Track Hyper | Nervous Son Masayoshi: Arm Lands on Nasdaq

Arm, almost all global smartphone SoC chip architecture suppliers, recently "promoted" to the largest IPO of the year.

Arm has completed its second listing and successfully landed on the Nasdaq, raising $4.87 billion, making it the fundraising champion since the beginning of 2023.

On the first day of trading, Arm's stock price rose by as much as 25%, with an IPO price of $51 per share and a closing price of $63.59, resulting in a market capitalization of $65 billion (approximately RMB 470 billion).

Based on Arm's latest annual net profit, the IPO price-to-earnings ratio reached 104 times, almost equaling Nvidia's static price-to-earnings ratio of 108 times.

It is worth noting that Masayoshi Son, the founder and CEO of SoftBank Group Corp., still feels uneasy. Why is that?

Profit or loss? An embarrassing uncertainty

Arm Holdings, a subsidiary of SoftBank, has successfully landed on the Nasdaq, returning to the public market after a seven-year hiatus. SoftBank, which holds 90.6% of the shares, seems to have made a net profit of $4.87 billion (based on a valuation of $54.5 billion).

In 2016, SoftBank Group Corp., the parent company of Arm, privatized Arm for $32 billion. Four years later, Masayoshi Son wanted to cash in some of his shares. At that time, Son wanted to sell Arm to Nvidia for $40 billion. However, due to regulatory restrictions, the deal fell through.

Faced with this setback, Son turned his attention to an IPO. There were also many obstacles during this period, such as conflicting interests with the UK government. The UK wanted to keep the meat on its own plate and sought an Arm listing in London. In the end, it landed on the Nasdaq, and the valuation actually shrank from $64 billion to $54.5 billion. This has put Masayoshi Son in an awkward position of uncertain profit or loss.

According to Arm CFO Jason Child, Masayoshi Son doesn't care about Arm's current valuation level. He firmly believes in Arm's future value and "cares more about the future price of Arm." Indeed, the current price of Arm does not give Son peace of mind because the profit or loss is uncertain.

Regarding this point, the opinions of some senior bankers can be taken as a reference.

These senior bankers believe that the company with the closest valuation to Arm is Cadence Design Systems, a circuit design company, with a price-to-earnings ratio of 35 times. Based on Arm's price of $51 per share, the company's price-to-earnings ratio is 29 times. From this perspective, Arm still has room for price growth, but not too much.

Currently, Arm is the supplier with an absolute ecological advantage in the main chip architecture for smartphones and PCs. Its mobile SoC chip architecture design provides technical support to almost all smartphone manufacturers in the Apple and Android camps.

Arm's main source of profit comes from patent licensing fees for its Instruction Set Architecture (ISA).

ISA is the interface between software and hardware, especially microprocessors. The reason why chips can execute specific instructions is because of the existence of ISA libraries. Arm ISA is the most widely used instruction set globally, competing with x86 (a complex instruction set architecture developed by Intel and later adopted by AMD) for PC and server CPUs. Now, chip design giants like Qualcomm are also strongly supporting the new RISC-V instruction set architecture, which directly competes with the Arm ISA.

Although the RISC-V instruction set architecture has developed rapidly due to its open-source nature, the lack of centralized software tools means that the richness and consistency of this new instruction set architecture still cannot match the Arm ISA. Therefore, the moat of the Arm ISA architecture still has a strong competitive advantage.

As for the main terminal required for the mobile Internet - the main mobile SoC chip (CPU) of smartphones, even with the help of the Arm instruction set, only a few companies worldwide (such as Qualcomm, MediaTek, Spreadtrum, and a well-known but undisclosed famous chip design company) can independently design it. The renowned Apple company has also been working for years to build independent CPU cores through the Arm architectural license agreement (ALA).

In addition to the ISA and ALA, Arm also has a patent licensing fee called TLA, which stands for Technology License Agreements.

TLA is a "tool" for customizing CPU cores, allowing buyers to purchase "ready-made" CPU core designs. This is slightly different from the standard Arm Cortex cores used in MediaTek's smartphone mobile SoCs; the CPU cores of Qualcomm Snapdragon SoCs (Kryo) contain some customization.

