ARM (Minutes): The company is evaluating the feasibility of entering the 'physical chip' market.

The following is the summary of ARM's FY2026 Q1 earnings call organized by Dolphin Research. For the earnings report interpretation, please refer to the article ARM: 'Chip Tax' Supports the Scene, Orders 'A Bit Cold' Amidst Burning Money?

I. $Arm(ARM.US) Key Financial Information Review

1. Key Business Progress

a. CSS: Completed the first three upgrades of existing customers to the next-generation platform. The next-generation CSS platform offers better value, functionality, and market launch speed, with royalty rates hitting a record high, laying the foundation for further investment confidence.

(Note: CSS, or Compute Subsystem, is a pre-assembled IP module that includes not only Arm CPU cores but also other IP parts, allowing companies to skip the integration phase and bring products to market faster)

b. Licensing Business:

- Annual Contract Value (ACV) increased by 28% year-on-year, far exceeding the recent high growth rate of '10+%' and the long-term expectation of 'mid-to-high single digits' (short-term growth benefits from new deals, long-term expectations remain unchanged).

- Remaining Performance Obligations (RPO) remained flat quarter-on-quarter, as new licensing deals offset the revenue recognized from previous licenses.

- Key transactions driving licensing growth include multi-generation GPU cooperation with leading smartphone manufacturers and expanded IP licensing and design service agreements with SoftBank.

2. External Environment Impact

a. Tariffs and the macro environment have limited direct impact on royalty income and licensing revenue, but the company has limited understanding of their indirect impact on end demand. Continued uncertainty reduces the predictability of short-term royalty income.

b. Licensing business customers continue to invest during short-term economic slowdowns due to the long chip development cycle.

3. 2Q26 Guidance:

a. Revenue is expected to be $1.01-1.11 billion (midpoint growth of about 25% year-on-year), with royalty and licensing revenue remaining flat quarter-on-quarter.

b. Non-GAAP operating expenses are expected to be $655 million (including deferred expenses from Q1 and foreign exchange impacts), and non-GAAP EPS is expected to be $0.29-0.37.

4. Future Outlook:

a. Based on a clear judgment of the customer design pipeline, contract royalty rates, growing demand for custom chips, and 'cloud-to-edge' AI demand, the company is confident in healthy growth over the next few years.

b. Plans to strengthen its advantages through increased R&D investment, support customers, seize opportunities, and ensure efficient AI operation on the Arm architecture.

II. Detailed Content of AMD Earnings Call

2.1 Key Information from Executive Statements

1. Cloud Computing:

a. Over 70,000 enterprises run AI workloads on Arm Neoverse data center chips, a 40% year-on-year increase, and a 14-fold surge since 2021.

b. Arm Neoverse processors support critical global AI infrastructure (such as NVIDIA Grace, Amazon AWS Graviton, etc.).

c. NVIDIA Grace Blackwell's energy efficiency is 25 times that of the previous generation x86 systems, and Arm Neoverse chips are expected to approach a 50% market share in hyperscale data centers this year.

2. Edge and Endpoint: AI extends to the edge, and Arm meets local real-time intelligence needs with energy efficiency and scale advantages. Computing platforms integrate AI acceleration and energy efficiency (such as Ethos U85 neural network processors, V9 processors), and companies like Apple and Samsung integrate them into high-end phones for efficient local AI experiences.

3. Ecosystem: Over 22 million developers worldwide (accounting for more than 80% globally) develop based on Arm, forming a positive cycle of 'more developers → more software → higher platform demand'.

4. CSS Business:

a. The first-generation CSS has served 5 customers, with royalty rates twice that of Armv9.

b. This quarter added 3 new licensing agreements (2 data centers, 1 personal computer), doubling the number of licenses year-on-year. Subsequent platforms (such as the Zena CSS optimized for autonomous driving) will enhance value, functionality, and market launch speed, with royalty rates hitting new highs.

c. Xiaomi XRING 01, Samsung Galaxy Flip 7 (equipped with Exynos 2500) are both based on the latest Arm CSS platform.

d. A mainstream smartphone manufacturer has committed to adopting its GPU platform by 2030, accelerating graphics and AI performance.

5. Future Plans:

a. Expanding cooperation with SoftBank to explore directions such as compute subsystems, chiplets, and complete end solutions.

b. Increasing R&D investment, expanding engineering delivery capabilities, leveraging CSS popularity and AI computing demand to drive strong company growth.

