Description

Idea: Long Qorvo (“QRVO” or the “Company”), a leading smartphone RF semi supplier, at trough earnings & multiple which should see a) shipments rebound next year from a China lockdown-driven low and b) secular content gains over the next five years from increasing 5G penetration, which requires more sophisticated RF modules. We are 65% above CY2023 Street Adj. EBIT in our base case and see 68% upside.

Company Overview:

Formed in January 2015 from the merger of RFMD and TriQuint, Qorvo is a leading manufacturer of RF solutions for mobile, infrastructure, and defense applications. Qorvo’s main business is RF front-end module manufacturing for smartphones. These modules enable smartphones to interpret RF signals. This makes up most of their Mobile Products segment, which did $3.5bn of sales in FY22 (3/22), ~50% gross margin, and 36% EBIT margin ($1.3bn of EBIT, 83% of total). The Company has a smaller infrastructure and defense segment (“IDP”) that is well positioned but not our area of focus given its size ($0.3bn of EBIT in FY22; 24% op margins and 17% of total EBIT). 

Industry Overview: 

We estimate the mobile RF front-end market at $15bn in 2022, growing to $18bn by 2025 (6% CAGR), and $20bn in 2028 (5% CAGR). We believe Qualcomm has roughly 27% market share (80% of this is from transceivers, which Qorvo does not make; excluding this, their share is ~5%), Qorvo has roughly 22%, Skyworks has 20%, Broadcom has ~15%, and the remaining 16% is split between Murata and other Asia-based component suppliers. 

Current setup – why is Qorvo cheap? 

Qorvo saw appreciable mobile revenue growth in 2020 and 2021, of +14% and +34% respectively, but the Street now estimates total sales to be down 10% in 2022 and a further -14% in 2023. Street rev estimates for the entire business for FY24 (ending March 2024) peaked at ~$5.6bn in Jan of 2022 and now sit at $3.8bn (or a 32% reset). 

The main culprit of the y/y decline in revenue for 2022 was China-based smartphone OEM weakness, who have seen rapidly declining shipments YTD. The Street has focused on this short-term unit trend.  

Consensus View:

8 buy / 17 hold / 2 sell. The Street broadly models a normal or below normal multiple on forward earnings, which we believe to be trough. BAML is one sell rated shop; this is their reasoning from early August (when the stock was at $100): “We reiterate our Underperform rating, cutting CY22/23E sales by 5%/4%, EPS by 12%/18% and wait for increased visibility into reacceleration of mobile sales. We take our PO to $90 (from $95), based on 12x CY23E EPS (versus 10x prior), roughly applying QRVO’s long-term P/E on what could be trough EPS levels.”

We believe the Market has grouped Qorvo into the broader group of semi peers, many of which are still over-earning due to COVID pull fwd, so it is applying a below normal multiple to already depressed fwd estimates. 

Variant Perception / Thesis: 

1. RF front-end technology is enjoying 5G upgrade-driven content growth, which should last for years to come. 

2. Qorvo is well positioned to meet the demands of this shift. 

3. The Market is currently valuing Qorvo’s business with a trough multiple on trough earnings as it has myopically focused on a macro-driven slowdown in smartphone shipments and misunderstands customer inventory dynamics. 

Thesis Point 1) RF front-end technology is enjoying 5G upgrade-driven content growth, which should last for years to come. 

5G launched commercially in late 2019 and has grown to roughly 50% of new smartphones sold today. This shift has driven significant content expansion for the RF front end industry. 

What is RF Front End Technology? 

The RF front-end (“RFFE”) module sits between the antenna (sends and receives signal from the tower) and transceiver/baseband (interprets/processes incoming signal). 

Without the RFFE, smartphones simply would not be able to connect to wireless networks; the antenna would not be able to communicate with the baseband as incoming RF signals would be weak and unrefined. Think of the RFFE as a massive funnel – the antenna lets in a flow of all incoming signals and the RFFE modules filter and amplify specific signals such that the transceiver/baseband can handle processing the targeted frequency. 

Qorvo makes everything in the above schematic besides the transceiver/baseband and the antenna (Qualcomm, Mediatek, and others make these). The boxed components in the aforementioned graphic are mostly packaged together into integrated “front end modules” which are designed to handle a specific set of frequencies (e.g. a low-band module, mid-band, and high-band module may all be included in a smartphone – each with filters & power amplifiers packaged together). To briefly explain some of these specific components: 

• Filters

  • Filters remove unwanted frequency from a signal to preserve the desired component.  They make up ~50% of the value of the RF front end. 

  • E.g., communicating using 5G C-Band spectrum, one has three separate frequency bands – each of which would need to filter out lower & higher frequency noise from the signal.

    • 

• Power amplifiers

  • Converts a low power signal into a high power one, i.e., amplification. 

• Antenna switches 

  • Directs the antenna to various nodes for transmission / reception: 

    • 

• Antenna tuner

  • Most prevalent in phones with metal cases; helps the antenna function. 

• Envelope tracker

  • Increases efficiency of the power amplifier (without one power drains faster). 

