Description

Business Overview:

Globalfoundries Inc. (NASDAQ: GFS) is an American semiconductor foundry, and it is also the only scaled pure-play outside of China and Taiwan.

The business was previously spun out of Advanced Micro Devices in 2009 and operates four fabs across the US, Germany, and Singapore. As an aside, there’s a great podcast by Business Breakdowns on AMD filled with a detailed history of the semiconductor industry and some of the strategic moves that various semiconductor manufacturers took to achieve their current positions.

Since spinning out of AMD, GFS has made a number of moves including partnering with Samsung on their 14nm plan in 2014 and acquiring IBM’s microelectronic business in 2015. More importantly, GFS made the move to stop investing in leading-edge chip development in 2018. Since then, the business has looked to optimize the production of lagging-edge, feature-rich semiconductors that populate several secularly growing end-markets.

Industry Overview:

Semiconductors are loosely divided into leading and lagging edge. Leading edge chips are the ones that populate our news feeds and require high levels of R&D and CAPEX. Lagging edge chips are chips used in situations where faster computing is not required. Most chips are lagging edge. These lagging edge chips are frankly everywhere in modern goods with things like the 2022 Toyota Prius having 1,400 chips and various refrigerators having 50+ chips.

In the fab industry, TSMC is the largest dedicated fab in the world, and it makes up nearly 60% of all global revenue share. Taiwan is also home to another fab called UMC which accounts for ~6% of global revenue shares and has historically had an infamous reputation for poor and late product delivery (more on this later). SMIC is China’s state-sponsored fab which makes up for ~5% of global revenue share. In Korea, Samsung is a large non-pure-play fab that accounts for 16% of global revenue share. Finally, the US is home to GFS (~6% of global revenue share) and Intel.

Typically, fixed costs are 60-70% of the cost per wafer given the large amounts of semiconductor fabrication equipment (lithography, etching, deposition, oxidation, ion implantation, diffusion, etc. machines) needed for manufacturing. This makes fab utilization rate the largest determinant of profitability, and usually utilization rates need to be above ~80% for a fab to achieve profitability. This has led to many fabs—including GFS—signing LTAs to ensure future capacity.

Thesis Summary:

Now that that long overview is done, the crux of our thesis revolves around GFS’ locations in America and Europe. We believe that GFS’ geopolitical position provides them a structural advantage over competing fabs given current international tensions and US-EU subsidies for local semiconductor development.

We also believe that GFS’ 2018 move was incredibly strategic. This decision allowed them to reduce their exposure to the costly—and risky—bets that leading edge foundries are exposed to. Additionally, it allowed them to develop a set of single-sourced, differentiated semiconductor solutions which helps them dramatically grow their ASP and renew customer contracts at better prices.

Thesis 1: GFS’ Geopolitical Advantages

Sino-American relations have deteriorated significantly within the last few years, and it’s no old news that Sino-Taiwanese tensions are a significant concern for TSMC customers. Over COVID, Chinese lockdowns dramatically halted US semiconductor production capacity, and the US became painfully aware of a potential weakness in their supply chain. Most semiconductor customers are heavily reliant on East Asian fabs, and Europe and the US combined only make up 22% of global semiconductor capacity.

GFS offers a unique solution to this problem where concerns over geopolitical tensions have caused customers to seek stable manufactures. Following the 2021 semiconductor chip shortage, customers reevaluated their chip source locations with companies like Ford and GM partnering with GFS to create exclusive deals to better secure their supply chain. US businesses are increasingly concerned about the durability of their supply chains, and we’ve included some interesting quotes below:

"The supply agreement with GlobalFoundries will help establish a strong, resilient supply of critical technology in the U.S."  - Doug Parks, GM EVP

“With accelerating demand for 5G, Automotive and IoT applications, a robust supply chain is critical for ensuring innovation in these areas remains uninterrupted,” - Dr. Roawen Chen, Qualcomm Technologies SVP

“Taiwan Sees Accelerating Shift of Production Away from China” - Bloomberg

We believe that this secular shift will help GFS retain old customers and attract new ones. With the decline in perceived stability of Asian manufactures, stable US semiconductor manufacturers are poised to benefit from this instability because they can provide customers with supply lines that are insulated from international tensions. Also, GFS and the other competing US company in Intel have little direct competition given that the two are fundamentally different fabs. Notably, Intel primarily competes in the leading-edge semiconductor market with the development of their new “Intel 7” process for 10nm nodes and their “Intel 4” process for 7nm nodes. Intel has been written up a few times before on ViC, and we believe that these write ups provide useful information if you believe that Intel is the better US-based fab for this tailwind play.

