STMicroelectronics ("ST") is a leading global analog and power semiconductor company. They operate a vertically integrated model under which they manufacture ~80% of products in-house and outsource ~20% of requirements to foundry partners. ST employs ~50k people. ST is incorporated in Netherlands. ST's headquarters are in Geneva. ST's front-end (i.e. chipmaking) facilities are primarily located in Europe (mostly France + Italy) with the sole exception being one fab in Singapore while its back-end (i.e. assembly + test) facilities are distributed more globally.

History

STMicro was formed in 1987 by the merger of two formerly government-owned semiconductor companies. These were the Italian SGS Microelettronica and the French Thomson Semiconducteurs. SGS Microelettronica was itself the product of a merger in 1972 between to local firms, ATES and SGS. Thomson Semiconducteurs was the semiconductor arm of Thomson-CSF (which was eventually rebranded as Thales). The company listed on the Paris and New York exchanges in 1994. The company took on its current name of STMicroelectronics in 1998. Thomson also sold its stake in the company in 1998. Carlo Bozotti became STMicro's CEO in 2005. Bozotti had joined SGS-ATES (later renamed SGS Microelettronica), a predecessor company to ST, in 1977, as a product and application engineer. STMicro formed a JV with Ericsson, ST-Ericsson, in 2009. The JV was a 50/50 partnership focused on wireless products. Nokia was a large customer. The JV reported PF 2008 revenues of $3.6b. It was a disaster and STMicro and Ericsson announced that they would wind up the JV in 2013 after failing to find a buyer and after five straight years of losses. The company changed course at the same time, moving away from leading-edge digital processors entirely and electing to focus on automotive, power and industrial end markets. Those remains the focus today. Jean-Marc Chery became STMicro's CEO on May 1, 2018. He took over from Carlo Bozotti. Chery had joined Thomson Semiconducteurs in 1986 and spent 32 years at the company prior to assuming the top job. His background relates more to manufacturing and design than it does to sales/marketing. He ran several of the companies' fabs over the years before becoming CTO in 2008, COO in 2014 and deputy CEO in 2017.

Business

ST reports under three segments. These are Automotive and Discrete Group (ADG), Analog, MEMS and Sensors Group (AMS) and Microcontrollers and Digital ICs Group (MDG). 

ADG is comprised of automotive ICs and MCUs as well as power/discrete semiconductors (SiC + MOSFET + discretes) that go into automotive/industrial applications. ADG revenues split ~70% auto, ~25% industrial and ~5% other. The automotive IC portfolio is broad and covers the powertrain (engine management systems), the chassis (airbags, power steering, anti-lock braking etc), ADAS (vision & radar), car body (HVAC, wipers, lighting, seat controls etc), infotainment & telematics (digital radio, navigation, wireless connectivity). The power/discrete portfolio covers silicon and SiC MOSFETs. ST was late to market in IGBTs (an alternative to MOSFETs for higher powered applications) and has chosen to focus on other products rather than playing catch-up. The result is that ST's IGBT portfolio and revenues are limited but that they are a leader in SiC. The power/discrete portfolio serves automotive as well as industrial/other end markets. The function of these power/discrete chips is to control and convert power within and around other chips. They are used to turn current on/off, convert current (AC to DC and vice versa), change voltage levels, modulate and control the flow of current and regulate it to ensure that devices receive a steady flow of precisely the current they need. 

The power semiconductor industry is benefitting from the transition to electric vehicles as EVs require substantially more semiconductor content than ICE vehicles and much of that step-up relates to power semiconductors. ST is particularly well-positioned as they have a strong expertise in Silicon Carbide (SiC). SiC is used in EVs for components like traction inverters (converts power from DC to AC to drive the motor), on-board chargers (manages the flow of electricity from the grid to the motor) and DC-DC converters (converts high-voltage DC from the battery to low-voltage DC for components). The strength of SiC relative to conventional Si is that SiC is able to operate at higher temperatures and has lower resistance, meaning less energy is lost. Together this means that the components can be smaller, saving both space and weight in the vehicle. There are three main steps in the production of a SiC product. You have to produce the raw SiC substrate. You have to turn that into a chip. You have to package it as a "module" that is ready to go into a vehicle. ST has historically done steps 2 and 3 while relying on partners like Wolfspeed and Rohm for raw SiC substrates. They are in the process of building and ramping capacity to do step 1 in-house too (targeting 40% of substrates produced internally by YE24). This is not straightforward. SiC substrate manufacturing is notoriously tricky but, assuming they are able to master the process, will have obvious benefits around value capture / margin stacking as well as surety and reliability of supply.

