ARM: Priced for perfection

Simple big idea

We believe that ARM is a fundamentally sound business trading at an unreasonable valuation post-IPO.

When given very optimistic growth rate assumptions such as a 40% base case EPS CAGR over 2023-26, the stock still suffers more downside than upside. For this reason, we believe that ARM is a sell, given that stress test parameters still fail to save the giant from multiple de-rating in line with comps over the coming three years.

Business overview

Business

ARM Holdings (NASDAQ: ARM) is a semiconductor and software design company based in Britain that services companies around the world and it went public on September 14, 2023. We recommend selling ARM as we believe that the share price is too volatile in the short term, but ARM has strong fundamentals and should perform well over a three to five year period. We’ve identified three core ideas that drive this short recommendation: rapid growth in three of Arm’s largest market segments which will increase share price, current overvaluation which could cause the share price to fall, and the impending threat of RISC-V architecture that could move the share price either way.

Edge computing and internet of things, cloud computing, and autonomous vehicles are three large market segments, making up 43% of ARM's 2025 projected revenue, that ARM is well-positioned to target. The common advantage ARM has in these three segments is that their chips are designed to be lightweight, but high-performant. ARM has created optimizations on both the hardware side through updating hardware architecture (most recently ARMv9) and on the software side (by building a unified ecosystem of products), and have been able to optimise machine learning model performance on technology as simple as a tiny Raspberry Pi.

Edge computing refers to the process of bringing data storage and processing capabilities closer to information being produced. In practice, this means shifting away from sensors which send data to a centralised data centre for processing in favour of adding the capability to process data directly to these devices. This leads to improvements in application performance, reduced network bandwidth requirements, and an overall more efficient process. ARM has the ability to re-architecture the internet for the edge computing space by adding an additional to the existing OSI networking model to handle low-latency requests and improving the hardware used in these embedded systems. In addition, the looming concern of data privacy and security is handled by ARM’s ecosystem of products, which come with a variety of security technologies built in and additional requirements handled on a customer-by-customer basis. ARM has already begun to capitalise on the rise of edge computing as each of the chips in ARM’s Cortex line (A, R, M) are optimised for certain use cases in the edge computing space. Edge computing is expected to make up 20% of ARM’s royalties by 2025.

Rather than maintaining on-prem data centres, more companies are examining pay-by-usage cloud computing as an alternative. Cloud computing allows for on-demand usage of shared IT resources and is the best way for a company to build a service without incurring significant overhead from physical servers. Since 2018, ARM has slowly gained traction in the core server market that was initially dominated by Intel and Amd, the x86 processors. Improvements in chip architecture can lead to an increase in performance per watt and decreases in total cost of ownership. However, ARM’s growth is bounded by fierce competition in this market and currently only has around 5% market share, compared to Intel’s 75% and AMD’s 20%. However, ARM is positioned to capitalise in the cloud computing space through ARM based SmartNIC accelerators. These chips are core to the new era of cloud computing and offer significant improvements to networking and security. ARM has aspirations in many parts of the cloud computing market, and cloud computing is expected to make up 16% of ARM’s royalties by 2025.

Autonomous vehicles (AV) is another growing market segment that ARM plays a key role in. Increasingly segmented hardware is now favoured by the AV industry; they are able to lower the number of parts needed per vehicle even as complexity of the system increases. These AV systems require the same lightweight and high-performance chips as edge computing. Computations need to be done on-vehicle due to latency issues, but many of these machine learning computations require more computing power. In addition, ARM unifies the complex moving parts of an AV system through a universal processor architecture, as no similar universal language exists for sensors to communicate. Again, ARM’s Cortex line of chips are used to handle the massive set of complex tasks onboard an AV, such as autonomous driver assistance system (ADAS) functionality, embedded control tasks, and traditional microcontroller tasks. ADAS currently uses an average of 13 ARM chips and this number is expected to increase as the AV space continues to grow. Autonomous vehicles are expected to make up 7% of ARM’s royalties by 2025.

ARM generates revenue through two main product offerings, technology licensing agreements (TLA) and architecture licensing agreements (ALA). The distinction between these two is that in TLA, users pay a licensing fee to use ARM’s existing chip designs (such as Cortex), which is either limited by a duration or a number of uses, while in ALA, users develop their own chip design using ARM’s architecture and pay an architecture fee to do so (such as in Apple’s M-line chipset). Thus, a core part of ARM’s revenue are royalty fees and it is bound to the same revenue cycles as the broader tech market. There exists both a yearly cycle due to product releases and a 5 year cycle where outdated technologies are updated. As such, ARM’s share price is largely unpredictable and we base our sell decision on the high premiums on ARM’s share price, which are discussed in the valuation section of this report.