2.2 Q&A

Q: Can you share Arm's strategy in ASIC and complete end solutions?

A: Currently, there is no specific information to announce, but further integration is the company's development direction. Customers (including new customers like CSPs, OEMs, and traditional customers) hope for a better starting point in SoC development due to chip complexity and development time, which has led to the creation of compute subsystems (CSS), whose success has exceeded expectations. Many chiplets currently under development are based on Arm's IP, and we have supported chiplet development through the Arm Total Design ecosystem. Based on this, the company is evaluating the feasibility of expanding beyond existing platforms into more subsystems, chiplets, or launching complete solutions. Internally, we either have in-depth knowledge and technology required for designing, implementing, and manufacturing products like chiplets, or we can directly apply this knowledge and technology.

Given the trend of complex chip development in the market, and Arm being the only platform covering computing from the smallest devices to the largest data centers (from milliwatts to megawatts), the company is uniquely positioned to provide solutions in a unique way and is currently exploring related possibilities in depth.

Q: Royalty growth was 25%, while last quarter you expected 25%-30%. Which end market caused this difference? Additionally, you mentioned that royalty might grow 10%-15% quarter-on-quarter in Q3 and Q4 this year. Is this trend still expected?

A: This quarter's royalty was close to expectations (about 1% deviation), at the lower end of last quarter's range. Smartphone growth was below expectations, although the growth rate was still much faster than the low single-digit market (market growth was slower than expected), but there was no weakness.

For the rest of the year, the company's expectations for royalty are similar to last quarter, with attention needed on the subsequent impact on smartphones.

Q: You mentioned that Arm's chip market share in hyperscale data centers is expected to reach 50% this year. What was this proportion last year? And how do you view the composition of these workloads running internally and externally?

A: Last year, Arm's market share in hyperscale data centers was about 18%, and this year it is expected to approach 50%, driven by two aspects: 1. In the general workload field, products like Graviton, Axion, and Cobalt continue to capture share from x86; 2. In the AI workload field, previous generations relied on NVIDIA Hopper (paired with external x86, Arm's share was nearly 0), now shifting to integrated Arm-designed Grace Blackwell (GB200/300), where Arm almost exclusively holds the share. Combined with AI data center growth and traditional data center share increase, the share is expected to approach 50%.

Q: This quarter's exchange rate impact on EPS was $0.01. You mentioned that next quarter's guidance includes exchange rate impacts but did not quantify it. Can you explain the expected impact of exchange rates on next quarter's EPS, and the company's hedging strategy and future exchange rate impact estimation method?

A: The exchange rate impact on this quarter's EPS was $0.01, and it is expected to be about $0.01 per quarter for the remaining three quarters of this year, totaling about $0.04 for the year. The company's exchange rate hedging started at the beginning of the year, with revenue settled in USD, and about two-thirds of operating expenses in EUR and GBP. Due to some unhedged expenses, the exchange rate impact is slightly larger than before.

Q: Since Neoverse chips will reach a 50% market share this year, it means x86 still holds 50%. Why do customers choose to retain x86? Will Arm's market share eventually stabilize at a level below 100%? How should we view the ultimate market share ceiling for Arm?

A: We are confident in the continued growth of market share beyond 50% for two reasons: 1. Arm architecture offers high customization capabilities, enabling unique designs at the chip, blade server, and rack levels, significantly optimizing total cost of ownership and maximizing performance, which is our unique advantage. 2. As AI data center share increases, adopting a single CPU architecture and unified software stack can simplify data center software management, forming a boost.

Q: You mentioned that ACV grew significantly this quarter, related to GPU transactions and additional licensing from SoftBank. Can you explain the extent of these two factors' impact on ACV growth? Additionally, the press release mentioned a GPU transaction involving a major smartphone OEM. Can you further explain this situation and the background of these two factors?

A: This quarter's ACV growth mainly came from three major CSS transactions (customers upgrading to the next-generation platform) and deepened cooperation with SoftBank: excluding other regional transactions, ACV growth was about 17%; with SoftBank cooperation, it reached 28%. The incremental licensing revenue from CSS transactions is partially included in this quarter and partially in RPO. Overall, global customers and SoftBank cooperation jointly drove ACV accelerated growth.

Q: You mentioned that the royalty rate for the second-generation CSS will be higher than the previous generation. Can you explain the extent of this rate change?

A: The royalty rate for new CSS transactions exceeds 10%, higher than the approximately 10% of the first-generation CSS.