Filters are the most important as they make up ~50% of the total RFFE value and are required to isolate specific frequencies. As more frequencies (termed by the industry as frequency “bands”) are supported, the number of filters increases. This is where the content story for RFFE comes in. 

Increasing RF content in 5G vs. 4G

Unlike many other semi components within a smartphone, where increasing capacity comes with reliable annual cost-downs due to Moore’s law (e.g., there is more DRAM capacity through smartphone generations, but relatively muted total DRAM dollar content as the cost per GB drops), RF is unique in that major tech jumps require additional modules, not just upgraded versions. This is simply since RF technology is based on receiving and transmitting various frequencies. The number of frequencies utilized has only gone up over time – to support more mobile data capacity. That is – as 4G came around, smartphones needed to support both past generations (3G and prior) and then add on 4G functionality. The same happens in 5G. 5G phones are simply able to receive and transmit in more frequencies than 4G phones.

This may sound overly simplistic – but the takeaway is very important. To support various frequency bands, the number of filters used in the RF front end needs to increase. Filters isolate specific bands, based on the physical properties of the component itself (see Appendix 1 for the technical details). To improve the frequencies supported by a RFFE module – one needs to add more filters (which, almost by nature, cannot filter out multiple different frequencies). Over time, more filters have been needed in each of the low, mid, and high frequency band areas:

5G continues this trend; it is additive through both new modules (solid blue boxes) and increasing complexity (blue and gray boxes) while 4G RF content remains:

Global 4G smartphones, to handle the full range of frequency bands, require ~40 filters. This moves to ~70 for 5G, or a 75% increase (see Appendix 2 for past jumps). To put some context to this, below we show the frequency bands supported by the iPhone 8 (2017 era 4G iPhone) and iPhone 14 (latest 5G iPhone). The number of bands supported went from 39 to 68. The addition of 5G added 28 new bands, and this did not reduce prior gen bands (prior gen bands actually increased from 39 to 40): 

Additionally, filters to support 5G are generally more complex. There are different types of filters commonly used for mobile RF:

• Surface acoustic wave (SAW)

• Temperature compensated surface acoustic wave (TC-SAW)

• Bulk acoustic wave (BAW) [similar in style is the film bulk acoustic resonator – FBAR]

SAW filters are the least expensive, followed by TC-SAW (same core design as SAW but manufactured in a way that makes the filter less sensitive to changes in temperature), BAW (and variant FBAR). BAW works better for frequencies above 1.5 GHz (performance of SAW drops off significantly above 1.5 GHz), justifying the price premium (Appendix 3). 

The logic easily follows from here that the shift to 5G smartphones will increase the number & performance of filters in a smartphone. This, in turn, leads to more dollar content, which means RFFE suppliers will see their markets expand nicely as the smartphone market mix-shifts to 5G. Qorvo has indicated it garners 40-50% more dollar content in 5G (sub-6GHz) than 4G. Our VAR has supported this directionally.

While not widely utilized today outside of the U.S., we believe 5G mmWave would represent an opportunity for incremental RF front-end content for Qorvo. We have heard this may be another ~20% uplift, but do not model this in our numbers, given uncertainty on adoption of mmWave and Qorvo’s exact solutions (most countries have only deployed sub-6 GHz 5G to-date). 

Comparison of the total RF content for high end phones across cellular technology and RF % of phone BOM: 

For these reasons, we are fairly confident about the content uplift for RFFE in 5G smartphones. The question then becomes, will 5G continue to expand? 

Why are we confident 5G will continue to expand? 

5G adoption is being pushed by both the smartphone OEMs and telcos. OEMs want to sell 5G as a way to spur adoption and replacements (5G is a better experience for the user, it is faster and lower latency). Telcos will increasingly need 5G frequency bands to support the demands of their network and want to offer high quality (speed & capacity) service to subscribers. 

We believe the 5G mix of new smartphones sold is already nearly 100% for mid & high-end smartphones (>$300). This segment of the market makes up roughly 52% of the total and corresponds to our estimate of total 5G penetration (of new sales) in CY22 being ~50%. It is just a matter of 5G tech migrating down to lower end phones that will drive the growth to 90%+ over the next five years or so (~10% of smartphones are less than $100, which may not be applicable for 5G tech). 

One of the largest gating factors to 5G tech migrating to lower end phones rests on the baseband/processor side of things (industry refers to this as the 5G System on Chip “SoC” as the baseband and CPU are commonly integrated) and is not predicated on price-downs for RF front end modules. It is just a matter of time for the major SoC suppliers to create a low end / lower priced offering. Mediatek (Mediatek and Qualcomm are dominant in the smartphone processor/baseband market) said it themselves in April of this year: 

“We do have -- as we said earlier and many times, we are -- we have a complete portfolio for 5G SoC which covers from -- all the way from flagship high end, mid end, entry, and we are going to have also, I would say, well, low -- entry level SoC, as we believe the penetration of the 5G into the smartphone will continue to increase across the different markets, different regions, different segments or products, and MediaTek is there and will be there to support all those different events. So we are working on that and the -- we will announce that in right time.” – Mediatek CEO

