Arm's impact on Nasdaq has become one of the most watched topics in the technology circle. Arm will be officially listed on Nasdaq on September 14. With a valuation of up to US$50 billion, it will become the world's largest IPO this year, and is expected to become the third largest IPO in the history of the US stock market after Alibaba and Facebook.
As a leader in the field of chip design, Arm is almost stuck in the "neck" of global technology giants. The Arm architecture is the most widely used CPU architecture in the world. It has produced hundreds of billions of chips so far, covering more than 99% of the global smartphone market.
Arm's listing has stirred up waves in the IPO market, which has been stagnant this year. At the same time, the hot AI wave is also constantly pushing up ARM's value.
However, although Arm holds architectural hegemony, it cannot make a lot of money. At the same time, its "AI quality" is insufficient. Can it support its high valuation?
Next, peel back the surface of the noise, delve into the prospectus, and see what kind of company Arm is.
The cornerstone of the industry is stuck in the neck of global technology giants
What kind of chip company is Arm? How did it rise and grow into an industry leader?
From the following introduction, you can understand Arm’s industry status and products provided. There is no doubt that Arm is the leader in the CPU industry. ARM is good at designing, developing and licensing high-performance, low-cost, energy-saving CPU products and related technologies. Its architectural advantages have made giants such as Apple, Nvidia, and Google its customers.
Arm's products have a wide range of reach. Smartphones, tablets, navigation systems and even network equipment almost all use Arm-based CPUs. About 70% of the world's population uses Arm-based products.
In terms of market share and shipments, as of December 31, 2022, Arm’s CPUs accounted for more than 99% of the global smartphone market, and cumulative shipments reached 250 billion. In the fiscal year ending March 31, 2023, chip shipments based on Arm architecture exceeded 30 billion.
1. Rise on the RISC route
From a historical perspective, the rise of Arm essentially lies in betting on the right RISC (reduced instruction set) route.
Arm was founded in 1990 as a joint venture between Acorn, Apple and VLSI. At the beginning of its establishment, ARM's goal was to "develop high-performance, low-power consumption, easy-to-program, and scalable processors." Over the past thirty years, this mission and market positioning have remained unchanged, and high performance and low power consumption actually follow the RISC route.
The basic architecture of the CPU is divided into two sets: RISC (Reduced Instruction Set) and CISC (Complex Instruction Set). The difference between the two is that CISC has high performance and high power consumption, while RISC performance is not as good as the former but has low power consumption. The representatives are Intel and Arm respectively.
In the PC era, computing speed and performance were pursued, so Intel's X86 instructions and architecture almost monopolized the entire computer chip market. In the mobile era, more emphasis is placed on the balance between computing power and power consumption. Arm, which uses low-power, high-performance RISC, is rapidly emerging. Arm CPU also plays a key role in the smartphone revolution.
2. "Licensing fee + royalties" two major ways to generate income
Looking further, Arm is in a special position in the industry chain and has a unique business model.
The entire chip industry chain is roughly divided into designing chips, manufacturing chips, and packaging chips. Before designing a chip, instructions and architecture are required, which is like a "design sketch."
To simply understand, Arm developed a CPU "design sketch", Apple and Nvidia bought the "design sketch" back, perfected and adjusted the design according to the needs, and finally found TSMC to manufacture the chip according to the design.
Without "design sketches", it is difficult to design a chip, and Arm is unique in the "design blueprint" of mobile devices. Arm pointed out in its prospectus that as the complexity of CPU design has grown exponentially, no company has successfully designed a modern CPU from scratch in the past decade.
Understanding Arm's cornerstone position in the industry chain, its business model is easy to understand. Arm makes money by selling "design sketches" (intellectual property), which mainly includes two aspects:
(1) License fees: Customers need to pay a fee to obtain Arm products ("design sketches") and then develop processors based on the Arm architecture.
(2) Royalties: When products based on the Arm architecture are designed and produced, customers need to pay royalties to Arm in proportion to the selling price of each chip. The commission is generally 1% to 2%.
In short, Arm not only sells "design sketches", but also collects taxes from the "finished products" produced by the "design sketches", and the "finished products" can bring a continuous stream of revenue.
In fact, there are different levels of "design sketches". Strong chip manufacturers need simple sketches to design chips, while less capable manufacturers need more precise sketches. At the same time, the architecture will also be different from old to new, which corresponds to different licensing restrictions.
Traditionally, most customers license Arm's products under the terms of the TLA. In 2019 and 2021, Arm launched the ArmFlexibleAccess and ArmTotalAccess protocols respectively to enable as many customers as possible to access Arm products in a way that best suits their specific business needs.
Holding architectural hegemony but not making big money
Although Arm holds architectural hegemony, its actual profits are not that huge.
1. Revenue and net profit are relatively small
Arm's performance has been poor due to the sluggish smartphone market. Operating income for the fiscal year ended March 31, 2023 was $2.679 billion, down from $2.703 billion in the same period last year, and revenue in fiscal 2021 was $2.027 billion.
