This year's largest IPO reveals more details: ARM plans to raise up to $4.87 billion, with major investors including NVIDIA, Apple, Google, and others.

The news of Arm's listing has always been a major event in the chip industry and even in this year's "stagnant" IPO market.

On Tuesday, September 5th, Arm submitted its IPO application to regulatory authorities. The application document shows that Arm's ADS will be priced between $47 and $51 per share, with a total issuance of 95.5 million ADS. The company plans to raise $4.87 billion in the U.S. IPO, which also means that Arm's valuation will reach as high as $52.3 billion.

According to the document, 10 of Arm's clients, including Apple, Nvidia, AMD, Google, Intel, MediaTek, TSMC, Synopsys, and Cadence Design, have agreed to become cornerstone investors in this issuance and have expressed interest in purchasing up to $735 million worth of ADS.

Arm's previous plan was to raise $8 billion to $10 billion, which is higher than the current fundraising scale. One reason for this may be SoftBank's recent decision to acquire 25% of Arm's shares indirectly held by its Vision Fund. The IPO is expected to be completed next week, and SoftBank will still hold 90.6% of the company's shares.

A total of 28 investment banks, including Barclays, Goldman Sachs, JPMorgan Chase, and Mizuho Financial Group, will serve as joint underwriters for Arm's initial public offering.

According to sources familiar with the matter cited by the media, the joint underwriters of Arm are expected to earn up to $100 million in fees from the listing. Arm plans to pay the underwriters 2% of the fundraising scale, which is between $5 billion and $5.4 billion (including a basic fee of approximately 1.75% and a bonus of 0.25%).

In 2016, SoftBank acquired Arm for approximately $32 billion. Although Arm's valuation is higher than the acquisition price, it still falls short of Masayoshi Son's target. When SoftBank recently acquired Arm's shares indirectly held by its Vision Fund, the valuation of Arm was $64 billion.

In addition, Arm may still adjust the fundraising scale based on the demands of investors during the roadshow. Arm is considering pricing its shares on September 13th and starting trading the next day.

Although it did not meet Masayoshi Son's standards, Arm is still the "hot cake" in the chip industry this year and will become the largest IPO globally. The target valuation also indicates that the market is quite optimistic about Arm's prospects.

James Anderson, one of the most famous technology investors in the UK, said that Arm has "missed many opportunities" since its acquisition by SoftBank, such as faster growth in the cloud computing field. He believes that although Arm has a strong presence in the smartphone market, the valuation is "more challenging than they imagined":

It is not yet clear whether Arm will play a key role in the expanding era of artificial intelligence, but I don't think it has a unique advantage in AI development.

Arm's Special Position in the Industry Chain

As a major player in the chip industry, Arm has equipped 99% of smartphones worldwide with its own chip architecture. However, Arm's brand awareness among consumers is not high, and its profits are not as optimistic as one might imagine.

Arm operates in the upstream of the chip industry, primarily by designing IP, including instruction set architecture, microprocessors, graphics cores, and interconnect architectures, and then licensing them to chip manufacturers to facilitate their chip design process.

For the fiscal years 2021-2023, Arm's annual revenues were $2.027 billion, $2.703 billion, and $2.679 billion, respectively. Its net profits were $388 million, $549 million, and $524 million, respectively.

In terms of revenue composition, Arm has two main sources of income: licensing fees (accounting for 40% of total revenue) and patent royalties (accounting for 60% of total revenue).

Analysts explain that the licensing fee model means that other companies using Arm's architecture to design chips need to pay a licensing fee, paying for each use. On the other hand, the patent royalty model involves chip manufacturers using Arm's architecture and Arm receiving a certain percentage of the chip's sales.

Among them, patent royalties are Arm's most profitable business, but the pricing is not high. This is an important reason why Arm's market share is as high as 99%, but its revenue and profit scale are not particularly large.

In 2022, the total value of chips using Arm's technology reached $98.9 billion, accounting for nearly half of the market share. However, Arm's profit was only $1.68 billion, which is 1.7% of the chip value.

Some analysts point out that the issue lies in Arm's charging target, which is the chip rather than the end-user smartphone. In comparison to other high-tech companies, when they license their patents to hardware companies, they charge fees based on the end-user products, which are generally much higher than Arm's fees.

For example, Qualcomm's proposed royalty rate is: for a 3G, 4G, 5G multimode phone that uses Qualcomm's standard essential patents, the royalty fee is 3.25% of the phone's original price. In other words, for a phone priced at 3000 yuan per unit, Qualcomm would charge a royalty fee of 97.5 yuan.

Can Arm Become the Next NVIDIA?

According to Arm's prospectus, in the fiscal year 2022, the company had 5,963 full-time employees in North America, Europe, and Asia, with approximately 80% of them focusing on research, design, and technological innovation. The proportion of R&D expenses to total revenue was 42%. Furthermore, Arm's R&D investment intensity has been increasing, with R&D expenses accounting for 37% and 40% of total revenue in 2021 and 2020, respectively. After SoftBank's investment in Arm, he ordered a significant increase in recruitment and allocated more resources to research and development. Sun's bet was that global chip demand would grow rapidly, and Arm would seize this opportunity. However, it now appears that, due to the market's rapid changes in recent years, there is no evidence that Arm has capitalized on the opportunity.

In the latest fiscal quarter ending on June 30, Arm's quarterly revenue decreased by 2.5% YoY to $675 million, and net profit decreased by over 50% from $225 million in the previous fiscal year to $105 million.

Since 2016, Arm's revenue has grown by 65%, slightly higher than the overall chip industry but far behind the industry leaders. The significant increase in research expenses has not yet translated into the profit growth anticipated by Sun. The bet on the Internet of Things has also failed. The vision of a future where connected refrigerators, doorbells, and other gadgets are ubiquitous has not materialized.

Arm still positions itself as a potential beneficiary of the artificial intelligence boom, stating that the growth of AI systems such as autonomous vehicles may lead to an increased demand for chips designed by Arm. The prospectus for Arm's IPO states, "Arm will be at the core of the transformation brought about by artificial intelligence."

Rene Haas, who took over as Arm's CEO last year, is also focusing on more advanced computing, particularly chips for data centers and AI applications:

We believe that if we push investments into certain instructions, such as SME and SVE - which are vector extensions for specific workloads on exascale processors - we can make a big impact in the exascale server market.

In a podcast, Haas said that in terms of market share based on shipments, Arm has the highest share in the GPU market, but "we will still stick to 'performance per watt is important' and not take the risk of making GPUs that consume hundreds of watts (similar to Nvidia's products)."

Arm's revenue for the most recent fiscal year was $2.68 billion, with a net profit of $524 million. Its historical P/E ratio ranges from 95 to 105, lower than Nvidia's historical P/E ratio of 117. However, compared to other chip manufacturers, Arm's target valuation is still significantly higher, and these chip manufacturers also have a large exposure to the smartphone business. For example, Qualcomm's historical P/E ratio is only 15.