The 5G mix of sales is already widespread in countries that invested first in infrastructure. We believe 5G is already 73% of new smartphones sold in North America (2Q22 according to IDC) and ~82% in China. Below is how we are modeling 5G adoption (mix of new units sold, not install base) vs. the 4G adoption curve: 

Putting it all together, we see a 7% CAGR in the RF front end TAM from 2022 to 2024 (our target year for valuation). Going out to 2028, we see a 5% CAGR. This is driven by only modest smartphone unit growth assumptions (all years below 2019, recovering from the 2022 low point) and continued content expansion as the 5G mix increases as a fraction of the total:

We believe Qorvo will be well positioned to capture this demand and grow nicely through the macro noise of smartphone shipments.

Thesis Point 2) Qorvo is well positioned to meet the demands of this shift. 

RFFE module making is largely a four-player market: Qorvo & Skyworks are nearly pure play vendors while Qualcomm & Broadcom have broad portfolios across semi/software and only play in certain areas. 

Unlike high-end processors manufactured under leading-edge design on silicon substrates, RF technologies generally encompass different materials & niche processes. In mobile applications, surface (SAW) & bulk acoustic wave (BAW) tech are used for filters, silicon on insulator (SOI) for antenna switching, gallium arsenide (GaAs) for power amplifiers; the back end packaging is difficult and leading RF front end manufacturers control their own test & assembly facilities. To illustrate this complexity, on just one RFFE module (low band frequency LHS and mid/high RHS below), four different types of manufacturing processes are used:

As such, creating RF chips requires expertise in niche semi material manufacturing processes. RF devices are usually highly technically differentiated from their manufacturing process / material selection / packaging – rather than through miniaturization of the chip (like memory, logic, etc.). 

Qorvo has three decades of experience working with these technologies and has internal manufacturing capabilities for filters and power amplifiers. Its formation has been years in the making and originated when TriQuint acquired TFR Technologies, a tech pioneer in BAW filters, all the way back in 2005. 

We believe Qorvo is specifically well positioned to capture future growth given its IP surrounding 1) BAW filters and 2) integrated module packaging. 

BAW filters

BAW filters, specifically, are difficult to manufacture. They incorporate more process steps and a different manufacturing technique entirely than SAW filters. 

At high frequencies, BAW filters become harder and harder to get right, due to basic physics. BAW filters work by taking an incoming electric signal in and then mechanically resonate at the desired frequency (a thin crystal layer on top of the silicon wafer creates this mechanism). Higher frequencies require a thinner crystal layer – much like a tuning fork shrinks for higher frequencies. As such, manufacturing high frequency BAW filters is very complex. For years only Broadcom an Qorvo could do this:

Skyworks and Qualcomm have been investing in BAW technology over time and they can manufacture BAW filters today, though we believe they are not quite to the quality of a Qorvo or Broadcom (which is considered the best). 

Integrated Modules

Given the increasing complexity of RF technology needed in 4G and 5G phones, it is almost required for smartphone OEMs to use integrated modules. Integrated modules combine filters, switches, and power amplifiers. 

A basic schematic looks something like this:

That looks complicated and yet this is only for one frequency band! Without modules, complexity would be unwieldly in the compact designs of smartphones (see Appendix 4 for a more complex module). 

Importantly, there are only few leaders in the RF space that have individual component manufacturing capability (ability to make high quality components), design expertise (know-how to combine components into an integrated design), and packaging capabilities (ability to package together the reference designs). That is to say – there are only a few major RF module producers in the world – they include Qorvo, Skyworks, Broadcom, and Qualcomm. 

The landscape for manufacturing just discrete RF components (i.e., a single filter, a single power amplifier) is larger, but these players do not effectively have the capability to compete in 4G/5G applications (due to the widespread use of integrated modules), which has consolidated the market meaningfully over the years.

Amongst the four major module makers, there are a few nuances that we believe diminishes the level of competition:

• Technology & market focus is different across players

  • Market focus

    • Broadcom and Qualcomm mostly focus on the premium smartphone market. Skyworks and Qorvo try to play everywhere. 

      • Qualcomm: 

        • “We've been focused on premium and high, and that's the value share of the Android market and technology investments, our differentiation and even Snapdragon brand matters…  and we continue to see Snapdragon as probably the synonymous with premium and high Android flagship” – Qualcomm 2Q22 earnings call

  • Tech focus

    • Skyworks and Qualcomm are behind in the development of BAW filters, which are instrumental for higher frequency filters. As such, Skyworks is often given the low band module in smartphones while Qorvo or Broadcom are given the mid / high band module. Said another way, Skyworks is allowed to participate in OEM designs still and is often spec’d in to level the playing field deliberately; they are not missing out on a full “BOM package” due to low quality BAW filters. 

    • Broadcom, on the other hand, does not manufacture SAW filters. It resultantly focuses on mid/high band frequency solutions. 

