When it comes to control, the only thing that is more powerful than monopolies and oligopolies is market share supported by quantity. Whether from an economic or technical point of view, the essence of the competition in semiconductors and even the entire IT industry is the victory of proportion and quantity.
Looking at the present, there is perhaps no technological force that can compete with ARM in numbers. Despite a successful IPO last fall, Japanese company SoftBank Group still holds 90% of the company's shares. Both the Motorola 68000 and the IBM Motorola PowerPC had the opportunity to dominate the client and embedded businesses, but were quickly defeated by Arm's attack. PowerPC also did its best to take over the data center, but Intel and AMD sprang up with the desktop-based x86 architecture, which pocketed most of the data center gains. As for the smartphone platform, it has been an advantageous battlefield for Arm since its birth, and the X86 has not been able to bite a piece of itself after several attempts.
Today, ARM has established itself in the data center space and has turned a new page in history. This is largely due to Amazon Web Services' enthusiastic adoption of the ARM architecture, first with the Nitro DPU and then with the Gr**Iton server CPU. Since then, the world's largest server and storage complex has gradually come under the control of ARM. Coincidentally, Alibaba Cloud also quickly followed up and launched its own Yitian CPU; Tencent releases Zixiao CPU; Microsoft, on its part, has a Cobalt CPU; Google is also expected to join the ARM server battle with its own "maple" CPU. Fujitsu, Nvidia and Ampere Computing have also come up with their own solutions to meet the needs of HPC and AI user groups through unique designs. In addition, Europe, China, and India** are also supporting ARM-based supercomputer and cloud infrastructure designs. The only exception is Meta Platforms, which seems to be planning to skip Arm and opt for the RISC-V architecture in the first original chip.
ARM may not be the least expensive architecture, but it is arguably the lowest-risk option of any computing architecture available. Wall Street analysts admit that this is also the main reason why ARM first attracted attention. Before introducing RISC-V in computing, there is always a need to evaluate and measure it; But ARM is different, and there are hardly any things that go wrong. Of course, many managers in the RISC-V ecosystem (including Tenstorrent and Sifive) are also gradually building licensing and concession subscription businesses, and like Arm, they are trying to carve out on smartphone platforms, hoping to participate in the data center market competition in the context of the macro recession.
Nearly a decade ago, we discussed various versions of the ARM architecture. Since then, the ARM ecosystem has grown in detail and attention, and has eventually become a semiconductor pillar that can compete with X86.
Everyone should still have the impression that Nvidia once quoted $40 billion to buy Arm or even the entire SoftBank, so it seems that the real value of Arm is as high as $60 billion. But on the face of it, neither the size of the business nor the profitability can support such exaggerated valuation figures. That's right, the ARM architecture is ubiquitous, and the technology highlights are not to be overlooked, and it's serving thousands of engineers and countless users. They have built an excellent RISC architecture and are self-improving by constantly evolving and expanding. But the valuation of 60 billion ......Let's just take a look, the current technology industry is obviously overzealous.
The chart below shows the trend in ARM's revenue from Q1 FY2017 to Q4 FY2024 (the latest data is based on the median revenue guidance released by the company's executives after yesterday**).
If the current trend continues, Arm's annual revenue will soon exceed $4 billion. Arm's prospects are also quite positive, with the ArmV9 architecture, which was unveiled a few years ago and is now being deployed on clients and servers, has already offered twice the royalties of the previous generation ArmV8 architecture. A few days ago, RENE HAAS, CEO of ARM, said in an interview with Bloomberg Broadcast that the ARMV9 architecture accounted for 10% of the company's revenue in the second quarter, and the proportion has grown to 15% in the third quarter. The growth is quite definite.
But don't be overly optimistic: the semiconductor industry, in which Arm is still located, is still in a difficult situation. In the face of the emergence of RISC-V and complete open source, the future fate of ARM is also uncertain. At present, the most reasonable tactic for ARM is to play the fusion solution between X86 and RISC-V, give full play to its easy-to-deploy architectural advantages, and help customers meet various chip needs with sufficient technical depth and rich SoC options. Fortunately, there are already countless companies involved in it, and I believe that ARM can grasp this precious living space.
In short, don't make mistakes, move forward steadily, ARM come on!
Returning to the topic, we continue to talk about ARM's performance and trends over the past eight quarters from the newly disclosed financial documents. From the fourth quarter of 2022 to the third quarter of 2023, ARM scored a total of 56$900 million in revenue, but net profit of only 5$7.7 billion, or about 101%。And judging from the chart above and the financial and market statistics table below, two of them even incurred losses, and the other quarter also had little to no gains.
Earning chip design licenses and royalties is a difficult and unstable business, but the IT world as a whole has never been stable. The current S&P 500 is trading at a price-to-earnings ratio of about 25, which means that the value of each average company** in the S&P 500 is approximately equal to its total profit over 25 years. ARM is not included in the S&P 500 index, but this principle still applies. At the time of writing, ARM's market capitalization grew by a staggering 50% in the first quarter of fiscal 2024, with its valuation rapidly jumping to $79 billion, driven by an upbeat outlook for the last quarter of fiscal 2024. In the past 12 months, ARM has earned a total of $85 million. At this figure, ARM's market valuation should be only 21$300 million. While the overall market situation in 2023 is not good, and technology companies tend to be able to support higher P/E ratios, 21$300 million isn't even enough to cover Arm's revenue over the past year — close to $3 billion.
