ARM: Market share in cloud services and automotive will continue to grow (FY25Q2 conference call)

ARM (ARM.O) released its fiscal year 2025 second quarter report (ending September 2024) after U.S. stock market hours on November 7, 2024, Beijing time. The key points are as follows:

The following is a summary of the ARM fiscal year 2025 second quarter earnings call. For the financial report interpretation, please refer to "ARM: AI Faith, Can It Support a Hundredfold Valuation?"

1. $Arm(ARM.US) Core Information Review of the Financial Report:

2. Detailed Content of ARM's Earnings Call

2.1 Key Information from Executives:

1) Business Progress

Overall Performance

Last quarter's performance exceeded the upper end of guidance.

The rising demand for AI has driven the demand for Arm's computing platform, with over 300 billion Arm chips shipped to date.

Royalty Revenue

Royalty revenue reached a record high, increasing by 23% year-on-year.

Version 9 architecture: Revenue share reached 25%, a significant increase from 10% in the same period last year.

Smartphone sector: Royalty revenue grew by 40%.

CSS Licensing: The number of CSS licenses doubled in the past year.

AI Applications and Collaborations

NVIDIA Grace Blackwell chip shipments, integrating NVIDIA GPU Blackwell and Arm CPU.

Microsoft Azure Cobalt and Google GCP Axion based on Armv9 data center services have been fully launched.

Collaborated with Meta to optimize Llama 3.2, using the Arm Kleidi library for faster on-device AI processing.

In the automotive market, CSS applications in ADAS (Advanced Driver Assistance Systems) and IVI (In-Vehicle Infotainment) are in strong demand.

There is a very high demand for v9 CPU acceleration in edge AI products.

Software Ecosystem

Arm has the world's largest hardware-supported software ecosystem, with over 20 million software developers globally.

Collaborated with GitHub to integrate Arm tools into GitHub CoPilot, further supporting developers.

Financial Performance

Total Revenue

Total revenue for Q2 was $844 million, exceeding the upper limit of guidance.

Royalty Revenue

Q2 royalty revenue reached $514 million, a year-on-year increase of 23%.

Smartphone royalty revenue increased by approximately 40% year-on-year, significantly exceeding the mid-single-digit growth in smartphone shipments, mainly due to the increased share of Armv9-based processors and higher royalty rates.

Market share in the automotive and cloud service provider sectors continues to grow, but the industrial sector is showing weakness due to inventory adjustments in the semiconductor industry.

Licensing Revenue

Q2 licensing revenue was $330 million, a year-on-year decrease of 15%, better than the expected 25% decline.

Annualized Contract Value (ACV) increased by 13% year-on-year, consistent with recent quarters.

Remaining Performance Obligations (RPO) increased by 10% quarter-on-quarter, some of which will be recognized as revenue later this year.

Q3 and Fiscal Year Guidance:

Q3 Guidance: Expected revenue between $920 million and $970 million, with a median increase of 15%; non-GAAP operating expenses expected to be approximately $525 million; non-GAAP earnings per share (EPS) expected to be between $0.32 and $0.36.

Full Year Guidance: Expected revenue between $3.8 billion and $4.1 billion, year-on-year growth of 18% to 27%.

- Full year royalty revenue is expected to increase by high double digits.

- Smartphone revenue growth will continue to be driven by Arm v9 chips and CSS promotions.

- Market share in the cloud services and automotive sectors is expected to continue to grow.

Networking: Q3 and Q4 are expected to achieve quarter-on-quarter growth.

Internet of Things (IoT): Recovery is expected next year.

The fluctuation range for full-year revenue guidance remains at plus or minus $150 million, mainly due to several large licensing transactions in the cloud sector, although the timing of the transactions and revenue recognition methods are still unclear, all are expected to be completed.

Full-year non-GAAP operating expenses are expected to be approximately $2.05 billion, a year-on-year increase of 19%, consistent with previous guidance.