• There are high switching costs within phone models / OEMs

• Smartphone OEMs generally do not want a dominant supplier, so they stretch share across players

• Manufacturing footprint 

  • RF front end modules are built on relatively niche manufacturing platforms. The four integrated module makers have internal fab and packaging capacity for these needs. As such, wild swings in market share are just unlikely – this is not a fab-lite designer that can grab a ton of share by having a way better design and getting an outsourced foundry to quickly ramp capacity or cutting prices rapidly. 

  • Within this manufacturing footprint, given the unique attributes of each process technology, there is limited interchangeability. To elaborate, BAW filters and SAW filters are manufactured using different processes – the former generally on 8-inch wafers and the latter on 6 inch (as SAW filters use lithium tantalate which are not widely available in 8 inch configurations). What this means is – someone like Skyworks who has focused on low frequency bands historically (due to its lack of quality BAW filters) would not be able to quickly take share in mid/high band (it would need to build a fab to do this). 

As a result, we believe the RF module makers generally maintain pricing discipline and do not make aggressive market share pushes. The last five years should provide evidence that the leading companies, in fact, have remained disciplined in terms of pricing. Gross margins of the two nearly pure play RF manufacturers – Skyworks and Qorvo – have been extremely stable over time (barring some utilization differences at times). 

Thesis Point 3) The market is currently valuing Qorvo’s business with a trough multiple on trough earnings as it has myopically focused on a macro-driven slowdown in smartphone shipments and misunderstands customer inventory dynamics. 

After Qorvo’s Mobile Products segment grew revenue by 14% in 2020 and 34% in 2021, the Market expects this segment to decline by ~16% in 2022. This is driven by a) weak smartphone demand in 2022 and b) channel destock. 

1. Weak smartphone demand in 2022

The Street and semi peers estimate global smartphone units will be down ~10% y/y in 2022 to 1.22bn (135mm unit decline), after growing 5.7% y/y in 2021 and -6.5% in 2020. While 2021 was a good year in terms of y/y growth it was still 1% below 2019; smartphones did not exactly get a COVID pull-forward / bump like most other semi end-markets (PCs and tablets did, for more obvious reasons). Esitmates have progresivel come down throughout 2022 – IDC was estimating units down ~6% as of 2Q22. 

Of the ~135mm unit decline in 2022, we believe China demand is driving 53mm of it (39%), Eastern Europe (incl. Russia) is driving 30mm (22%), and 52mm is RoW. (39%). 

China demand

China smartphone demand was up 1% in 2021 y/y (IDC), though choppy on a % growth from 2019 basis by quarter. While demand seemed to be improving into the end of 2021, giving OEMs hope of solid y/y growth in 2022, parts of China (namely Shenzhen and Shanghai) entered COVID-related lockdown again in March of 2022, leading to factory shut-downs (limiting smartphone production) and a broader damper on demand. This got worse into 2Q22, as in April, Shanghai expanded the lockdown across the entire city (25mm population) wherein many neighborhoods were either in a “control” lockdown (they could only collect essential supplies in the neighborhood) or “prevention” lockdown (only travel within a given street). In May, multiple areas in Shanghai were put into a “silent” period – which banned all entry and exit – including delivery shipments! 

These conditions led to a physical restraint (production and demand issues) but also cascaded into plummeting consumer confidence in China broadly:

Along with this, smartphone demand plummeted, with 1Q22 down 14% y/y (11% below 1Q19), 2Q22 down 15% y/y, with expectations our of -15% in 3Q22 and -20% in 4Q22. Here is what this looks relative to 2019 (see Appendix 5 for shipments by month): 

In total, we expect shipments to be down 16% y/y in China, to ~277mm units. China has not shipped this low of a figure since 2012 – when penetration was just 22% (and nearly doubled the next year) and the install base was just ~300mm. Building up to “underlying” demand from an assumed replacement cycle length (2016 base of 35 months which we grow 3% per year), we believe demand would have been closer to 358mm in CY22 (29% above the actual shipments, which were held back due to lockdowns/macro):

We expect shipments to improve in 2023, and while it is not directly a driver of our QRVO model, assume China demand grows 3% in 2023 and 3% in 2024 (largely in-line with sell side global smartphone models we have reviewed). Our conversations with China-based smartphone OEMs supports this view. 

3% growth in 2023 and 2024 would place China demand at just 294mm units in 2024, or down 20% from 2019! We estimate an install base of 1.2bn smartphones in 2024 and ~284mm replacement demand units, or an implied 51-month (4.25 year) replacement cycle. 4.25 years is an elongated replacement cycle (and we believe conservative) – it was just 3.3 years in 2019 for China (based on our estimate) and it is around 3.4 years in developed markets today. 

Said another way, we are at a massive trough in China smartphone demand today. The sheer size of the install base and conservative assumptions around the replacement cycle would suggest underlying demand is much greater than is currently being realized. 