Historical rate references: V8 was 2.5%-3%, B9 about 5% (double that of V8), first-generation CSS about 10% (double that of B9). Previously thought 10% was close to the ceiling, now confirmed there is still room for improvement, with the specific proportion to be clarified later, but indeed higher than expected, which is positive.

Q: SoftBank expanded licensing transactions. Can you explain the situation of this transaction (whether it targets next year)? In accelerating this cooperation, will the company benefit from ASP or only from licensing transactions?

A: The Stargate project, jointly invested by SoftBank and OpenAI, plans to expand computing power to 10GW in the coming years, involving a large amount of computing demand and design opportunities. All related opportunities in this project are based on Arm as the core CPU, and Arm will provide solutions for it, with related work already started. Specific product and timeline details cannot be provided at this time.

Q: You mentioned that Arm's market share will reach 50%, and currently, there are major customers like Google, Amazon, and Microsoft, and possibly another cloud service provider (CSP) joining soon. Will Arm launch commercial CPUs into the market in the future, or only provide licensing platforms for internal use?

A: Unable to respond to whether specific products will be launched. However, market share growth is not significantly related to this strategy. The company is highly confident in the trend of Arm software investment and the roadmap for providing product technology in various forms to increase share.

Q: In recent weeks, hyperscale enterprises have raised their capital expenditure targets for this year and mentioned further growth next year. What impact does this have on the royalty growth of Arm's existing processor ASIC designs this year and next year? What impact does it have on longer-term CSS licensing opportunities?

A: The continuous increase in AI data center capital expenditure strongly drives Arm's technology products and royalty rates.

The second-generation CSS has been launched, and its positive impact on royalty rates is beginning to show; with the implementation of higher-rate new-generation CSS in the future, royalty growth is expected to further expand.

Current related demand has not decreased, as AI will cover various industries and enterprise applications are still in the early stages. Computing power and model performance promote each other, and Arm, being at the core of computing, is the preferred platform in the ecosystem involved in model evolution, with clear long-term benefits.

Q: Can you talk about the situation of Arm's business in China? Especially the GPU export control in the past month, will it make Arm China's contribution in the data center field more significant than expected three months ago?

A: Arm's market in China is in sync with the global market, with development momentum in smartphones, autonomous driving, data centers, etc., consistent with the global trend, as China highly relies on the Western software ecosystem based on Arm.

The GPU export control in the past month (especially related to H20) has no significant impact on Arm's business.

Arm China's business growth is stable, with Q1 revenue accounting for 21%, up from 15% in the previous quarter and 14% in the same period last year, consistent with the global growth trend.

Q: Regarding the adoption of Armv9, it was about 30% last quarter. What is the current proportion?

A: Armv9 adoption rate was slightly over 30% at the end of last year, and subsequent updates will be annual, with the next data to be announced at the end of the year. This quarter's royalty growth was 25% year-on-year (18% last quarter), indicating the continued growth of v9 and CSS proportions.

We no longer frequently disclose specific proportions, as it may cause misunderstandings: v9 has iterated to the fourth generation (such as some mobile phone products), with each generation's rate higher than the previous one and updated annually, its rate growth faster than v9's adoption rate, and the two are not directly corresponding, needing to be viewed separately.

Q: Its edge capabilities are outstanding. Is it necessary to improve performance to exceed 4 TOPS? What are the long-term applicable scenarios for such MPUs, and can they be used in cloud environments to achieve the required performance levels?

A: Although the roadmap has not been announced, performance is expected to exceed 4 TOPS, not a data center product, focusing on low-power, small-size scenarios (such as wearable AI acceleration devices), with broad application space and considerable long-term potential.

Q: The first CSS transaction has been completed, with broad application opportunities in the automotive field. Is it positioned for autonomous driving or more towards GenEv and ADAS? When will the growth of the licensing business start, and will it be achieved this year?

A: Customers have already signed up, and the automotive market is an important application scenario, helping customers lacking SoC experience accelerate time-to-market. Mainly targeting L2 - L4 level ADAS applications, handling control and data layer work related to autonomous driving. With the integration of IVI and ADAS systems, the investment in Arm-based software is significant, and the launch of Zena is logical and progressing ahead of schedule, with strong momentum.

Q: What is the timeline for subsequent licensing transactions?

A: We have a very strong transaction reserve in this area and will announce relevant information at the appropriate time.

<End Here>

Risk Disclosure and Statement of this Article:Dolphin Research Disclaimer and General Disclosure