Under the same conditions, TLA is more expensive than ALA, and the royalty fees are higher. This is because TLA can provide higher added value and reduce the massive design work for chip design companies. In other words, TLA directly provides chip designs that meet specific personalized requirements for chip design companies.

For example, Xiaomi, Vivo, and OPPO, including OPPO, can complete the customization design of dedicated chips (ASICs) through Arm TLA technology licensing. This benefits both parties: as mobile phone manufacturers (Xiaomi/OPPO/Vivo), they benefit from Arm's rich IP and have more choices to meet specific requirements, reducing development costs and shortening time to market. For Arm, it means higher profit margins.

Concern: Arm's Role in the AI Wave

Although purchasing the right to use the Arm ISA architecture or obtaining the ALA architectural license agreement and TLA technology license agreement can bring convenience to chip design companies or smart terminal companies, relying heavily on the Arm instruction set architecture or technology licensing means enduring Arm's dominant style of making demands.

For example, Qualcomm, a major customer of Arm, was originally a supporter of Arm's ready-made CPU cores. However, after acquiring Nuvia, a startup company that designs fully customized IDC (data center) CPUs based on the Arm Phoenix core, Qualcomm plans to use the Phoenix core in future Snapdragon APs. This could result in Arm not receiving patent licensing fees from Qualcomm for using its ISA instruction set architecture. Currently, the two parties are in litigation. However, there are not many companies capable of doing this. Designing a CPU is extremely difficult, to the point where the cost and effectiveness of developing a self-designed CPU core are daunting or discouraging.

In fact, Arm could also design its own chips, focusing on IDC CPUs, supplying Marvell, Google, Amazon, Microsoft, and Meta. Since Arm owns the CPU instruction set architecture, it is not difficult to design chips based on its own technology instruction set architecture and outsource production to foundries such as TSMC.

In addition to its traditional business and the potential new business brought by designing its own IDC CPU chips, Arm's prospectus repeatedly emphasizes the important role it will play in the rapid development of AI and machine learning (ML).

Clearly, Masayoshi Son's shaping of Arm's future value is closely related to AI. In fact, investors are also concerned about the extent of Arm's involvement in this AI boom.

In its prospectus, Arm states, "Every smartphone can run AI applications, and each generation of Arm processors has designed key parts of the acceleration algorithm."

To be honest, Arm's relationship with AI is not so close, but it is not completely unrelated either. Nvidia, the biggest beneficiary of AI, needs to match its chips with Arm's energy-efficient CPUs, and CPU design relies on Arm's instruction set architecture.

Currently, the most needed processor for generative AI is the GPU, while Arm provides the instruction set architecture for CPUs. Therefore, the relationship between the two is not significant. Therefore, Arm's prospectus can only discuss the match between CPUs and GPUs from that perspective: CPUs are crucial in all AI systems, whether they fully handle AI workloads or work in conjunction with co-processors such as GPUs or NPUs.

What could be the reason behind Masayoshi Son's efforts to "ride" the AI wave?

SoftBank Group previously acquired a 25% stake in Arm from the Vision Fund for $64 billion. This deal means that if Arm's valuation does not exceed $64 billion at the time of listing, SoftBank will bear the loss. Currently, Arm is valued at $54.5 billion, still $9.5 billion short of $64 billion. Therefore, unless the stock price continues to rise in the future, it is difficult to determine whether Masayoshi Son will make a profit or loss from Arm's listing.

Therefore, "riding" the AI wave and convincing market investors of this has become Masayoshi Son's focus.

Furthermore, Arm also needs to increase its revenue scale. Based on the valuation of $64 billion, Arm's revenue needs to exceed $1.7 billion by 2023. It seems inevitable that the patent fees for CPU design instruction set architecture in mobile SoCs will increase from the current 3% to 5% for the latest high-end chips.

As the largest source of Arm's revenue, the market for smartphone chip designers and terminal manufacturers is still relatively sluggish. Although there have been signs of a slight rebound in the supply chain in the past two months, it is still difficult to confirm the strength and sustainability of this rebound. Therefore, it is difficult to say whether Masayoshi Son will truly profit from Arm's successful IPO. This still depends on the speed of recovery in the smartphone market. So, whether investors can recognize Arm's core position in this wave of AI and become the focus of Masayoshi Son's concern remains to be seen.