Eastern Europe

Additionally, the war Russia brought upon Ukraine has severely diminished smartphone demand in Eastern Europe (incl. Russia). After being up 5% y/y in 2021, shipments in Eastern Europe declined 20% y/y in 1Q22 (would include just ~one month post Russia invasion) and went on to decline 32% in 2Q22. We estimate 2022 full year shipments to be down roughly 34% y/y, corresponding with roughly 30mm units of destroyed demand (2-3% of global smartphone demand). We do not expect these units to come back in the near-term but 2023 should look like the run-rate of demand and therefore not drive much incremental demand destruction into 2023 (with potential upside into 2024). 

2. Channel destock in China

In addition to the lockdown and war-driven sell-through demand destruction, for most of 2022 and into 1Q23, Qorvo is under-shipping to this sell-through demand as its Chinese OEM customers work down excess inventories. Chinese OEMs have excess RF component inventories as they placed orders in late 2021 thinking 2022 was going to be a nice year of growth, which of course was disrupted both by domestic issues and international issues (Chinese OEMs have a large presence in Eastern Europe). 

This is how Qorvo’s CFO described the situation on the C2Q22 earnings call in early August:

But I think given the nature of the events that are impacting that channel inventory, being macro and out of our control, it's hard for us to pin it down precisely. But given that backdrop with COVID lockdowns in China, the war in Ukraine, high inflation and all the other global macroeconomic challenges, obviously, we do see the inventories higher than normal, especially within the Android ecosystem. And most of the companies that we're selling to there were planning ordering and producing for much higher growth than the industry is currently experiencing. So it will take some time to bleed down that inventory, which is, again, why we're bringing factory utilization down to respond throughout our fiscal year, but we're not putting an exact time frame on it at this point.” 

Combined with sell-through being down in CY22, this channel destock dynamic has resulted in massive y/y declines for Qorvo’s business. Their revenue from China was down 45% y/y in C1Q22 and another 47% y/y in C2Q22. China was 38% of total QRVO revenue in CY21 (and 36% in the prior two years). See below for longer-term historicals and fwd estimates based on loose guidance: 

Sell-side largely has no idea how to model this impact and the Company’s disclosure does itself no favors. For starters, Qorvo does not disclose revenue by customer quarterly and only discloses 10% customers annually, of which it has two: Apple (43% of Mobile Products revenue) and Samsung (14% of sales). One must make estimates on quarterly builds and content dollar per device for each Apple and Samsung to back into “Other Android” revenue, of which only two of the ten sell-side models we reviewed attempt to do. Within “Other Android,” we estimate ~90% of this comes from China-based OEMs. This is reported as “China revenue” by Qorvo, as their definition is based on the headquarters of the OEM. 

To a casual observer, this dynamic is likely misunderstood, as one needs to piece together bits and pieces of disclosure from calls, 10-K’s, and 10-Q’s, to assess exposure by customer (and it is nearly impossible to garner precise exposure by actual end-market). Most analysts just hard-code revenue growth assumptions; this brute force method misses the nuance of China vs. non-China customers (i.e., channel de-stock vs. non-destock dynamics) and the nuance of 5G vs. 4G content and this mix-shift driver.

Outside of Apple (HQ in the U.S.), Samsung (HQ in South Korea), and Lenovo (HQ in Hong Kong – unclear how Qorvo treated this for reporting purposes historically), nearly all of the major remaining smartphone OEMs in the world are headquartered in China. The top 14 smartphone brands made up 95% of total shipments in 2Q22 according to IDC. We believe 52% of total shipments came from OEMs headquartered in China, or roughly 147mm units in 2Q22. For clarity, of this 147mm, we believe ~63mm units (or ~22% of global units) shipped to end customers actually within China, with the remaining 84mm units shipped elsewhere (no presence in the U.S, large presence in India, eastern Europe, parts of W. Europe, and Latam). 

There are only four Chinese OEMs that matter given their size: 1) Xiaomi (14% of the global market), 2) Oppo (8.6%), 3) Vivo (8.6%), and 4) Honor/Huawei (7.1% but growing). Collectively these three groups make up 38% of the global market (compared to ~52% for China OEMs in total):

This is to all point out that really only four OEMs are driving the China revenue line for Qorvo. None of them are >10% revenue contributors for Qorvo, but in speaking with Xiaomi and Oppo procurement contacts, we believe we are able to get a good gauge on the current situation (whereas, again, sell side is mostly just hard coding total revenue growth assumptions). 

Our view on how this shakes out 

The way we have attempted to build to “underlying” (meaning normalized, excl. any inventory noise) revenue for Qorvo is by doing a bottoms up build of shipments to China OEMs (actual historicals and then our estimates for the go forward based on previously discussed regional growth assumptions) and estimated content $ by RF technology. Actual shipments times our estimate of $ content would therefore be the revenue based on “sell out” of smartphone shipments. 

To get to our estimate on content $, we assumed that 2019 was a normal year with respect to channel sell-in/out dynamics, took the reported revenue from China-based OEMs by Qorvo, and divided by our estimate of total Chinese OEM shipments. 2019 was mostly all 4G units, as 5G had just launched towards year-end, and 3G was de minimis, so we can back into a reasonable estimate for 4G content based on rev/units. From there, we used feedback from VAR (and the Company) of a 50% premium for 5G content vs. 4G. Therefore, we can build up to an expected revenue estimate based on 3G, 4G, and 5G units multiplied by their respective content values. From there, we can compare our estimate to the actual reported China revenue by the Company to get to an implied over-ship (selling more into the channel than they shipped out) or under-ship (selling less into the channel then they shipped) amount.

