格芯

GlobalFoundries raised its 2026 Communication Infrastructure and Datacenters segment revenue growth outlook to 50s from high 30s and the Internet of Things segment to 10%-15% from midsingle digits, but this was overshadowed by a weaker outlook for smart mobile sales. The stock fell 5% on Wednesday.

Why it matters: The market was disappointed by the cut in smart mobile sales guidance to a midteens decline from a high-single-digits decline, as GlobalFoundries said in May that its premium-skewed smartphone exposure and wins in wearables should help it outperform the broader smartphone market. The stronger outlook for the CI&D segment reflects faster-than-expected capacity expansion and an additional $30 million of service revenue from newly acquired assets. We view the upside surprise as modest given the aggressive increase in capital expenditure budgets by hyperscalers. Management's explanation for IoT strength is simplistic, as smartphone and industrial IoT markets have little overlap. We believe industrial chipmakers are placing more orders with GlobalFoundries, driven by automation and energy infrastructure demand, regardless of consumer smartphone demand.

The bottom line: We maintain our fair value estimate for no-moat GlobalFoundries at $60. Following a roughly 50% decline from its June peak, we view the shares as attractive. The market appears overly cautious about competition, as other foundries also offer silicon photonics products. Our long-term expectation of a 12% revenue CAGR remains unchanged despite lowering our 2026 revenue and gross margin assumptions to reflect a weaker smartphone outlook. We expect stronger growth in the CI&D and IoT segments to offset mobile weakness beyond 2027.

Our fair value estimate is $60 per share, which implies an adjusted 2026 price/earnings ratio of 26 times and price/book ratio of 2.6 times. We expect the firm to increase revenue at a 11.8% compound annual growth rate over the next five years. GF’s growth drivers include 5G-related chip content (RF) as well as Wi-Fi, Bluetooth, NFC, optical modules, power management, and image sensors. Automotive is another bright spot, thanks to strong traction with automotive customers for an array of chips supporting autonomous, connected, and electric vehicles. Thanks to the fast growth of optical product revenue, AI revenue will exceed 30% of the total by 2030. The communications infrastructure and data center segment is the fastest-growing one, for which we expect CAGR to exceed 30% up to 2030.

We expect IFRS gross margin to increase from 25% in 2025 to about 40% by 2030 as the firm should be able to increase utilization in its fabs, serving more lucrative markets in AI data centers and satellite connectivity. This level of gross margin is comparable with GlobalFoundries’ peers, such as UMC and SMIC. We also think the firm can achieve GAAP operating margin expansion from 12% in 2025 to 27% in 2030.

The main upside risks to our forecasts are better-than-expected execution in improving gross margin and a more bearish geopolitical outlook between the US and China. For GF to meet its long-term gross margin target of 40%, we believe it has to overcome structural cost disadvantages by improving its product mix more than peers like UMC and SMIC. GF faces higher labor, construction, and transportation costs due to higher wages and being located further from some its supply chain partners. However, GF could benefit from geopolitical tensions if customers reallocate orders out of East Asia over longer periods.

We don't believe that GF has a moat. GF’s most advanced manufacturing process, or node, is over a decade behind leaders like TSMC and Samsung Electronics. Given GF's focus on specialty processes based on legacy process nodes, we do not believe it has meaningful defenses to protect pricing from supply and demand shocks. Customers have no difficulty securing alternative supplies in most nodes that GF offers. GF turned its first annual GAAP profit in 2022, but we do not expect excess returns on invested capital consistently.

As foundries debut more advanced nodes, the cost per area increases, but transistor density rises even more, resulting in a net reduction in cost per performance. However, few companies can offer the latest process nodes because semiconductor manufacturing is capital-intensive and requires methodical planning and execution to keep chip costs at reasonable levels. Ballooning development costs for advanced nodes mean only the largest foundries can afford the scale required for a decent return, in turn justifying premium pricing that is the main source of excess returns. We think GF’s small market share (5% in 2025) means it has insufficient resources to uphold such an investment cycle.

