Broadcom is out. NVIDIA should probably get out, too. Competing with MediaTek and Qualcomm (QCOM 1.45%) in basebands is quite a challenge, indeed. The former is dominating the low end while the latter owns the high end. However, Marvell (MRVL 3.17%) could actually give MediaTek a run for its money.

China is just now beginning its transition to 4G LTE, where Chinese carriers use a different flavor: TDD-LTE. Here in the U.S., domestic carriers prefer FDD-LTE, although Sprint has been rolling out TDD-LTE in select markets since it offers greater spectral efficiency. TDD-LTE separates downlink and uplink signals by time, whereas FDD-LTE separates downlink and uplink signals by frequency. That means that FDD-LTE requires paired spectrum channels, while TDD-LTE only needs a single channel.

Transitioning to 4G LTE is easier said than done, though. Qualcomm is the clear leader in LTE basebands (both discrete and integrated), but even the mobile chip giant is experiencing a "slower-than-expected ramp of LTE sales in China," according to CEO Steve Mollenkopf. Qualcomm does have momentum with LTE design wins and expects LTE device sales in China to be back loaded toward the end of the year.

According to a recent research note from Nomura, Marvell is farther along the LTE learning curve than MediaTek, particularly with FDD-LTE. China is still in the early stages of TDD-LTE deployment, so many Chinese smartphone vendors still need FDD-LTE modems for models shipping into the U.S. and Europe. Marvell's modems were certified on AT&T's network last September, and got a stamp of approval for Verizon Wireless just last month.

Furthermore, Broadcom's public exit could also be an opportunity for Marvell to win more business at Samsung. Qualcomm will also vie for more share at Samsung, especially since its Snapdragon chips frequently power any devices that don't use Samsung Exynos processors.

Marvell is already starting to build momentum. Last quarter, CEO Dr. Sehat Sutardja noted better-than-expected demand for the company's LTE offerings. Mobile and wireless revenue jumped 30% sequentially as a result of this LTE strength, much of which came from China. This business is now a third of overall revenue.

Excluding Qualcomm and MediaTek, baseband challengers are collectively spending $3 billion in R&D chasing the LTE baseband market, which is expected to generate less than $3 billion in total revenue this year. The real question is whether Marvell can achieve sufficient scale to make its R&D investments pay off. Broadcom obviously couldn't make the economics work out favorably. Can Marvell?