Full-year non-GAAP EPS is expected to be between $1.45 and $1.65.

2.2 Q&A Analyst

Q: Regarding the relationship and litigation situation between ARM and Qualcomm, can you clarify the recent media reports? The reports state that ARM has canceled Qualcomm's architecture license, possibly related to Qualcomm's product release based on Nuvia's design and the upcoming court hearings.

Regarding the financial data, will the above situation impact future revenue recognition and operating costs? Is there a need to reduce revenue recognition, or will increased legal fees lead to higher operating costs?

A: Regarding the litigation, I can share limited information as the case is still ongoing. Qualcomm needs to obtain ARM's contractual consent to transfer the Nuvia license, but has not received that consent, thus constituting a breach of contract. We have sent Qualcomm a notice regarding the cancellation of the architecture license, but that license has not yet been actually canceled. Obtaining transfer consent is a fundamental requirement in our licensing agreement, and we must ensure fairness and protect the Arm ecosystem that relies on these licensing agreements

Regarding the financial impact, ARM's financial forecasts and guidance have already taken into account the possibility of not prevailing in this case, so there are currently no additional adjustments. As for expenses, there are no changes in revenue recognition or operating costs, and fees will continue to be charged at the existing ALA royalty rate.

Q: You mentioned that cooperation with CSS is quite active. Although it was originally expected that this year's order volume would be lower than last year, given the strong demand for customer design activities and the visibility of renewals throughout the year, can the team continue to increase backlog orders in this fiscal year?

A: We have observed that the demand for Arm technology in the market is very strong, particularly due to the improvement of the ecosystem and the overall growth in demand for Arm technology. The application of AI is not limited to training in data centers but also includes inference in data centers, networks, automotive, PCs, mobile phones, and wearable devices (i.e., edge devices). The demand for computing resources in the market is rising to support the operation of small or large language models on top of existing computing needs. This has driven a broad demand for R&D and innovation across all markets and industries, allowing us to seize this platform opportunity.

Additionally, our number of CSS licenses has doubled this year, exceeding expectations. For example, MediaTek's first CSS chip design is a typical case. Overall, the demand for computing, AI, and the promotion of CSS have collectively driven the growth in licensing demand. Jason also mentioned that based on the strong growth we have observed, although there may not be a significant increase in backlog orders, we have raised our licensing revenue target for this year by 45%, far exceeding expectations at the time of the IPO. The main growth driver is the increase in royalties, which achieved a 23% year-on-year growth this quarter, which is very encouraging.

Q: Regarding the licensing business. You mentioned signing six new ATAs, and the number of CSS licenses has also doubled. Initially, you mentioned that there were about 50 potential customers for ATA, and now it seems that this number is increasing, as there are already over 40 licenses. How do you view the growth potential in this area in the future? Will there be a slowdown in growth as licensing shifts to royalty income?

A: Through the ATA (Arm Total Access) licensing agreement, we have the opportunity to expand the breadth and scale of the ARM technology portfolio. ATA grants customers broad access to ARM technology, including a large IP portfolio and annual chip production and tape-out rights, allowing for adjustments to the licensing scope based on demand. As a result, customers can obtain more value than with a one-time license.

Theoretically, most customers can transition to some form of ATA licensing. The main reasons customers prefer ATA are: first, they are certain they will use ARM technology; second, engineers can easily access a wide range of IP for experimentation and evaluation; third, they can lock in R&D costs, as they will continue to purchase these IPs in the future.

For ARM, ATA means more stable revenue, as licensing is more repeatable, and a broader IP portfolio typically increases engineer usage. Therefore, most customers will ultimately choose some form of ATA licensing, and this broad IP portfolio will also drive higher royalty income in the future

We previously stated that approximately 80% of customers may be suitable for ATA authorization, and recently this proportion has exceeded 50%. The annual fee or price increase for ATA authorization is about 7%, with an average contract duration of around three years. Therefore, even if all customers adopt ATA, there will still be renewals and annual fee growth. Additionally, with the introduction of new technologies and products, the authorization can be further expanded. Thus, there is still significant room before TAM saturation, and there are multiple growth points in the future to expand TAM.