ST has historically been the sole-source supplier of SiC modules to Tesla. The main module they sell today is the traction inverter. They also sell the onboard charging module but that's a smaller business line. ST has said that SiC will make up ~$1.2b of revenue in 2023. They've also said that this splits 75% auto / 25% industrial. Then within auto, they've said that Tesla will be ~70% of SiC revenue (i.e. 90-95% of auto SiC) though will exit the year at less than 70% as other customers come online and start to ramp up (Hyundai/Kia is one large recently disclosed new customer). The fact that ST has been supporting Tesla since 2018 and played its part in helping Tesla ramp volumes exponentially while staying ahead of competitors in terms of efficiency/range should speak to the expertise around SiC, SiC chips and SiC modules that sits within ST today. The OMDIA data suggests the SiC market will be ~$2.8b in 2023. That implies that ST has 40-45% market share. ST is targeting $2b of SiC revenue in 2025 and has suggested that this target is well covered by existing design wins.

ST's participation in the ADAS ecosystem is also worth flagging. ST and Mobileye have been partners since 2005. The two companies essentially codesign Mobileye's EyeQ series of chip which sits behind all of Mobileye's ADAS solutions. STMicro has historically manufactured the chip and sold it to Mobileye. Mobileye's latest chip (EyeQ 5) is manufactured on TSMC's 7nm FinFet node so ST has shifted from a "codesign + manufacturing" role to a "codesign + manage manufacturing outsourcing" role. I'd initially worried that the entry of TSMC into the equation here would begin a process in which ST was ultimately squeezed out however having spoken to both companies it seems that there's still sufficient value-add all-around even that the current set-up can be expected to persist for the foreseeable future. 

AMS is comprised of analog, smart power, low power RF, MEMS sensors and actuators, and optical sensing solutions. The analog piece is ~40% of segment revenues and covers general purpose as well as application-specific products. STMicro has historically had more success with general-purpose analog and in particular in areas like smart meters, motor controls in industrial tools, computer peripherals and wireless charging. The MEMS/sensors piece is the other ~60%. The company don’t disclose it but a reasonable guess would be ~40% of the segment is sensors and ~20% is MEMS. ST's sensors cover a wide range of applications including front- and world-facing sockets in smartphones (for facial recognition, biometrics, ambient light sensing, camera focusing etc). Their highest-profile socket in this area is the FaceID chip that they have supplied Apple ever since it was launched in 2017. They serve some of the same use cases (i.e. facial recognition/biometrics) in PCs. In the automotive setting they can be used for driver/occupancy monitoring as well as LiDAR. The MEMS products end up in a broad range of application fields, including mobile, gaming, computer, automotive, industrial, healthcare and IoT.

MDG is largely comprised of general-purpose microcontrollers (MCUs) but there is also some contribution from specialty memory and RF products. MCUs make up ~80% of segment revenue with speciality memory another 5-10% and RF 10-15%. MCUs are best understood as basic, power-efficient CPUs that can control the physical actions of an object (be it a kettle or a robot or a conveyor belt or a washing machine). The MCU typically includes memory for temporary data storage and storing the program code itself. The MCU market is often segmented according to the size of their data bus (essentially how many bits of data they can handle at once). The three major tiers are 8, 16 and 32 bit MCUs. The 8/16 bit MCUs are slower, typically attach less memory and are often used in applications where the program size is small, and the complexity is low to moderate. The 32 bit MCUs are suited for more complex applications with more demanding software requirements. STMicro sells all three but is really focused on the 32 bit MCU market. This seems sensible as 32 bit products are taking greater and greater share of the overall MCU market as more and more products become sufficiently complex to warrant the beefier chipset. The specialty memory part of MDG is not worth that much time as it's mostly a bundled sale alongside the core MCU franchise so they can almost be thought of as a single product. The RF franchise really sells into two end markets, wireless infrastructure (for ex, beam formers for base stations and small cells etc) and LEO satellites (SpaceX is a large customer - STMicro makes components that go into both the satellites and the antennae). It doesn't produce anything that goes into smartphones.

The three segments make up ~43%, ~22% and ~32% of LTM revenue, respectively. In terms of LTM EBIT, they account for ~48%, ~17% and ~18%, respectively.

Thesis

The crux of the thesis here is that ST is just way too cheap (~9x 2023 P/E ex cash) for a good business run by a good management team with a pristine balance sheet (net cash) and several significant growth vectors that should allow it to deliver industry+ and GDP++ revenue growth for the next 3-5 years (and likely beyond).