Industry

The global semiconductor IP market has been valued at US$ 6 Bn in 2022, and it is projected to grow at a CAGR of 6.7% over the forecast period of 2023-2032. By the end of the forecast period, the market is expected to reach a size of US$ 11 Bn.

The Semiconductor IP Market share is well-consolidated. Large-scale vendors are capable of backward and forward integration. Many players also operate the business in both national and international territories. The major players are adopting strategies such as product innovation and mergers & acquisitions to stay ahead of their competition.

The Semiconductor IP Market players are Arm Holdings Ltd., Synopsys Inc., Imagination Technologies, and Cadence Design Systems, Inc. These four companies together hold 65-70% of the market’s licensed IP cores. The other key players are Fujitsu Limited, CEVA Inc., Achronix Semiconductor Corporation, Faraday Technology Corporation, Lattice Semiconductor Corp., eMemory Technology Inc., Rambus Inc., etc.

Industry headwinds:

Demand Deterioration.

After a surge in demand due to the pandemic, there's now a reduction in demand, especially in markets like PCs and smartphones.

Excess Inventory.

There are increasing levels of surplus inventories throughout the semiconductor supply chain because of weakening end demand.

Supply Chain Challenges.

There's a move towards increasing regional diversification of supply chains. Customers are looking to have their semiconductor supply come from multiple regions, which might lead to inefficiencies or challenges in logistics.

Political & Geopolitical Issues.

The semiconductor industry has been thrust into the limelight due to ongoing tensions between the US and China.
About 70% of semiconductors are manufactured in Taiwan and China, leading to supply-chain vulnerabilities and perceived national security risks.
The semiconductor sector has become a part of recent US legislation, like the CHIPS Act and export restrictions on US companies producing semiconductors in China.

Regulatory Restrictions.

The US government now has the authority to review any semiconductor-related technology sold to China.
New restrictions are in place to stop the shipment and manufacture of chips that could be used for military applications in China. This hampers China's aim of becoming a prominent player in the semiconductor manufacturing sector.

IP issues.

As semiconductor IP becomes increasingly valuable and complex, there is a growing concern about protecting intellectual property from unauthorized access, replication, or reverse engineering. The fear of IP infringement and security breaches can lead to reluctance among companies to share or license their valuable IP, hindering the growth and adoption of semiconductor IP solutions.

Industry tailwinds:

Reinvestment, R&D and leaning into the future.

The overall value creation from semiconductors has accelerated rapidly in the last five years according to a recent McKinsey report. The industry will likely continue to consolidate since any new big player would be too difficult to create. 2020 will have the second highest total dollar value of mergers of the past seven years. The US semi industry as a whole allocates 16% of revenue towards reinvestments into R&D. That’s second only to the pharmaceuticals industry at 21%. As for leaning into the future, many of the best semiconductor-related businesses are also cutting edge innovators in Artificial Intelligence and Machine Learning.

Secular trends / cyclicality .

Semiconductors used to be considered cyclical because big single product releases would drive up demand that could quickly be followed by dry spells. Today, the demand for electronics is much wider and new digital products are introduced on a near-continuous basis. The integrated role of semiconductors in everyday living is taking some of the cyclicality out of this very capital intensive business.

Today, most of our favourite trends and themes, digital health solutions, public cloud, digital advertising, AI, etc, coincide with a long-term tailwind for the semiconductor industry.

In particular, AI represents a large opportunity for semiconductors as some expect the semiconductor industry to capture up to 50% of the AI-related revenue pie. One can reasonably assume that more semiconductors will be needed in order to match the continuous demand for cutting edge technology to support current and future aspirations of current and future entrepreneurs.

Increasing demand for advanced semiconductor technologies

The growing demand for advanced semiconductor technologies, such as System-on-Chip (SoC) and Application-Specific Integrated Circuits (ASIC), is a significant driving factor for the global semiconductor IP market. As companies strive to develop innovative products with higher performance and functionality, the need for semiconductor IP cores to accelerate design and reduce time-to-market becomes crucial.

Rising adoption of IoT and AI applications

The proliferation of Internet of Things (IoT) devices and Artificial Intelligence (AI) applications drives the demand for semiconductor IP. IoT devices require specialised semiconductor IP to enable connectivity, sensor integration, and low-power operation. Similarly, AI applications rely on semiconductor IP for high-performance computing, machine learning, and neural network processing. The increasing adoption of IoT and AI technologies across various industries is fueling the growth of the semiconductor IP market.