While we hazard there is moderate quarterly noise, the result suggests significant over-ship by Qorvo into the channel in the first nine months of 2021, culminating in a peak cumulative $351mm of excess inventory in 3Q21. This excess channel inventory looks to have been worked down partially from 4Q21 through 2Q22 (largely in-line with VAR commentary), and from 3Q22 through 1Q23, we estimate the Company will under-ship sell-out demand (which is at a trough itself!) by 20-40% per quarter until neutral conditions are reached in 2Q23: 

The Company has, in effect, already guided to this abysmal performance through C1Q23. Two quarters ago, the Company guided to C4Q22 being a low point for their China business at ~20% of sales (roughly corresponds to $200mm of sales, which is in-line with our estimate above). QRVO subsequently lowered this guide last quarter, to just 10% of sales. We believe this timing and magnitude is roughly right; they will likely start to grow nicely on a y/y basis in 2Q23 through the end of 2023. We think Chinese rev can be somewhere around $1.2bn for CY23 (compares with $1.12bn in CY19, $1.35bn in 2020, $1.72bn in 2021, and $0.85bn in 2022). 

The Company has not offered any guidance after C1Q23, which we think is largely due to its fiscal year timing (we are in FY23 which ends March 31, 2023). On its C2Q22 earnings call, mgmt. guided directly to C3Q22 and then gave an indication for how the rest of the fiscal year would look (which, as described above, would encompass effectively the worst of the under-ship dynamic).

We think this dynamic is highly misunderstood by the Market. Due to all the moving pieces with 5G mix growing considerably in 2020 & 2021 and channel inventory dynamics, it is hard to know what “normalized” China revenue looks like. Most of the Street does not break out China in their modeling. We think some of the Street is crudely using CY19 as a base year for China rev to normalize to. That may sound like an elegant heuristic to avoid any whiplash effect from COVID – but it is completely wrong. 5G had just launched in 2H19, and we estimate global 5G shipments were only 16mm in CY2019 (9mm excluding Samsung). That is just 1% of the ~681mm shipments by China-based OEMs in CY19. 5G mix will be ~53% in CY23 for China-OEM Android, and simplistically, that should produce a 26% uplift to revs w/ a 50% content premium for 5G phones (52% increase in 5G mix * 50% ASP uplift = 26.0% blended ASP increase from CY19 to CY23; only to be offset by lower total units). 

We think CY2023 will be a massive growth year for Qorvo’s China OEM business, as it laps the trough end-customer shipments & under-shipping dynamics in 2022. This will lead to nice overall growth for Qorvo’s mobile business in CY2023. We estimate reported China revenue will go from -51% y/y in 2022 to +41% in 2023 (and +20% in 2024 as C1Q23 still has some under-ship in it): 

Below – one can see the impact to overall Mobile sales growth from China OEMs. China contributed 10 points of rev growth in CY20, 14 points in 2021, and we believe will be a 25 point headwind for 2022 (as is implicitly guided to by the Company). This will turn from a headwind to a tailwind in CY23, where we estimate (even with further destock in C1Q23), China growth will contribute 10 points of rev growth:

If we are right on China, then for Street is implicitly modeling RoW smartphone revs (Apple, Samsung, and a small amount of other Android) to massively decline in CY23. 

This makes little sense, as Apple is relatively stable – consensus estimates for Apple have iPhone units flat in CY23 vs. CY22 (ex- a recent production issue) and based on recent teardowns we believe Qorvo’s share is the same. Samsung should continue to grow nicely along with 5G penetration and share gains. See recent comments from Qorvo: 

• “And to your point, we are winning, we are taking share, gaining content at large customers, including the Samsung and Honor, as you mentioned. And we have reason to believe that those gains, especially in Android, will continue. These customers are adding the content to achieve the performance improvements that are embedded in 5G and other functionality. And we've been very encouraged by design activity and support of future architectures.” – Qorvo at a GS conference on 9/13/22

• “Of note, Qorvo more than doubled revenue year-over-year at Samsung with growth across multiple product categories. We also expect content gains as the year progresses across on our smartphone portfolio with opportunities spanning multiple products and technologies … As an example, take Samsung, where we had previously been underrepresented and where the combined opportunity extends for years … Samsung is going very well for us. And part of it is their realignment of their product portfolio and their technology strategy lining beautifully with our road map. We've had, as you know, very good long-term relationships there. So the pump was primed, and we're really excited to ramp across our full integrated module portfolio as well as power management, tuners, and so forth.” – QRVO on 5/4/22

Taking a step back – let’s talk some about total smartphone shipments. 

Our view on total smartphone shipments

We have total smartphone units growing 6.5% in CY23 y/y (1.19bn to 1.27bn) and another 3.0% in CY24 y/y. 

