Qualcomm is battling it out on all fronts with competition becoming more fierce. In mobile, MediaTek is becoming far more aggressive on designs and node transitions. MediaTek will be the first user of TSMC’s N4 process. Samsung is licensing RDNA graphics to beef up their internal SOCs. Google is working with Samsung on their own in-house custom Tensor chips, and Apple is chugging away on their modem goals. Despite all these challenges, Qualcomm isn’t backing down. They are committing to far more aggressive SOCs in the future even if the short term is weak. More importantly, they are accelerating gains in the RF Front End (RFFE) market. This is a nearly $20B market that Qualcomm is rapidly gaining share in. SemiAnalysis believes it will be larger than $8B revenue market for Qualcomm in 2025.
RFFE is a broad field encompassing everything from the antennas catching wireless signals all the way up until the transceiver and modem. Qualcomm has a lockdown on modem technology where they sit multiple generations ahead on implemented 5G 3GPP Release schedule, but in the RFFE business, they only hold around 20% share. Qualcomm started building out their RFFE business quite a while ago, but it has really supercharged with growth in the last few years with the transition to 5G. To bring leading radio technologies to the market requires a lot more than just implementing modems and calling it a day. It means regulatory approval in dozens of different locales, working with over a hundred telecom companies, ensuring form factors and thermals are optimized, while keeping costs in check.
RFFE is not a single technology, so classifying it as a monolithic entity to make general statements about is difficult. This AnandTech article from 2013 by Brian Klug, who now works at Apple, is a fantastic summary of the basics. Suffice to say, there are even more changes and developments since 2013. The capabilities required here are varied and difficult. Every 6 months, some part of the RF stack evolves. Qualcomm is attempting to come to market with a vertically integrated stack from modem to antenna modules which reduces the required engineering effort on the part of clients. Qualcomm does not have offerings in every part of the stack, in fact there are still many areas in RFFE unsatisfied by Qualcomm. Today's announcement is about plugging one of those holes with a leadership product.
Qorvo, Skyworks, Akoustis, and Broadcom would be considered the main competitors of Qualcomm, but there are others in the space as well. One area Qualcomm has been attempting to be competitive and exceed the competition is with RF filters. 5G is very complex partially due to all the variety of frequencies used for communication. The overlap of various bands and sources of interference makes accurately and effectively maintaining signal integrity difficult. Filters are used for rejecting everything outside a given band and to isolate frequency ranges.
One such example of this type of issue exists with GlobalStar’s n53 band (11.25MHz of spectrum 2483.5MHz to 2495MHz). We debunked the ridiculous rumor about the iPhone 13, satellite internet, and $GSAT well before the iPhone announcement.
The iPhone being supplied with a modem that supports this band is nothing special in of itself. 2.4GHz WiFi routers for decades have been able to use this band with mild modifications due to the proximity of their operating spectrum. This spectrum is rife with interference, with low quality WiFi and Bluetooth operating outside their normal bounds and microwave leakage.
The GlobalStar n53 band was considered low quality due to the overlap of heavily used spectrum. A filter’s main purpose would be to isolate the specific frequency being used and to ensure communications can be maintained over a given spectrum. Qualcomm released their ultraSAW filters last year which are used for 600MHz to 2.7GHz. One of the side effects is that GlobalStar’s n53 band became more usable.
Qualcomm is now extending their filter technology with ultraBAW. This filter operates from 2.7GHz to 7.2GHz. Qualcomm has gone from severely lacking in RF filters to having high performance filters from 600MHz all the way to 7.2GHz. This includes support for upcoming band deployments such as C-Band including the upcoming Wi-Fi 6E standards. The ultraBAW filters can support channels as wide as 300MHz on downlink, meaning Qualcomm will be able to achieve some nutty bandwidth figures in future devices even when relegated to only sub-6GHz. This innovation will need to come with others across the RFFE stack, but a big hole in Qualcomm’s offerings has been plugged. The ultraBAW filters will be shipping in volume by H2 2022. It will be available as discrete filters and as part of modules.
SemiAnalysis believes it will first appear with the mid-cycle Q3 2022 smartphone update. This is somewhat unrelated to the RFFE situation, but Qualcomm seems to be in a bit of a predicament with MediaTek becoming more aggressive on application processors. Our sources report that Qualcomm will be countering this push by releasing 2 flagship SOCs next year. First will be on Samsung’s 4nm node in Q1 followed by an update with TSMC 4nm in Q3. The architecture should be similar, but process node difference will drive discrepancies in efficiency and clock speeds.
This filter will likely help Qualcomm penetrate deeper into Wi-Fi. Broadcom isn’t as well prepared on a baseband and RFFE front for the 6 to 6E transition. Other firms generally lack either the RFFE aspect or baseband aspect. Apple will move to in-house 5G modems and Wi-Fi basebands, but ultraBAW would allow Qualcomm to possibly stay in the supply chain long after the transition. SemiAnalysis asked about the manufacturing partner(s) this product, but we were only told they would not be disclosed.
Qualcomm’s overarching goal is not to offer components for smartphone OEMs, IoT, Automotive, etc to build radio solutions. The real end goal is to offer pre-qualified, pre-approved vertical RF solutions from the modem to antenna. This extends beyond client side. They want to offer this vertical solution for small cells with leadership products such as FSM200xx. In general, Qualcomm will continue to gobble up SAM in RFFE. SemiAnalysis believes Qualcomm will move well above 20% share in RFFE in a few years. The RFFE will help stem the upcoming share losses on the application processor side and allow for meaningful earnings growth even as smartphone volumes stagnate.
This article was originally published on SemiAnalysis on October 20th 2021.
Clients and employees of SemiAnalysis may hold positions in companies referenced in this article.
But how much RFFE share can Qualcomm gain independent of bundling with the baseband/AP? With an Apple internal modem, MediaTek gaining share, and more use of internally developed APs, that would seem to put a glass ceiling on Qualcomm's RFFE share, unless it can stand alone vs. its competitors (Broadcom, Skyworks, Qorvo, etc.).
Hi Dylan, thank you for the valuable insights into the ultraSAW and ultraBAW filters. I've been following Qualcomm closely for some time. In your article, you mentioned there remains several components in RFFE that Qualcomm is still not able to provide or is lacking in leadership. Which are some of these main components, and who are the respective leaders for each of those components - Skyworks, Broadcom, Qorvo, Murata etc? Thank you very much, much appreciated!