TI pushes head with OMAP, calls baseband a distraction: Page 2 of 3

November 11, 2011 //By Sylvie Barak
While Texas Instruments pushes forward quickly and aggressively with its OMAP mobile processor, some still wonder whether the firm's “strategic decision” to leave out baseband could come back to bite it.
solid on the surface, however, analysts in the space have cast doubt over TI’s dismissal of baseband.

Analyst Will Strauss of Forward Concepts believes Goren is simply “towing the party line,” and that while it’s certainly true that modems and application processors are on different road maps, integration of the two is a growing trend, especially in terms of cheaper, lower-end phones.

Analyst Jim McGregor of In-Stat agreed, saying that not only would it be cheaper and less of a battery drain, but that Moore’s law actually facilitated it.

“TI’s argument doesn’t hold water, because it’s not a question of ‘if’ baseband should be integrated onto the chip, but when,” said McGregor adding, “if you’re going to be playing in the smartphone market, you need baseband.”

While it’s true that some devices don’t necessarily need cellular connectivity today, McGregor posited that as carriers began to think more in the direction of pooled data plans to connect up all of a person’s devices, that tenet would become less true.

“There will always be room for a stand-alone OMAP4 or 5, but the number of stand-alone sockets will likely not grow as fast as the integrated pair numbers,” agreed Strauss, adding that while the integration trend lay more at the low-end of the smartphone spectrum, Qualcomm's Snapdragon S4 processor with inbuilt LTE modem would more than match up to OMAP5-level capability.

“No matter how good OMAP is, there will always be competitors that are in the same "horsepower" category on the same die with the modem,” Strauss said.

Interestingly, said Strauss, the "modem being a distraction" was Intel's mantra back when the firm was pushing StrongArm apps processors with a modem acquired from DSP Communications. When this initiative failed, after a billion dollars or so worth of investment, Intel sold the pair to Marvell for $600 million - a fraction of the $1.7 billion it had acquired DSPC for.

Seeming to regret

Design category: