Who is the semiconductor company with the biggest bubble in the world?

Some say it's Nvidia.

However, Nvidia's dynamic valuation is only 60 times, and with its still impeccable GPU products and monopoly market position, it's hard to say that this valuation is ridiculously high. What's more, the future of AI is a sea of stars.

More importantly, Nvidia is fully capable of delivering results, with a gross profit margin of 90%, net profit of 50%, and a net profit of tens of billions of dollars a year, which is not far from Microsoft's level. You can say that Nvidia's valuation is not cheap, but it seems inappropriate to say that it is a bubble.

If you compare it to another semiconductor company, Nvidia's valuation can be said to be woefully low.

Because of that semiconductor company's dynamic PE, it's 10 times that of Nvidia. In less than a year, its market capitalization has quadrupled to nearly $200 billion. One year's revenue, at the level of 3 billion US dollars, net profit is only about 0.3 billion US dollars. Calculated, PE is close to 600 times.

This company is called ARM.

01

Towards a monopoly?

Whether a business makes money or not, the business model is a decisive factor.

For example, a monopolistic business model can generate amazing profits. Large technology companies such as Nvidia, Microsoft, and Google all belong to the monopoly industry to a greater or lesser extent. At least in their field, they hardly see that competitors can challenge their position.

Although it faces anti-monopoly regulatory risks, if it is because competitors are really unable to challenge it, there is actually no need to worry about this kind of monopoly; it's all about paying fines on a regular basis. This is basically the case with large US technology companies.

ARM's business model is unique. It mainly uses authorized semiconductor companies, original equipment manufacturers (OEMs), etc. to use its architecture to design and manufacture chips, that is, sell IP cores (instruction set+microarchitecture) and collect licensing fees and royalties.

Although there is no direct chip design, the public architecture provided by ARM greatly reduces the threshold for chip development and preserves a certain amount of R&D space. Downstream chip designers can improve the upper limit of chip performance on their own, in addition, it also has the characteristics of high performance and low power consumption. Therefore, many chip companies are using the ARM architecture, including Apple, Qualcomm, and Nvidia.

Currently, 99% of smartphone baseband chips are designed using ARM architecture. Apple is leading this trend.

However, with Apple using the ARM architecture on PC chips and Amazon's IDC server chips, ARM has begun to break out of the smartphone field and launch an impact on the backyards of Intel and AMD, and has already achieved certain results.

At the beginning of last month, at the build conference, Microsoft announced that it would launch more Windows on ARM products, and ARM reached its most glorious moment in history.

In addition, in the AI era, the problem of power shortages has already come to the table, so there are voices that ARM is leading the trend of squeezing Intel and AMD's x86 architectures out of the market and dominating the world with its ability to watch over the market — low power consumption.

However, this statement is too optimistic because ARM is likely not to get to the point of monopoly.

02

An old opponent fights back

Traditional PC chip manufacturers, Intel and AMD are also developing AI PCs.

ARM may still have an advantage in terms of power consumption, but PCs are not mobile phones; PCs can be connected to a power source, so the demand for power consumption is not as high as that of mobile phones. It is difficult to clean up the x86 architecture just because of the low power consumption. In fact, if there is a better foundry process, the power consumption of the x86 architecture may not be completely unmanageable.

In terms of software ecology, ARM is still not comparable to x86, because x86 has been accumulated since the 1980s. However, ARM started with the Apple M1 chip before officially establishing the PC software ecosystem.

ARM will occupy a certain share in the PC field, and this amount is still quite impressive. Counterpoint predicts that the share of ARM-based laptops will increase to 21% in '25 and to 25% in '27.

Currently, the overall sales volume of consumer electronics, such as PCs and mobile phones, is saturated. Although supported by AI, it is only a structural replacement, not a double of the total volume.

ARM's deterministic incremental market actually comes from Windows on ARM, and Microsoft and Qualcomm have already launched corresponding products. Exactly how much it can grow depends on the shipment volume of such AI PCs.

However, it is still very difficult to increase profits several times by relying only on the speed of encroachment as shown above.

If profits can't be doubled quickly, how can ARM convince the market that a valuation of 600 times is not high?

ARM's revenue and profit growth performance over the past few years has had a growth rate, but it's hard to say that it is explosive. How can it prove that it will perform beyond the sky in the future?

Let's take a closer look at ARM's main revenue streams:

The first is license revenue. Customers pay fees to obtain licenses for ARM products and then develop processors based on ARM architecture. According to ARM's official website, its latest billing plan is: chip manufacturers can choose to pay an access fee (access fee) of 0.075 million US dollars or 0.2 million US dollars per year to obtain one or all of ARM's design plans to put into production chips.

Second, there is royalty revenue. When the design of a processor based on an ARM product is completed and production begins, ARM will collect an IP license fee from the customer, and then charge royalties based on the number of chips. Normally, it is 1%-2% of the chip price. The FY23 average rate is 2.7% of the chip price.

Therefore, if ARM wants to increase revenue, there are two ways. One is to harvest more chip design vendors, so there will be more license revenue and IP licensing fees, and the other is for design manufacturers to sell more chips based on the ARM architecture, so they charge more royalties based on the number of chips.

Looking at it now, both paths are possible.

ARM's authorized partners are growing at a rate of 30-40 per year. And with the success of Apple's ARM-based PCs and the addition of Microsoft, it will drive a number of PC manufacturers to adopt ARM architecture-based CPUs.