The opportunity exists, I believe, because the LO community is still somewhat scarred by ST's prior life as a low margin, low RoIC, very cyclical business with meaningful revenue exposure to Apple and the HF community is short the stock on the thesis that auto/industrial demand is unsustainably high as customers restock/hoard inventory and that this will ultimately roll over in a big way. The other worry is that China's ongoing push to internalise its semiconductor ecosystem and aggressive investments in lagging edge capacity will mean cutting out firms like ST.

I think there are very good reasons to argue that ST today doesn't look like ST 5+ years ago and is capable of higher-than-historical through-cycle margins. The company has been strategically and successfully shifting its focus to higher value-add, higher margin segments and deemphasising lower margin, more cyclical, more commoditized products. The picture is muddied somewhat by various segment reshuffles over the years but I estimate with reasonably high confidence that what is now ADG and MDG was 35-40% of ST's revenue in 2007/08 whereas those two segments are 65% of LTM revenue. These are the quality assets in ST. I estimate (again, with reasonably high confidence) that ADG has grown at a 6% organic CAGR and that MDG has grown at a 10% organic CAGR over the last 15 years. The rest of the business ex wireless is effectively flat over the last 15 years. The wireless business was ~30% of ST's revenue in 2009 and went to zero over the course of 2012/13/14. If you were to apply those growth rates to the portfolio as it stands today, it implies that the company should be capable of ~6% growth. I think that's on the conservative side as the electrification angle has only emerged in recent years and should lead to higher growth rates on the automotive side than we've seen in the past.

I think that the company's progress pre COVID also supports the hypothesis that ST is delivering structurally higher margins today than it did in the past as opposed to just benefitting from a strong end market and tight supply conditions. The company abandoned their prior strategic goal of becoming the "undisputed leader in multimedia convergence" back in late 2012. Their GM was 32% the following year. ST's GM had reached 40% in 2018 and 39% in 2019. The company was well on its way to mix-shift its portfolio towards higher value, higher margin products. The EBIT margin had followed suit, rising from -3% in 2013 to 14% in 2018 and 12% in 2019.

The argument that ST's margins are sustainable is also supported by what we can observe at various peers. For ADG, the closest competitors are probably Infineon and ON Semi. They're expected to make high 20s and low 30s EBIT margins this year, respectively. For AMS, the portfolio is fairly broad so singling out any directly comparable peer is tricky. That said, I think comparisons to NXP (lots of general purpose analog) and Skyworks (lots of exposure to smartphones and, in particular, Apple) are useful. Those two companies are expected to make low 30s and high 20s EBIT margins this year, respectively. For MDG, the closest competitors are. Microchip and the embedded processor segment within Texas Instruments. Those are expected to make mid 40s and low 30s EBIT margins this year. The fact that every large, relevant competitor of ST's is making EBIT margins either close to or higher than 30% seems supportive of the idea that ST itself can also reach >30% EBIT margins over time. This would also be higher than the market is pencilling in currently with the street expecting margins to plateau at 27-28% in 2025 and beyond.

Apple was ~25% of revenue in FY22. I estimate that will be ~12% in FY24 and probably drops below 10% in FY25/26. This revenue is spread over multiple products and doesn't represent overly concerning concentration, in my mind. It's also worth noting that (a) ST has supplied the chip supporting FaceID ever since Apple introduced it and (b) the Android ecosystem hasn't really adopted FaceID so there's not another vendor out there really focused on this product and socket. That's not to say they can't or won't lose the socket. It's just to say that ~12% of revenue from Apple across a number of products and where there seems to be some level of stickiness for the largest single product strikes me as very manageable.

ST's three major end markets are autos, industrials and consumer. They also make stuff that ends up in telco infrastructure, PCs, servers, etc but those exposures are relatively small. The consumer market (smartphone in particular) has been in a downcycle since H2 2022. Their AMS segment which is the most consumer-exposed has seen Y/Y declines the last three quarters and last quarter reported $990m of revenue vs $968m in Q3 2019. This market is already well into its downcycle. The industrial market had been strong until recently but we've now started to see signs of weakness emerge. On their most recent earnings call, ST talked about weakening industrial demand in China and consensus now has Y/Y revenue declines for the next four quarters such that by Q4 2024 MDG will have grown revenue at a ~13% CAGR vs Q4 2019. The segment which continues to hold up extremely well is ADG which grew 30% Y/Y in Q3. The outlook over the next 6-12m seems fairly healthy as the segment is still supply constrained as it pertains to SiC, will begin ramping OEMs other than Tesla in earnest over the next 6-12m and as the Mobileye business continues to ramp. To hit consensus' ~10% growth in 2024, I estimate that the entire segment ex Tesla and ex Mobileye only needs to grow ~5%. That doesn't seem too great of a stretch considering the electrification tailwind. This is all to say that concerns around the cycle seem manageable - one of ST's segments is already well into a downcycle, one has just entered a downcycle (but numbers have come down as a result) and one is still healthy but the structural trends appear very powerful and numbers don't seem crazy.