The following is old news, but for the sake of being thorough, GFS has a unique opportunity to benefit from US support for domestic semiconductor efforts. The US Chips Act was passed in late 2022 which provided $39b in incentives for production on US soil and a 25% investment tax credit. Europe passed a similar act which also provides subsidies to manufacture chips. These benefits are largely geared to domestic fabs as they force foreign fabs to make a long list of concessions (profit and IP sharing, prohibitions from investing in China). With GFS and Intel being the only major players without scale in China—and especially given that Intel does not compete in the lagging edge market—we believe that GFS is a well-poised to benefit from these substantial policies.

Thesis 2: Usefulness of Lagging Edge

Since 2018, GFS has abandoned their pursuit of smaller nanometer processes and has instead focused on optimizing their 12+nm technologies. These are high growth end markets which include smart mobile devices, various smart home appliances, industrial sensors, Wi-Fi infrastructure, etc. This focus allows GFS to be more strategic about their R&D and CAPEX spend given the limited nodal evolution that they have to take.

Additionally, GFS’ focus allows them to better retain key customers and guarantee that they do not fall below operational capacity. Many lagging edge facilities that are attempting to enter the leading-edge chips market have traditionally been plagued with timely delivery issues and poor-quality chips, and we believe that GFS’ dedication to the lagging edge allows them to carve out a unique niche for themselves. Notably, GFS’ ability to continually deliver these chips has allowed them to have 65% of their volume be single-sourced and enjoy LTAs (~50% of their revenues are from LTAs) that allow them to lock-in capacity for 3-5 years. Additionally, they’ve been able to command ASPs in a nice sweet spot between leading and lagging edge (~3000 for GFS vs ~4500 for leading edge players like TSMC and ~2000 for other lagging edge players).

There are some concerns that traditional leading-edge players might transition down. That said, TSMC’s expansion plans have exclusively remained in the leading-edge space and the probability that they transition down is low given the demand for 3-5nm chips. Additionally, even if a player were the transition down the lagging-edge, GFS’ first-strike advantage allows them to better price their offerings and continue winning that market. Notably, greenfield buildouts are extraordinarily expensive while adding modular capacity is much easier given higher capital efficiency and economies of scale. Some estimates place modular expansions as 20% more efficient than greenfield ones. 

Valuation:

GFS divides their revenues by end markets (smart mobile devices, comms, home and industrial IOT, automotive, personal computing, and other). We have little advantage in projecting out end-market growth compared to consensus, so we pulled consensus forecasts for these various growth markets which implied out a 7.6% top-line CAGR for the entire business’ top-line growth until 2028E. This set of forecasts allows us to get to an 83% implied utilization rate by 2028E. Of course, semiconductors are notoriously historical, but we believe that we are not near the top of the semiconductor cycle. The pandemic meant that consumers overspent on electronics, and since then, worldwide semiconductor sales have lost much of their pandemic gains in the last few years. This gives us some confidence that we are not forecasting from the top of what is a highly cyclical business.

Management is targeting a 40% gross margin rate for the business, and we approximately reach that target by 2028E as well. Notably, we gave the business 4% in annual fixed cost inflation, had D&A roll-off, and calculated variable costs as a function of wafers produced. 

Finally, given how sensitive valuation was to CAPEX, we built CAPEX based on expansion guidance into 2028 with different costs for modular and greenfield. We assumed an 8% expansion cost inflation and various subsidies per region. Below, we have provided some detailed estimates:

These assumptions give us a 12.0% IRR.

Risks

1. Intel entering into the lagging-edge market

Outside of TSMC, Intel looks like the potential greatest risk to this GFS thesis. Intel has repeatedly been aiming to become a leading outsource fab and they are also likely to benefit from many of the secular tailwinds and subsidies that GFS is. That said, Intel is extremely early stage and potentially half a decade behind GFS’ lagging-edge production quality. GFS’ operational successes, LTAs, and the fact that they don’t need to have large greenfield expansions also help insulate them from this risk. Finally, Intel has been vocal about catching up to TSMC in leading edge chips, and it seems unlikely that they can divert the resources they need to optimize lagging edge production as well.

2. Continued Unprofitability

GFS has historically had issues with profitability and maintaining high utilization rates. That said, we believe that 2018 was a fundamental year for the business which has allowed them to dramatically taper off CAPEX and R&D spend. Additionally, they’ve been able to command higher ASPs in recent years and their differentiated product delivery abilities in lagging edge should become more apparent in the near future. Finally, this management team has been expanding only for proven demand which helps mitigate some concerns.

Catalyst

Increased utilization rates, further US + EU specific benefits for local semi players, increased ASP growth