Specifically, Arm's revenue sources are mainly divided into licensing fees and royalties, of which 40% are licensing fees and royalty income accounts for approximately 60% of revenue.
Among them, royalties are ARM's most profitable business. However, since most products based on the Arm architecture are MCUs (microcontrollers), the prices are very cheap. On the other hand, Arm takes a commission from the chip price rather than charging for end products, so the revenue it brings is not high.
In fiscal year 2023, Arm shipped more than 30 billion units, charging an average of about US$0.1 per chip. In 2022, the total value of chips based on Arm technology will reach US$98.9 billion, accounting for nearly half of the market share. But Arm’s royalty income is only US$1.68 billion, accounting for 1.7% of the chip value.
However, the advantage of royalties is that they are developed once and taxed for life, and old chips can provide an ongoing revenue stream. In fiscal year 2023, approximately 46% of Arm’s royalty revenue comes from products released between 1990 and 2012.
For example, ARM's ARM7TDMI architecture chips sold to Nokia and Texas Instruments in 1997 are still being shipped in 2020 and have continued to make money for more than 20 years.
In terms of expenses, as a chip design company, Arm's research and development costs are not low, accounting for about 40% of revenue on average.
This has resulted in Arm not being able to make a lot of money. Arm's net profits for the 2021-2023 fiscal years are US$388 million, US$549 million, and US$524 million respectively.
It has made a total of US$1.4 billion in the past three years, which pales in comparison with the profits of its customers. Qualcomm made a net profit of US$13 billion last year, and Nvidia earned more than US$4 billion last year.
The prospectus shows that the increase in net profit margin in the past two years is mainly due to the layoffs and restructuring plan implemented in March 2022. Judging from the implementation results, Arm has basically kept the total number of engineers unchanged, while significantly reducing the number of managers in the UK headquarters.
2. Insufficient say in the industrial chain
Fundamentally speaking, Arm cannot make a lot of money, probably because it does not have enough say in the industry chain.
Some analysts pointed out that in the entire industry chain, the most profitable ones are the chip companies that control the software ecosystem. They directly determine how competitive a certain chip is in the market.
According to a report by Wall Street analyst Robert Castellano, Nvidia sold the GPU chip H100 for US$40,000, TSMC earned US$1,000, and SK Hynix, which provides memory chips, earned US$2,000. The remaining over 90% of the revenue almost entirely belongs to Nvidia.
In addition, ARM has always been in a neutral position, and its architecture standards are open to major companies for authorization, causing its position to be unstable. On the other hand, its competitor Intel not only designs IP, but also owns its own chip factory. It monopolizes the ecosystem by forming the Wintel alliance with Windows, thus gaining control of the industry chain.
How good is the “quality of AI”?
Arm's listing coincides with the artificial intelligence boom, and enthusiasm for AI has driven Nvidia's stock price to rise sharply. In order to increase the valuation of Arm, Masayoshi Son is crazy about AI.
Masayoshi Son believes that ARM can achieve "exponential growth" with AI, and told investors that Arm is in the center of artificial intelligence-related companies and can produce synergistic effects.
1. The quality of AI is insufficient
However, judging from the prospectus, Arm’s AI is not of sufficient quality.
Arm mentioned in its prospectus that with the rapid development of artificial intelligence (AI) and machine learning (ML), Arm also plays a vital role in these fields and cooperates with many leading companies to handle AI work.
“CPUs are critical in all artificial intelligence systems, whether fully processing artificial intelligence workloads or combined with co-processors such as GPUs or NPUs.” However, Arm focuses on the architectural foundation of CPUs, rather than the architecture of GPUs and AI-specific chips required to create large models.
In the risk warning, Arm admitted that emerging technologies, such as AI and ML, may use algorithms that are not suitable for general-purpose CPUs (such as Arm's processors). Therefore, in chips based on Arm architecture, its processor may become less important.
2. Upgrades around AI and ML
In other aspects, Arm’s latest Armv9 architecture has made certain upgrades around AI and ML.
In May of this year, Arm officially launched its latest architecture, Armv9. This is the company's first major adjustment and change to its CPU architecture more than ten years since the launch of the Armv8 architecture in October 2011.
For developers and users, the most significant feature of the new Armv9-compatible CPU is SVE2 as the new benchmark after ARMNEON technology. SVE2 was released in April 2019 and is designed to accelerate high-performance computing and has great gains when handling 5G, VR and AR, and ML task loads.
Arm believes that ML workloads will become increasingly common in the coming years. Accordingly, any performance- or power-focused device will need to run ML workloads on dedicated accelerators, but most of them will still choose to adopt a smaller range of ML workloads running on the CPU.
3. Key flaw: Insufficient computing power
In fact, Arm's disadvantage in the field of AI is also obvious, that is, its computing power is not powerful enough.