The balance sheet is in great shape. ST reported ~$5b of cash against ~$2.5b of debt as of Q3. Their debt is well-termed out and/or low-cost so they are generating significant net interest income (~$200m run rate) in the current interest rate environment. ST will generate significant FCF this year, even as they add a large amount of capacity and then CapEx should fall fairly meaningfully in 2024 as they move out of the "build" and into the "ramp" phase at various new facilities so FCF is very likely to expand next year.

ST has a sensible strategy for the Chinese market. They are trying to create, where possible, a portfolio of products that can (a) leverage their IP, (b) take advantage of plentiful, cheap domestic foundry capacity and (c) look and feel "made in China" to satisfy customers who want to "buy local". For example on SiC, they announced a JV with Sanan ($10b mkt cap electronics company) for SiC for Chinese customers. The JV will buy substrates from local suppliers. The JV itself will build the SiC device (they are building the fab now). The JV will then leverage ST's packaging facility in China for final assembly and test. Then over time Sanan plans to build its own facility to produce SiC, which it could then sell to the JV. The total cost of the fab is expected to be ~$3.2b and the local government is contributing a portion of the funding. They are also leveraging local foundries for some of their MCUs that get sold to Chinese customers, supporting their desire to "buy local" while, again, taking advantage of plentiful, cheap domestic foundry capacity.

Upside

I assume ADG grows 10% in 2024, AMS is -7% (or flat excluding a very low value-add piece of assembly work that they did for Apple in 2022/23) and that MDG is -4%. Thereafter I assume ADG grows at 8% (a bit faster than history given electrification), AMS grows at 2% and that MDG grows at 8% (a bit slower than history given market share gains are layering onto a higher baseline today than they were in the past). That all nets out to ~7% revenue growth in 2025 and beyond. That leaves me at $21.6b of revenues in FY27. I have them just squeezing past their 50% GM target that year (50.2%). I've got OpEx growing at 6% which enables modest leverage and sees EBIT of $6.3b and EBIT margins hit 29% in FY27. That yields $6.7b of PBT and $5.9b of PAT. If we look at EPS excluding net interest income, it's $5.9 in FY27. I don't factor in any buybacks and only assume modest dividend growth which implies that the net cash position continues to grow. I have it reaching $15 per share at YE26. If we value the stock on 14x operating EPS (its historical multiple over the last 10y) that's $82. Then you'll get ~$1 of dividends along the way and have ~$15 of net cash on the balance sheet, which all adds up to $98. That's +134% vs the current price or a 31% IRR. You're creating the business at ~5x FY27 earnings, ex net cash. If I'm even close to right on the numbers it's very hard to see how you lose money at that kind of a valuation, making the risk/reward seem very favourable here.

Risks 

The cycle can't be discounted but I think you're buying at a valuation that leaves ample margin of safety on a 1+ year view.

The capital allocation at ST is not awful but leaves something to be desired. The dividend is paltry and they do buybacks to offset dilution. The willingness to build up cash on the balance sheet is somewhat understandable given the company's tumultuous history as well as the capital requirements of their growth plans but it still feels overly conservative to me. I would expect the dividend to grow over time but not explosively and I wouldn't expect the philosophy around buybacks to change. 

They could do a silly deal. ST has a long history of growing organically rather than by acquisition and I've specifically asked the CEO on this point and he was crystal clear that this is not the plan. That said, French companies with healthy balance sheets have a special knack for unexpected and unwelcome deals so never say never. 

I could be wrong about the sustainability of current margins. If demand contracts more severely than expected and firms are desperate for order, of course pricing can come under pressure. This is mitigated by the stickiness within markets like automotive and industrial where components, once designed in, are rarely swapped out but again, never say never.

The Chinese investment in the lagging edge has more downside than upside for ST. They are trying to be smart about taking advantage of that cheap capital - leveraging it via JVs or by using cheap local capacity - but there's clearly longer term risk that this investment supports the emergence of local firms who release competing products and chip away at ST both within and outside China. I'm not too worried about this. The manageable % of revenues that actually come from domestic Chinese customers, the quality of ST's technology and their "almost made in China" approach all mitigate the magnitude of this risk. There's also upside in parts of the portfolio if semiconductor ecosystems truly fragment and Chinese firms are unable to compete ex China.