice to stay or not in the Baseband business. They overlooked the 3G. The cost was going to be very high to play catch-up in 2008/2009.
BTW, in 2015, 75% of the mobile market will be shared between those vendors who have both the apps processor and the baseband. TI will find a market share in the remaining 25% unless they decide to come back to the Baseband market,
Integrated Mobile processor is going to be way forward, just think what all companies are doing other than TI, Nvidia with accquisition of Icera in process of making integrated solution, ST-E Nova thor already had one, Intel with accquisition of Infineon Wireless will some day integrate it in their Atom, Renesas Mobile with accquisition of Nokia wireless modem will integrate it, Broadcomm and Qualcomm already doing it. That leaves 2 players Samsung and Apple and both of them design their own application processor and does not sell it to anyone so they can continue on that path.
Problem with TI is they never had good 3G baseband and they were too ignorant to leave this market now it's extremly difficut for them, they don't have RF, 3G baseband , LTE baseband and worst no crediable baseband IP available in market. Just think 28 nm integrated baseband + Application processor is going to strom market next year from Qualcomm , i think after that in 2013 all will be ready.
Perhaps Qualcomm prefer Baseband integration because they own so much wireless IP. They can afford to throw everything but the kitchen sink into their Baseband because they don't have to pay royalties.
When LTE is more widespread perhaps a Software Defined Modem will be very computationally demanding, but once you license one it will cover all the legacy comms. Perhaps TI, Apple and others expect to do just that. On the other hand they may expect to pick up some Baseband cheap as the market consolidates.
I believe Will Strauss of Forward Concepts is simply "towing a different party line" and that while integration of baseband modems and application processors is an available solution, especially in lower-end phones, the predominant trend is the transition from lower-end phones to smartphones. Of the top 4 smartphone suppliers, only HTC appears to be committed to an integrated baseband/processor solution. Apple and Samsung are investing in their own application processors that include integration of other innovative features but not baseband modems. For the next 3 years, I think the remaining smartphone suppliers are more likely to choose application processors coupled with stand-alone baseband modems in an attempt to keep pace with Apple and Samsung on innovation. That would leave 50% of the smartphone processor sockets open for competition for companies like TI, Samsung, Qualcomm, NVIDIA, Broadcom, ST, eventually Intel, and others. I'd say execution is a bigger determining factor of who will win in this space than the baseband modem strategy.
TI isn't alone who calls baseband "a distraction." Freescale made a similar call a few years back. And I actually agree with both of them. Keeping up with the latest baseband development could suck the life out of you. But on the other hand, once you are out of the baseband business, you are pretty much telling the wolrd that you are no longer a serious mobile handset chip vendor. You need to reposition your apps processor for consumer devices other than smartphones or feature phones. I think both TI and Freescale have great apps processors. But where those chips will shine is probably in design wins other than phones.
Blog Doing Math in FPGAs Tom Burke 21 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...