Q: Regarding royalty income, you have raised this year's growth target to a high double-digit figure. But could you elaborate on why the v9 conversion rate and proportion have stagnated this quarter? What adjustments need to be made for overall royalty growth to address this issue?

A: Regarding the transition to the v9 architecture, the adoption has been very promising, especially with rapid growth on mobile. All Neoverse series products are based on v9, and v9 is gradually being applied in the automotive and IoT sectors. Additionally, Apple’s first batch of products based on v9 has begun shipping. Although growth may not be linear between quarters, royalties will continue to grow over multiple quarters.

Unlike the v7 and v8 versions, the royalty growth of v9 throughout its lifecycle will be more dynamic. First, the v9 architecture will continuously upgrade in terms of technology and product performance; for example, smartphones produced in 2025 and 2026 can achieve a 15% improvement in performance and power consumption, thus enabling higher value pricing. Second, v9 introduces CSS (Customer Customized Solutions), with royalty rates higher than the standard v9 core, even exceeding twice the standard rate. Therefore, as the adoption rate of v9 increases, royalty income growth will be significantly higher than before, primarily driven by value pricing and the transition to CSS.

Q: The networking and data center business has accounted for about 10% of your business portfolio over the past 12 months. What is the importance of this segment as a driver for royalties and licensing in the next 12 months?

A: From a macro perspective, the adoption of Neoverse in networking and data centers will bring Arm a very strong growth trajectory, not only in the next 12 months but also for several years to come. Specifically, in the general computing field, Graviton has been in use for several years, while Microsoft Azure's Cobalt and Google GCP's Axion have now officially launched, allowing users to purchase instances directly. Based on the application of Graviton, we are very optimistic about the development prospects of Azure and GCP, especially since these chips improve power consumption and efficiency by 50%-60% compared to X86.

Additionally, NVIDIA's Grace Blackwell advanced training and inference chips are also on the rise, and these chips are part of the ARM CPU solution. The advantage of this solution is that general computing and AI training or inference centers in data centers can share a large amount of foundational OS and workload software, thereby reducing total cost of ownership (TCO), which we are very much looking forward to

Regarding the data infrastructure and data center market, network growth is expected to remain slow, but the data center, especially the cloud computing market, is performing strongly. With the further deployment of Cobalt and Axion, as well as the promotion of customized silicon chips by other manufacturers, we expect the data center market to accelerate further in the second half of the year.

Q: What specific factors contributed to the better-than-expected performance in licensing revenue? Additionally, how is this related to the unfulfilled performance obligation (RPO) metric? How should we view growth for the remainder of this year? Do you still believe the $1.7 billion target can be achieved? Is it possible to pull some licensing deals into this quarter? Thank you.

A: As for the part that exceeded expectations, our revenue was about $35 million higher than anticipated, while we originally expected a 25% decline, but the actual decline was 15%. This is mainly due to a few very large deals from the same period last year, which created a high comparison base. Regarding the year-over-year decline in RPO, we set a historical record last year, and this quarter is the second highest level ever, with a 10% quarter-over-quarter increase.

Overall, this is quite a good growth. Earlier this year, we effectively amortized the RPO portion of the large deals signed last year into revenue. Therefore, overall, the RPO trend is very healthy, and the year-end forecast range remains reasonable. Apart from the deals we expect to complete in the fourth quarter, we do not anticipate any significant deals at this time. If circumstances change, we will inform you promptly.

Q: How do you view the adoption trend of CSS in smartphones, and how will it develop from this year to next? Is it possible for CSS to capture 50% market share in the coming years?