In other words, more ARM-based PCs will be shipped.

However, old rivals are also struggling to fight back, and victory did not fall on ARM's side.

03

New opponents are catching up

In 2023, a new rival for ARM appeared.

The reason is that ARM is too aggressive in increasing revenue.

In 2022, ARM proposed new regulations. Starting in 2024, if you use an ARM public CPU, you must use an ARM public GPU, NPU, and ISP together. This new regulation essentially requires these customers to choose between existing solutions and ARM. Either abandon ARM and find a new CPU architecture, or replace all self-developed or third-party chips and switch all to the ARM public architecture.

In March 2023, ARM plans to change the licensing model, following Qualcomm's example, changing the pricing basis from chip sales price to terminal sales price. Each phone charges a license fee of 1-2% of its sales price. Although the ratio has not changed, the actual price increase has reached about ten times.

But this has touched the core interests of users, especially the 10-fold increase in price, and no customer can stand this practice.

As a result, Samsung formed a team to develop its own CPU core architecture to prepare to replace ARM chips with self-developed Galaxy chips in mobile phones and laptop products in 2027; Qualcomm has also accelerated the decision to leave ARM and is preparing to use the self-developed Nuvia architecture in the next generation Snapdragon 8 Gen 4 as soon as possible. Nuvia, which was acquired by Qualcomm, has already purchased part of ARM's IP, and both parties are filing lawsuits over whether Nuvia's IP still requires payment of licensing fees to Arm.

The most aggressive response was the establishment of the RISC-V software ecosystem (RISE) based on a streamlined instruction set in June 2023.

Members include Google, Intel, MediaTek, Nvidia, Qualcomm, Samsung, etc. The joint venture was founded in Germany to accelerate the commercialization of future products based on the open source RISC-V architecture. The initial focus of the application will be automotive, but will eventually expand to mobile devices and the Internet of Things.

Unlike ARM, RISC-V has the advantages of being open source and does not belong to any company or country.

Currently, the number of chips shipped by RISC-V has reached 10 billion, which is still very small compared to ARM's cumulative shipment volume of 250 billion chips, but the growth rate is very fast. The RISC-V International Foundation predicts that the RISC-V processor IP penetration rate will increase from 1% in 2022 to 16% in 2027.

Arguably, ARM's aggressiveness has forced out a strong competitor.

With old rivals fighting back and new rivals catching up later, the problems ARM has to face will become more and more difficult.

04

Softbank control panel?

Other than fundamentals, many have discovered a key factor supporting the rise in ARM's stock price.

SoftBank accounts for 88.7% of ARM's shareholding structure. Coupled with the fact that it was only listed in September last year, the current circulation market is very low, accounting for about 12% of the total share capital.

This raised questions from the market; SoftBank is deliberately controlling the market. Because in the financial market, there are many similar cases.

Examples include Evergrande Auto, a Hong Kong stock company, and Vinfast, a Nasdaq electric vehicle company. The experience of these companies is very similar. They all have very high shareholders' shareholders' holdings. After the stock price rose sharply by 10 times in a short period of time, it suddenly plummeted by 90%, or more, and finally left behind a scum.

However, ARM already has mature products, business models, and performance. Unlike these two car companies, they take a car case and say they are a new car builder; they are simply drawing big cakes and cutting chives.

Moreover, SoftBank's Sun Zhengyi is not in the same class as Xu Jiayin. He is the owner of a world-famous technology investment company. They do all serious business; they are by no means deliberate, and there is no need to make a fortune by cutting chives through the capital market.

ARM's sharp rise also has objective factors. For example, AI is too hot, the enthusiasm for capital speculation is too high, and when you see a ticket involving AI, it goes viral.

Therefore, the sharp rise in ARM's stock price is blamed on Sun Zhengyi's deliberate control of the market, which is unbiased. In reality, he doesn't have to do anything; he can just let the market go wild.

In February, there was an opinion that the ban on SoftBank shareholding was lifted, and Sun Zhengyi might sell ARM drastically due to the need to return funds. As a result, there have been no results until now.

The likely reason is that Sun Zhengyi is not as poor as he was two years ago. That being the case, there's no need for a major sell-off, as ARM clearly continues to benefit from the AI boom.

Who can't get along with money at this time?

Furthermore, even if there is a capital requirement, there are many ways for Sun Zhengyi to return funds through ARM holdings, such as pledges, forward contracts, etc., without directly selling in the secondary market, causing stock price fluctuations.

As a veteran of the capital market, Sun Zhengyi knows these principles better than anyone else.

05

epilogue

The US stock market is still very hot. As far as the market is concerned, it is difficult to say that it will collapse in the short term.

Because of several important factors supporting the US stock market, I can't see any advantage. Macroeconomic data is very good, liquidity is also expected to cut interest rates. The profit effect of AI is also strong, and there is a lot of enthusiasm for increasing capital.

In fact, as long as capital is still piling up for AI, ARM will also get a share of the pie, plus the circulation market is relatively small. Even if it continues to rise, with a market capitalization of 250 billion or more, I'm not surprised.

Because bull markets are like this, they have happened countless times in history.

We just want to remind investors that when the valuation is hyped up to 600 times, you shouldn't simply be led by the market's desperate voice to make you go all in, but should just calm down and think about it. Is this really reasonable? Is it possible that there is a huge bubble?

As Barao said, I'm afraid that others are crazy.

It might make you a little less money, but it's also worth it if you can avoid bigger losses in the future. (End of full text)