Market dynamics

ARM is the dominant player in the semiconductor IP industry, holding double the market share of its nearest competitor. It has also been a share taker of the last last two years. While this positions it well for continued fundamental growth, even if this fundamental growth occurs to the tune of 40% EPS growth annually, the multiple compression on sky-high semiconductor valuations coming out of the 2022 semiconductor cycle high due to COVID-induced supply constraints still result in a negative base case IRR.

Key Risks:

Geopolitical risk:

Exposure to China:
ARM Holdings, has a significant exposure to the Chinese market, which presents both opportunities and challenges. China is one of the largest markets for smartphones and other consumer electronics, many of which are powered by ARM's technology through its licensees. This makes the Chinese market a substantial source of revenue for ARM, as the demand for ARM's energy-efficient processor designs is high in a country that is the world's largest manufacturer of electronic devices. However, ARM's exposure to China also comes with risks. The geopolitical tensions between China and Western countries, particularly around technology transfer and intellectual property, can create a volatile business environment. The U.S.-China trade war has already shown that companies can be caught in the crossfire of tariffs and trade barriers, which can disrupt supply chains and affect revenues (ARM lost 63$ million in royalty revenues in the previous fiscal year due to Biden Export controls.
Moreover, China's push to become self-sufficient in semiconductor technology poses a long-term risk to ARM's business model. The development of domestic chip architecture, like RISC-V, could potentially reduce China's reliance on ARM's technology, impacting ARM's market share. Despite these risks, ARM continues to engage with the Chinese market, navigating the complex regulatory and political landscape. The company's technology remains integral to many Chinese products, and ARM's strategy has often involved partnerships and joint ventures to maintain its presence and influence in this key market (especially through its adjacent entity ARM China).
Arm China:
Arm China is a joint venture established between ARM Holdings and a consortium of Chinese partners, designed to enable ARM to effectively serve the local system of partners and chip manufacturers in China. This is critical because Arm must license its technology to Arm China which is its biggest customer is then responsible for distributing its products throughout the region. Therefore some risks of Arm Chian include:
Regulatory risk: The Chinese government has significant influence on business operations in China so it can at any time shift its attitudes towards Arm China thus altering how the company can obtain revenues in the region.
IP Theft: Ensuring the protection of IP rights within China is a big concern for ARM Holdings as if the Chinese government uses its authority to obtain these products it could result in many competitoros accessing ARM’s technology and thus undercutting and stealing the company’s market share.
Corporate Governance: Arm Holdings attempted to remove Allen Wu from his position as CEO of Arm China over many years but remained unsuccessful until 2022 when he was ousted from the board. Nonetheless, Wu maintains control over the company’s operations in China despite ARM Holding’s efforts and the appointment of an interim CEO. This may result in immense losses of revenues in the future if Wu seeks to separate Arm China from Arm Holdings furthering uncertainty regarding the company’s cash flows in the future.
Export Controls:
The Biden administration's export controls, particularly those aimed at curbing China's access to advanced semiconductor technology, can have several implications for ARM Holdings' business. These controls are part of a broader strategy to protect national security interests and maintain technological leadership. Here are some of the risks associated with these export controls for ARM Holdings.
Entity List Restrictions: If the U.S. Department of Commerce places certain Chinese companies on the Entity List, ARM Holdings may be prohibited from licensing its technology to these companies without a special license. This could directly impact ARM's revenue if the companies affected are significant licensees of its technology.
Deemeed Export Controls: These controls regulate the release of technology to foreign nationals within the United States. ARM would need to ensure that its technology is not inadvertently transferred to individuals associated with restricted entities or countries through its workforce or partnerships such as certain Chinese companies or authorities.
Foreign-Direct Product Rule: This rule extends U.S. export control regulations to foreign-made products that are based on U.S. technology or software. As ARM Holdings licenses its designs to be used in products worldwide, this rule could affect its ability to license to companies that might use its technology in products that are sold to restricted entities.

Industry Specific Risk:

The semiconductor IP industry comprising of companies such as ARM, Qualcomm, Synopsys, and others faces immense risks today including rapid technological development of open source technologies, customer consolidation, market saturation, regularity compliance, global competition, IP theft, and more. Despite this ARM has positioned itself as a market leader in the semiconductor IP industry, especially for CPUs used in mobile devices with their licensing of IP and the creation of an ecosystem enabling a vast array of companies that produce chips for phones, tablets, and increasingly other devices used in the IoT and the automotive sectors to select their products over other companies. Having said all this the 3 main industry-related risks that can be correlated to ARM Holdings' position today and in the future include the consolidation of customers of ARM holdings, the development of RISC-VC and x86 and the aforementioned outcomes of changing trade policies and their implications on ARM’s revenue streams.