Global smartphone shipments were in a significant growth phase in the early part of the 2010’s as smartphone penetration was growing from a base of 50% in developed markets while China was at just at 11% in 2011; this led to a 19% shipment CAGR from 2011 to 2017.

Growth began to taper out in the late 2010’s, as developed markets reached ~100% penetration and China reached ~78%; meanwhile smartphone innovation slowed, elongating the replacement cycle from ~2.8 years in 2011 to ~3.6 years in 2019 (or around 1.2 months per year). In 2022, we estimate an installed base of smartphones in China of 1.18bn (83% penetration), developed markets of 0.98bn (106% penetration; business users have multiple), and RoW of 2.51bn (46%), or ~4.68bn combined. Assuming the replacement cycle extended by 1.2 months per year as it had been pre-COVID, we estimate a 3.9-year replacement cycle in 2022). This would equate to 1.20bn units of replacement demand. Currently, we estimate 2022 global shipments will come in at 1.23bn units – that is 2.6% above our estimate of replacement demand alone. 

Outside of North America, Western Europe, Japan, and China, there is still scope for meaningful increase in smartphone penetration from the current ~46%. This “RoW” group has grown ~40ppt from a 6.6% smartphone adoption rate in 2011. Penetration has been increasing ~200bps per year in recent years in this population (400bps in 2018, 240bps in 2019, 220bps in 2020, and 170bps in 2021). Assuming this continues to increase at just 1.5% per year, demand from new smartphone users in RoW – excluding any growth from China or developed markets – would be ~110mm units (RoW population of 5.4bn * 1.5% penetration increase = ~80mm units; population growth of 1.3% = 72mm growth * 44% smartphone penetration = 32mm new smartphones for a total of 112mm). 

That means assuming NO penetration growth from here in China or developed markets, total “underling” demand for smartphones should be somewhere at least around 1.30bn units (1.20bn replacement units + 0.1bn new units). We estimate 2022 will come in 8% below that. 

We have total global shipments down 12% in 2022, recovering up 6.5% in 2023, and 3% in 2024. We think this translates to the replacement cycle stretching from 43 months in 2019 to >50 months in 2024.

Valuation

Compared with the S&P 500, QRVO used to trade at a ~20% discount in 2021 on an NTM P/E. At the same 20% discount, Qorvo would trade for 13.3x P/E currently. If this deserves a 13x P/E, for the current price to be right, EPS would need to be $6.90 – which compares with CY18 ($6.81), CY19 ($6.78), and CY20 ($9.43) [all using current share count retrospectively]. That is to elucidate that a LOT of pessimism is already baked in here. 

In our base case, we have Qorvo putting up $12.42 of Adj. EPS in CY2024, driven by normalization of current under-shipping to China as well as the continued secular growth of 5G units. 

To bridge to this 14% mobile rev growth from CY22 to CY24:

• Units

  • We estimate 1.308bn total smartphone shipments in CY24, which is 10% above 2022 levels but 5% below 2019. 

• Content

  • We model Qorvo’s content with Samsung increasing 10% from 2022 to 2024, driven by a 28 % point increase in 5G mix. We model Other Android content growing 6.6% from 2022 to 2024, driven by a 18 % point increase in 5G mix. We have Apple content flat as the iPhone has already mostly transitioned to 5G. This blends to a 2% content increase. 

• Channel under-ship normalizing

These assumptions get us to $3.43bn of mobile products revenue in CY24. 

We think 31% a OPM for the mobile segment is a reasonable assumption and at the low end of management’s target (Appendix 7 for historical gross margin structure vs. factory utilization). 

For the IDP segment, we model this growing to $1.15bn in CY24 – just 7.8% above CY21. Mgmt. guides the IDP market to grow LDD over time (this is conservative). We assume a 25% operating margin here, in-line with CY22e, which is below mgmt’s target operating model of >30%. 

Including bear, downside, and bull case (which we can detail in Q&A), this this works out to a blended PT of $147, or up 63% from the current price. 

Risks

• Near term quarterly noise – they could disappoint next quarter on worse than expected China OEM orders

  • Mitigant: Much pessimism is baked in here. Some context:

    • After QRVO’s Mobile Products segment saw continuing rev growth in C1Q22 of 7% y/y (56% 2 year stack and an all-in beat to Street), the Co guided the segment down 12% y/y in C2Q22 (total rev down 8% vs. Street expectations) incorporating their “current view of the COVID lockdowns in China, the war in Ukraine, and existing industry supply chain constraints.” This sent the stock down 15% over the proceeding week. 

    • On 8/3/22, they beat this updated guide by 1% when they reported C2Q22, but implicitly guided to high teens decline in the mobile segment y/y and total revs down 10% y/y. This was in-line with Street estimates at the time of reporting. 

    • On the 8/3/22 call, the Company guided to back half revs of the fiscal year 2023 (C4Q22 and C1Q23) down 10% compared with 1H. This translates to the second half of their fiscal year down 16% vs. Street estimates. The stock was flat after these results, however, which sheds some light on just how poor sentiment was going into results. 