It can be seen that considering ARM's revenue**, its profitability is obviously insufficient, or sustained profitability is insufficient. Although Arm's revenue has been growing steadily, it is still far from being able to match Nvidia or even AMD in the data center space.
Then let's take a fairer approach. Assuming that ARM generates $200 million in revenue for the fourth quarter of fiscal 2024 ending in March this year, which is $200 million higher than the median reported expected revenue; Let's pick 22 againThe profit ratio of 9% is almost equivalent to the highest earnings figure in the company's history (but not the highest revenue-to-net profit ratio). Even if this assumption is made, the market value of ARM can only be calculated to be $7.1 billion, which means that the market value at the time of ** is even as high as 111 times.
So if the hype bubbles in the United States, Europe and Japan** are removed, the "real" value of ARM should be higher than $7.1 billion, but less than the actual market value of $79 billion. (Sorry, this range is too large for little reference.) ）
But other similar manufacturers are similar. At the time of writing, Nvidia's market cap is $1$73 trillion. In the 12 months to October last year, the full year for which data is available, Nvidia's revenue was 448$700 million, with a net profit of 188$900 million. Based on the price-to-earnings ratio of the S&P 500 index of 25, Nvidia's market capitalization should be around $475 billion. Even if Nvidia's sales are to exceed $60 billion and its profit ratio rises (and is largely driven by the data center business) to about 50%, then Nvidia's market capitalization at a price-to-earnings ratio of only about $760 billion — still equivalent to the S&P average of 23 times. If this comparison, why is ARM's valuation ratio higher than that of NVIDIA?
Let's look at the real numbers. In the third quarter of fiscal 2024, which ended in December last year, ARM reported royalty income of 4700 million US dollars, a year-on-year increase of 106% and 3$5.4 billion, an increase of 184%。Total revenue grew by 138% and came to 8$2.4 billion. ARM is more optimistic about the latest quarter ending March this year, with year-on-year revenue growth at the midpoint of 277% counts as 8$7.4 billion. In contrast, revenue declined sequentially in the same quarter of the prior year, while profit and loss were essentially flat (total sales of 6.).$8.4 billion, net profit of $3 million). It is difficult to predict the net profit for the new quarter, and the final result may be between $100 million and $200 million. The financial data released after Arm's re-listing in September last year was rather limited and not enough to narrow the data range.
But on the positive side, there are also benefits to ARM's high market capitalization: in addition to the 31 that it currently has in handWith $500 million in cash and investments, the company also has financial leverage at its disposal. ARM has the financial means to drive continuous innovation and deliver more for data center, client and edge use cases.
ARM servers are gaining more and more coverage in the data center field, and the market influence is also increasing. As a new data center architecture option, the ARM camp has brought together 15 million developers and a large number of software solutions. In fiscal year 2023, ARM shipped more than 30.6 billion chips, and by the third quarter of fiscal year 2024, the total historical shipments exceeded 280 billion units. It's because the ecosystem is so large that we've always believed that ARM will have a place in the data center.
Arm doesn't specify the percentage of revenue from each customer platform, but we can see this in the following chart presented in the latest financial session:
We already know the specific revenues for these two fiscal years, and we have used a protractor to determine the degrees of each part of the two pie charts. Convert the readings into percentages to calculate the corresponding revenue for each platform, as follows:
One of the most interesting things we found was that cloud and web platforms accounted for 11 percent of revenue in fiscal year 20231%, and we naturally can't wait to find out what the share will be in FY2024.
The core driver of ARM's growth is the continuous improvement of the market coverage of its products. All of this will raise the cap on licensing and royalty revenue, while increasing licensing revenue among your audience**:
The total addressable market for ARM royalties is also growing, as shown in the chart below:
In terms of cloud and networking, which represent data centers, Arm's revenue is only 10% of the total available market (the total available market size here is purely based on the value of the chip, not the royalties that Arm can actually receive). In addition, the approved period of this part of the figures is a natural year, not a fiscal year, and should not be confused with comparison).
Arm said that its licensing and royalty space has continued to grow over time compared to other types of architectures and devices, and its share of total royalty revenue has continued to increase
This gives the ARM business a fairly favorable market position, just as enterprise annuities are to corporate employees. Chip products were designed a long time ago, but are still running in embedded, edge, and client devices, accumulating royalties for companies.
The reason why the ARM architecture regards servers as the forefront of the business is naturally related to this situation. In fiscal year 2023, 46% of Arm's royalties come from server chip designs a decade or more ago, which has also become the ballast stone for the long-term stability of Arm's business.
One of the main trends is that more new companies are moving to the ARM architecture and the licensing system is gradually shifting from flexible access to total access, which in turn generates more revenue.
Even if the customer has acquired all of the intellectual property rights from ARM to accelerate their designs and no longer rely on the Total Access license, the license will still provide ARM customers with almost all of the original intellectual property (except for the architecture license), ensuring that customers continue to pay royalty streams for devices developed in-house or sold externally. With the move to ARMV9, the royalty ratio has also doubled (ARM is also selling full architecture licenses at the same time).
All of this will provide the ARM architecture with great stability, allowing more customers to create compute engines with ARM on servers, storage, or switches in the data center. ARM will also be well-funded to continue to maintain its lead over the RISC-V camp in terms of innovation. The only regrettable thing may be that Masayoshi Son, the head of SoftBank Group, has once again made a lot of money by speculative operations.