Consolidation of ARM’s customers:
Consolidation of ARM customers is a rising issue with many semiconductor manufacturers acquiring others in recent years. This along with rising trends of customer acquisition in the low-rate environment until recently may erode ARM’s profits as customer consolidation results in negative outcomes for ARM including higher bargaining power of buyers, more room for in-house software development, reduced diversification of revenue streams, and volume discounts. This already presents a risk to ARM’s business since even today 57% of the company’s revenue came from 5 customers including ARM China which accounted for around 20% of the company’s revenues. (see below for some examples of recent consolidation of SMC manufacturers).
Intel’s purchase of the Mobileye for $15.3 billion
Analog Devices purchase of Linear Technology for $14.8 billion
Toshiba purchase of OCZ Technology Group for $35 million
Development of RISC-VC and x86 as subsidies for ARM:
Competitive Pressures from RISC-VC: RISC-V is an open-source instruction set architecture (ISA) that enables companies to design their own processors free of licensing and royalty fees. This could reduce the demand for ARM's proprietary ISA, as companies looking to cut costs might opt for the free RISC-V ISA instead. The presence of a free if not as technologically advanced software like RISC-V could pressure ARM to lower its licensing and royalty fee’s hurting top line and bottom line both.
Market share in EME’s: RISC-V has the potential to gain significant market share in emerging markets, where cost sensitivity is higher. This could limit ARM's growth opportunities in these regions.
Ecosystem shifts: If current customers of ARM switch their ecosystem, towards one more geared towards open-source software this may cut ARM entirely out of the supply chain in the future, thus costing it immense revenue streams. Thus ARM must remain technologically advanced than these open source competitors to ensure it is not undercut. However, it is important to note that this may be harder to do in the future as recently a bunch of customers and competitors of ARM holdings formed a consortium aimed at advancing the use of RISC-VC software. SMC industry players including (RobertBosch GmbH, Infineon Technologies AG,Nordic Semiconductor, Qualcomm Technologies, and NXP®) came together to invest in a company aimed at advancing RISC-V adoption. In addition, Google, Samsung, Qualcomm, and Nvidia formed a consortium in 2020 to develop RISC-V-based technologies as they were worried about Nvidia’s bid for ARM as this would have offered the chip manufacturer a competitive advantage if anti-trust legislation had not dissolved the merger.

Risks to a sell thesis on ARM:

Acknowledging all the aforementioned risks on ARM as risks to a long thesis our thesis on the company is sell therefore it is integral to mention risks related to a sell thesis on ARM. The primary risks to this thesis include exponential growth of the industry, Licensing model resilience, and Acquisition potential.
Growth of the Industry and ARM Holdings:
The semiconductor IP industry is poised for exponential growth, driven by the relentless expansion of technology into every facet of modern life. As the leading provider of chip architecture for a wide array of devices—from smartphones to the burgeoning Internet of Things (IoT)—ARM Holdings is strategically positioned to capitalize on this surge. The company's scalable licensing model means that as demand for semiconductor IP intensifies, ARM's revenue potential escalates with each new chip produced by its licensees. This growth is not just in volume but also in value, as ARM's advanced IP for next-generation technologies commands higher royalties. Moreover, ARM's energy-efficient designs are pivotal for emerging tech trends like electric vehicles and sustainable tech, further cementing its relevance. As the market recognizes ARM's central role in powering the future of technology, investor confidence should rise, leading to an appreciation in ARM Holdings' share price, reflecting the company's growing intrinsic value resulting in automatic correction of a sell thesis into a long one.
Resilience of ARM’s Licensing Model:
ARM Holdings has an extremely resilient revenue stream as part of its licensing model, underpinned by its strategic positioning in the SMC ecosystem. The company licenses its IP to a diverse array of clients across multiple industries, ensuring a broad revenue base that can withstand sector-specific downturns. This model offers scalability, allowing ARM to benefit from the growth of its licensees with minimal additional costs for each unit their clients sell. The upfront licensing fees provide immediate revenue, while the running royalties create a long-term, predictable cash flow. ARM's continuous innovation leads to the development of new IPs that keep the company at the forefront of technological advances, ensuring ongoing demand for its licenses. Additionally, the energy efficiency of ARM's designs aligns with the global shift towards sustainability, further securing its market position.
Acquisition Potential of ARM:
As evidenced in the NVIDia incident, ARM is a business with immense acquisition potential. This presents a risk to the sell thesis as if ARM is poised to be acquired this would probably send its share price higher, thus hurting our sell thesis. A potential acquirer could be a large semiconductor company seeking to expand its IP portfolio, a tech conglomerate diversifying into hardware, or a financial institution looking for a high-value investment. ARM's vast ecosystem of partners and licensees adds to its allure, offering immediate access to a broad market. Moreover, ARM's expertise in creating chip designs for next-generation technologies makes it a strategic asset in the race toward a connected and smart technology landscape. The acquisition would likely come with regulatory scrutiny, but the strategic benefits could drive a premium valuation, making ARM a lucrative target for industry giants aiming to secure a leading edge in the semiconductor space.