The AI server field is mainly dominated by x86 architecture and GPU architectures such as NVIDIA Turing, Volta, and Ampere. Arm architecture is also gradually emerging in the server field, especially in lightweight and low-power servers, but its share is very small.
AI has very high performance requirements. The X86 architecture can use extremely high-performance CPUs, coupled with large-capacity memory and high-precision graphics cards, to continuously provide computing speed. The RISC route taken by Arm determines its performance shortcomings. If the performance cannot be further improved, the shortcomings of the Arm architecture will become more and more obvious.
Therefore, Arm seems to be on the edge of the AI boom, and it will be difficult to reap the dividends.
Growth is getting harder and competition is getting more and more volume
In recent years, Arm's growth space has become smaller and smaller, while competition has continued to increase.
1. Growth is getting harder and harder
In order to win the hearts of investors, Arm must find ways to generate higher revenue, but whether it goes left or right, there are minefields.
In order to increase revenue, Arm hopes to change the royalty rate from the chip price to the terminal price, and at the same time force "bundled sales" of CPU, GPU, and NPU.
But this is facing a lot of pressure, which in turn forces customers to re-develop their own architecture. It is reported that Qualcomm will bid farewell to the Arm architecture and adopt the self-developed Nuvia architecture for the first time next year.
Therefore, Arm has decided to focus on designing its own chip products. Some sources pointed out that Arm wants to design its own processors and give priority to using its most advanced architecture.
This is equivalent to stepping on the customer's territory. In the past, chip design manufacturers cooperated with them partly because of Arm's neutral position. If Arm steps into chip design and produces its own chips, it will be equivalent to stepping on the customer's red line.
These two "growth paths" are full of dangers. If Arm insists on doing so, it is likely to bankrupt the long-established ecosystem and accelerate customers to competitors.
2. Competition is getting increasingly fierce
For example, the open source and free RISC-V architecture is eyeing it and is constantly trying to break into Arm’s existing customers.
The RISC-V architecture has become a "rookie" in the field of chip architecture by virtue of its advantages of open source, simplicity, and modularity. The open source nature of RISC-V has attracted the attention of many companies, especially for small companies and start-ups, as it can more easily obtain chip design authorization, thereby reducing costs.
Arm also mentioned in the prospectus that many of our customers are also major supporters of the RISC-V architecture and related technologies. If technologies related to RISC-V continue to evolve and market support for RISC-V increases, our customers may choose to use this free, open source architecture instead of our products.
In addition, it should be noted that nearly a quarter of Arm’s revenue comes from the Chinese market and it relies heavily on Arm China. However, ARM does not have any direct management rights over ARM China, nor does it have a control seat on the board of directors of ARM China.
Therefore, if ARM wants to maintain its position, it must continue to invest in developing new architectures. Arm also admitted that research and development is the core of its business and the key to its future success.
However, this is easier said than done, and further innovation will become increasingly difficult and costly to achieve. Developing new products requires investing a lot of money and time, generally requiring five or more years of development, and there is no guarantee that the expected results will be produced.
It can be seen that the road ahead of Arm is not a smooth one.
It’s more expensive than Nvidia, is it worth it?
In fact, this is not the first time it has been listed. From 1998 to 2016, Arm was listed on the London Stock Exchange and Nasdaq, and was later privatized by the controlling shareholder SoftBank Group for US$32 billion in 2016.
When the acquisition agreement was reached, SoftBank founder Masayoshi Son was very excited. He said that Arm would be SoftBank’s future.
Seven years have passed. Due to the current poor investment returns of the SoftBank Vision Fund, investment losses in 2022 alone will exceed US$39 billion. In addition, the plan to sell Arm failed. Masayoshi Son needs to raise funds through the Arm IPO and at the same time prove his investment vision.
Based on the upper limit of US$51 per share, Arm is valued at US$54.5 billion, significantly higher than SoftBank’s initial acquisition price. This may become Son’s biggest earnings.
If priced at the midpoint of the IPO guidance price range, ARM's price-to-sales ratio is about 18 times, while only Nvidia's price-to-sales ratio among the components of the Philadelphia Semiconductor Index is higher than this level.
However, many analysts do not buy into SoftBank's optimistic expectations and believe that the valuation of US$50 billion may be too high. ARM is not NVIDIA. From some perspectives, ARM is almost as expensive as NVIDIA. However, the speed and extent of its business growth, revenue and price-to-earnings ratio are difficult to support Masayoshi Son's vision, and it is difficult to support a valuation of US$50 billion.
Investors generally use the valuation of EDA software companies to benchmark Arm. Compared with the valuations investors give to leading companies in the chip design field such as Synopsys and Cadence Design Systems, Arm's current valuation level is much higher. If measured using the price-to-sales ratio standards of these listed companies, Arm's valuation will be between US$32 billion and US$43 billion.
Although analysts believe the valuation is on the high side, market demand is booming and it was 10 times oversubscribed on Tuesday. Arm is expected to go public on September 14 (Thursday) and will face a "test" from the market.