GF’s smaller presence also limits the extent to which per-unit research and development costs can be spread. GF’s most advanced node is 12 nanometers. In contrast, global leaders like TSMC and Samsung provide 3 nm products and have publicized road maps to 1.4 nm scheduled for 2028 or 2029. Industry publications suggest leading foundries can deliver manufacturing advancements at least until 2040. As of the time of writing, UMC and GlobalFoundries have said they will not engage in R&D of sub-12 nm nodes owing to insufficient returns. Currently, most sub-12 nm products are used on smartphone-related applications, with Samsung and TSMC effectively splitting the subsegment.

We believe GF is unlikely to close the gap with global leaders even if it wants to. In addition to high R&D costs required for advanced nodes, advanced plants are expensive to build. GF plans $700 million in capital expenditures in 2025, but this is miles away from TSMC’s and Intel’s capital spending of tens of billions of dollars each year. Our base case assumes GF will continue to optimize 12 to 40 nm processes for its various applications, like automotive and industrials.

Unlike leading foundries, GF does not enjoy special relationships with partners that can help build up know-how of advanced processes. For instance, TSMC, Samsung, and Intel invested in equipment vendor ASML to help the latter commercialize extreme ultraviolet lithography machines. These unique relationships help secure the latest equipment that comes with the best performance, the lack of which puts GF at a disadvantage in terms of production yield and speed. Assuming GF has a business case for acquiring state-of-the-art equipment, customers have no reason to pay a premium to help it learn the ropes when they can get the same product elsewhere more reliably.

GF’s average selling price is more than double that of other listed foundries, save for TSMC. Higher ASPs are a major factor behind an economic moat, but not sufficient in the case of GF. GF currently enjoys some advantage compared with SMIC and UMC as it has a larger contribution from the 12/14 nm node, but we think the edge is waning, as SMIC has helped Huawei produce 7 nm chips advanced enough for smartphones, and UMC has announced a partnership with Intel to roll out 12 nm offerings by 2027.

Although GF’s focus on developing higher-value differentiated or specialty products at older nodes may lead to some growth and margin expansion, we view consistent excess ROICs as unlikely. We now view the returns during the pandemic-catalyzed chip shortage as unlikely to recur. New plants have been announced, and when they come online, we think the mature foundry market is unlikely to fall into a deep shortage. GF also does not have cutting-edge business, such as new processors for iPhones and data centers, which we view as more economically resilient. Specialty products typically run on older nodes that use fully depreciated equipment, which translates to lower manufacturing costs. A common example is selling automotive chips that must adhere to stricter requirements instead of a chip for a consumer Internet of Things device. Specialty products tend to innovate more slowly than their microprocessor cousins, but they still face obsolescence risks, and these risks have been heightening in recent years. For instance, TSMC’s new sites in Japan and Germany offer new 7 to 28 nm capacity to service automotive and industrial customers, which threatens 12 to 65 nm alternatives by GF.

Profitability has been lackluster for GF. The company had accumulated losses of over $15 billion leading up to its initial public offering in October 2021. GF’s gross and operating margins have fared the worst compared with foundry peers both before and after going public. While GF’s financials have improved since listing, its cash margin—measured by adding back depreciation and amortization to gross profit—remains at the bottom. GF pitches itself as the only major pure-play foundry that has no production in China and Taiwan, but TSMC’s, Samsung’s, and Intel’s expansions are blunting the appeal of its proposition. By expanding in the US, Europe, and Japan, other major foundries will level the playing field by addressing customers’ needs for supply chain resiliency.

An unexpectedly strong boom in low-power, untethered devices will boost demand for GF’s specialty processes.

GF can continue its practice of using long-term agreements and single-sourcing contracts to mitigate volatility in its financial results.

Non-Chinese potential entrants are dissuaded by high capital requirements and lack of talent.

Single-sourced products may offer limited assurance as customers can find cheaper or superior substitutes from peers.

Chinese foundries are rapidly expanding to benefit from new demand for always-connected devices and shield the country from geopolitical risks.

GF’s absence in East Asia puts it at a cost disadvantage in utilities and labor.