Market perception

The current market perception of ARM is uncertain, with many analysts assigning hold ratings but a hopeful price target. Below are some statistics from the Wall Street Journal: a large number of analysts are recommending a buy/hold due to uncertainty in ARM’s stock, but stock price projections seem to be positive.

Third parties seem to have an overall neutral to negative outlook on ARM because of heavy competition in the processor space. One week after ARM went public, two Susquehanna analysts assigned a “neutral” rating with a $48 price target, as they believe the shares are currently valued properly. They also expect the growth of ARM to significantly slow; they will soon be unable to continue raising royalty rates and many of their end markets are experiencing slowing growth. During the same week, Needham and Company also provided a “hold” rating and Bernstein provided an “underperform” rating with a $46 price target.

However, we believe that the market has overvalued ARM. The market is expecting ARM’s premium to expand to 100%, which we believe should only be about 20%. We discuss this further in the valuation section of our report. Thus, we recommend selling ARM. Looking at the share price of ARM since its IPO, it has continued to oscillate and has not seen significant movement in either direction, which we expect to change soon.

The impending threat of adoption of RISC-V will be one of the largest factors in determining the share price of ARM in the next few years. Introduced earlier in this report, RISC-V poses a large threat to ARM as the architecture is royalty free. Qualcomm, Google, Intel, and others are already testing RISC-V technologies. However, it is unlikely that this will create a big impact in the short term; ARM has been around since the 1990’s and has not been adopted in many spheres yet, thus it is unlikely that RISC-V will begin eating away at ARM’s market share in the near future.

Financial analysis

ARM has historically grown topline at a 15% CAGR over 2021-23, and bottom line at 16%. This mid-teens growth rate is nothing special in the semiconductor industry. Our three year projections for top and bottom line growth are a 23% and 40% CAGR respectively, a very generous acceleration from historical growth rates.

This growth is primarily driven by:

Scaling of the newer automotive, IoT and Infrastructure revenue segments as software is increasingly integrated into vehicles and household appliances.
Revenue share shift from Licence & Other to the higher margin Royalty revenue segments driving margin expansion.

The assumption of a 40% EPS CAGR from FY23-26E should be viewed more as a stress test than as a conservative estimate. Even with this very optimistic estimate, ARM’s valuation still has far more downside than upside, which will be analysed in the valuation section.

Valuation

Comps table

Comps are currently trading at a median P/E ratio of 100x vs 124x for ARM, a 24% premium. Consensus expects this premium to expand to over 100% over the next three years as multiples in the semiconductor industry de-rate due to slowing topline growth.

We believe that ARM does not warrant a 100% premium to comps, especially when the comp set includes heavyweights such as AMD and Nvidia. ARM is expected to have a 30% net income margin in 2026 vs 29% for comps, and it is expected to underperform comps in terms of 2026E EPS CAGR. ARM’s mediocre fundamental performance fails to justify such a steep premium over comps.

We believe that the current ~20% premium over comps that reflects ARM’s dominant industry position and potential for margin expansion is more reasonable than a 100% premium. Assuming a 20% premium over comps in 2026, we expect a base case P/E multiple of 25.8x. Combined with the generous 40% EPS CAGR over 2023-26, this produces a -12% IRR over the period despite very optimistic assumptions.

The risk/reward ratio on a short position on ARM is attractive at 22:0 (bear case vs bull case), meaning that even if the bull case of an unrealistic 45% EPS CAGR takes place over the next 3 years on 31x P/E, the short position will still not have suffered any of the stock’s upside. A short position on ARM may facilitate exposure to asymmetric risk/reward brought about by an overvalued IPO.