A: The value proposition of CSS mainly lies in shortening time-to-market and enhancing confidence in design and performance. The smartphone market is characterized by very tight product cycles and annual updates. Product release times often align with major events (such as MWC or China's Double Eleven), so manufacturing cycles have also lengthened under increasingly complex application processor designs and more advanced manufacturing processes. Any technology that can shorten chip design time is welcomed by the market.

In the early stages of CSS introduction, the market was skeptical about its applicability in the smartphone market, but actual adoption has been very positive. In summary, I believe CSS is likely to capture 50% market share because it can go into production immediately after development is completed, meeting the demands of this competitive market.

Q: The smartphone portfolio in the Android market is also shifting towards this entity. Have you observed any substantial impact on your company's licensing revenue in China? Additionally, what is the current penetration rate of v9 in China?

A: We have indeed observed strong growth in overall smartphone shipments in the Chinese market, especially significant growth from local brands like Xiaomi, OPPO, and Vivo. Our team also attended OPPO's new product launch last month. Firstly, the Chinese market is growing strongly, and consumer preference for local brands is evident. In the high-end flagship market, almost all models will adopt the v9 architecture in the next year or so, and v9 will gradually enter the mid-range market Therefore, the Android market in China is performing well, and the vast majority of products are transitioning to v9, with high-end models fully adopting v9.

Q: Should we expect the proportion of the v9 combination to continue to grow? If not, what factors might hinder growth? Macroeconomic factors? AI factors?

A: The growth of v9 will be quite predictable. This time last year, the proportion of v9 was 10%, and now it is 25%. It can be confirmed that a year from now, this proportion will be far higher than 25%. Factors driving growth include:

PC market: As more Windows and Arm products are launched, more manufacturers will introduce v9 chips;

Mobile market: A typical product penetration model, where flagship features gradually spread to mid-range models, and mid-range features then trickle down to low-end models. A year from now, it is expected that most smartphones will be equipped with v9;

Data centers: All data center products use v9;

The update in the automotive sector will be relatively slow, mainly limited by the introduction time of IVI (In-Vehicle Infotainment) and ADAS (Advanced Driver Assistance Systems) products, but progress in this area is positive.

Quarterly, the growth of v9 has stabilized recently, due to the rapid growth in the mid-range smartphone market, which has not yet popularized v9. However, this is only a temporary phenomenon, and the long-term trend remains upward. We have a clear vision for product releases in the next 3 to 5 years, and collaboration with chip manufacturers and OEMs ensures visibility in the release schedule. Therefore, there is no need to focus too much on quarterly fluctuations; the overall trend is continuous growth.

Q: Since the beginning of this year, the AI PC market has not performed as expected, and AI has not driven the upgrade cycle of smartphones. When do you expect the shipment volumes of PCs and smartphones to grow? 2025 or 2026? Additionally, you previously mentioned a target of 50% market share in the PC market; is this still your long-term vision?

A: Regarding the 50% market share target for the PC market, I still stand by this prediction. For the Windows ecosystem, the key is to have a comprehensive SKU product line. Currently, various types of devices can be seen in retail stores, such as ultrabooks, low-end models, mid-range models, gaming laptops, etc. I believe we will see Arm architecture devices filling these SKU gaps in multiple areas. For example, when entry-level ultralight devices based on Arm or Arm-based gaming laptops are launched, it will drive market growth. We are very confident about the growth of the PC market in the coming years.

As for whether AI demand will drive this growth, I think it is important that these PCs have the computing power needed to run AI applications. Currently, the value positioning of AIPC is still in the early stages, but there is no need to focus too much on this. Typically, such technology cycles will first have a technological reserve that exceeds demand, and then application demand gradually catches up. Therefore, these devices are largely prepared for the future. Overall, we are optimistic about market demand and firmly believe in the growth trend of Windows on Arm

Risk Disclosure and Disclaimer for this Article: Dolphin Research Disclaimer and General Disclosure