    • While the market did not react negatively to this updated guide, QRVO has traded down a further 20% since then, directly in-line with broader semi’s. We would posit that QRVO already guided to much worse trends before this recent leg down of broader semi’s de-rating. Qorvo’s end-markets are smartphone focused and have idiosyncratic outsized exposure to Chinese OEMs which were impacted by exogenous events. As such, we think there is a good case to be made that Qorvo is already in the process of bottoming (over the next quarter or two) AND has already taken its medicine on the C2Q22 fwd guidance. 

    • We additionally note that C2Q22 was Grant Brown’s first quarter as interim CFO (now confirmed permanent CFO), so incentives do not seem to exist for much optimism embedded in guidance. 

• China competition

  • There are two emerging domestic RF suppliers in China, OneChip and Maxscend. These players could take share over time. 

  • Mitigant: None of the procurement contacts we spoke with indicated the domestic suppliers would take share from Qorvo any time soon, stating they were not close on the technology and only really supplying low end discrete components. 

• The one area we identified the Chinese suppliers may have won some share is on the low end phones for low band frequencies.

• Qualcomm competition

  • Over the past two years, Qualcomm has invested further in the RF front end. In September of 2019 they bought out the portion of their JV with TDK (discrete RF maker) that they did not already own. Over the subsequent two years they talked up RF front end more and have grown revenue in their RF segment nicely. There is risk to continued market share encroachment by Qualcomm. Qualcomm makes the SoC (baseband / main application processor for the phone), and it has looked to bundle RFFE content with the SoC in the past. 

  • Mitigant: Qualcomm reports revenue from transceivers and mmWave modules when it reports RFFE to investors. Qorvo does not make transceivers and does not yet make mmWave modules (these are all currently for the U.S. iPhone model). We estimate that ~80%+ of Qualcomm’s RFFE revenue comes from transceivers and mmWave modules, and therefore their growth has not really come from share gains from Qorvo. 

  • Mitigant: It sounds like some suppliers are actually switching away from Qualcomm now

• Apple Customer Concentration

  • Mitigant: while Apple is 1/3rd of Qorvo’s revenue, on the core RFFE tech QRVO’s share is fairly low. As such, there is likely room to displace Broadcom (have the mid/hand band module on the iPhone) or Skyworks (have the land band module on iPhones) over time, rather than the other way around. We have decent content in the iPhone w/r/t antenna tuners. 

• Excess inventory at customers other than Apple and Samsung

  • Apple business has been fairly stable – units have not whipsawed and Apple is notoriously good at supply chain management. Qorvo’s revenue with Apple is not up significantly and looks to be mostly in line with Apple units sold

• Samsung is less clear as Qorvo has been gaining share (as Samsung has been mix shifting to 5G). QRVO’s revenue went from ~$300mm with Samsung in FY21 to ~$500mm in FY22, largely on the back of these share gains. It is possible that some portion of this is excess channel sell in, but we have not heard of excess inventory of RF components at Samsung

Appendix

Appendix 1 – filter technical details

Appendix 2 – past frequency band jumps by cellular technology generation 

Appendix 3 – BAW vs. SAW

Appendix 4 – complex module by Qorvo

Appendix 5 – Mscience monthly android sales

removed for VIC post purposes

Appendix 7 – Gross margin vs. factory utilization

Appendix 8 – IDP segment 

Qorvo is well positioned in IDP. 

The Infrastructure and Defense Products Segment has several key end markets:

• Network infrastructure (we estimate roughly 70% of IDP)

  • Base stations (20-30% of network infrastructure)

    • Gan on SiC and SOI – differentiated products for power amplifiers

    • Has similar dynamics to cellular RF where content for 5G increases vs. 4G. Massive MIMO antennas have 10x the content. QRVO’s GaN based offerings are taking share due to superior efficiency. 

      • 

  • IoT / Home Wifi (~50% of network infrastructure)

    • RF/Wifi/Bluetooth for IoT

    • Power amplifiers and BAW filters for wifi applications in the home

  • Optical (10% of network infrastructure) 

    • Products which aid in data center, long haul, etc. Most products here amplify optical signals. 

  • Auto (10-20%)

    • Cellular / other connectivity applications in the car

    • Power semis (Silicon Carbide based converters for EVs and other auto applications)

• Defense and Aerospace (we believe roughly 30% of IDP)

  • Radar

  • Satellite Communications

  • Electronic warfare 

Qorvo’s chart of addressable market from its 2018 investor day (it has not provided a similar graphic more recently):

Disclaimer: This memorandum is for discussion purposes only and is not intended to be, nor should it be construed or used as, financial, legal, tax or investment advice or a general solicitation. This memorandum is as of the date posted, is not complete and is subject to change. The data contained herein are prepared by the author from publicly available sources and the author's independent research and estimates. Certain information has been provided by sources believed to be reliable, but has not been independently verified and its accuracy or completeness cannot be guaranteed and should not be relied upon as such.

Catalyst

• inventory destock ending
• demand recovery in china
• earnings beats
• continued growth of 5G