Arm: The Silicon Blueprint - [Business Breakdowns, EP.200]

At a Glance

Guest Jay Goldberg (CEO of D2D Advisory) and host Zack Fuss take an in-depth look at Arm's business model, history, and competitive landscape. Amid discussions on strategic inflection points, technological evolution, and market competition, the most consequential takeaway from the entire episode is: Jay Goldberg believes the reasonableness of Arm's current valuation (~$150 billion, 30x price-to-sales) stems not merely from the AI boom, but from its structural expansion from a dominant position in the single mobile phone market into broader markets such as data centers and autos, and its systematic improvement in value capture (moving from roughly $0.07 per chip toward higher levels).

Key Themes

1. Arm's Business Model: "Blueprint" and "Standard Component" Designer for the Chip Industry

Jay Goldberg argues that Arm's business model essentially involves licensing its core "instruction set architecture" (ISA) to chip design companies, akin to a "blueprint" or "standard component" provider.

• Mechanism Breakdown: Arm does not manufacture chips itself; instead, it licenses its IP (intellectual property) to customers such as Qualcomm, NVIDIA, and Apple. Its revenue model consists of two parts:

1. Upfront License Fee: An entry fee customers pay to obtain the right to use the IP.

2. Royalty: Every time a customer sells a chip based on Arm IP, Arm collects a percentage royalty, averaging roughly 7-9 cents per chip.

• Historical Analogy and Core Value: Goldberg explains Arm's value using an architect analogy — an architect is not rewarded for designing the best water pipes, so they simply adopt standard water pipe blueprints, focusing instead on truly differentiating elements like the building's exterior and lighting. Arm's role is to provide these "standard water pipe blueprints", handling the generic, low-value-added mathematical calculations and basic functions in chip design, allowing customers (e.g., Qualcomm's modems, NVIDIA's graphics processing) to focus on their own differentiating advantages.
• CPU vs GPU Relationship: The CPU is a general-purpose processor, while the GPU is an accelerator dedicated to specific tasks (e.g., graphics, AI). In complex systems, even in high-performance computing nodes, the CPU remains indispensable as a "control node". For example, in NVIDIA's high-end AI systems, its self-developed Grace CPU is based on the Arm architecture, which constitutes an important "attachment point" for Arm in the AI wave.

2. From Acorn to iPhone: The Triumph of RISC and a Historic Turning Point

Jay Goldberg points out that Arm's success stems from its adoption of the RISC architecture and its ability to seize the historic opportunity to expand from smartphones to data centers.

• Historical Context: Arm's predecessor was the chip design team of the British company Acorn, later funded by Apple to develop chips for the Newton product. Although the Newton was unsuccessful, its revenue supported Arm's early development. The real turning point was Nokia, whose 8110 phone (the phone from The Matrix) became the first Arm-powered mobile phone. Thereafter, Arm established dominance in the mobile phone market based on its low power consumption advantage over x86.
• Mechanism Breakdown of RISC vs CISC: RISC (Reduced Instruction Set Computer) and CISC (Complex Instruction Set Computer) are two different chip design philosophies.
• CISC (x86): Provides complex single instructions (e.g., square root), but execution is complex and power-hungry.
• RISC (Arm): Provides a small set of simple instructions; complex operations must be combined using multiple simple instructions. This initially required more memory, but as memory costs fell, its advantages of low power consumption and high flexibility became prominent.
• Key Conclusion: Goldberg emphasizes that the true advantage of RISC lies not only in power consumption, but also in its "flexibility". Arm's licensing model allows hundreds of customers to customize designs based on its architecture, in stark contrast to the closed ecosystem of x86, which is tightly controlled by Intel/AMD.

3. SoftBank Acquisition, NVIDIA's Ambition, and the "Awakening": A Strategic Inflection Point

Jay Goldberg argues that NVIDIA's failed acquisition of Arm unexpectedly "awakened" SoftBank, making it realize Arm's huge potential was undervalued, thereby bringing new growth momentum to the company.

• Historical Event Analysis:

1. SoftBank Acquisition: 10 years ago, SoftBank acquired Arm for approximately $32 billion. At the time, the market was generally puzzled, believing the price was too high. Under SoftBank, Arm entered a "sleeping period" because there was no longer the pressure of quarterly growth from the public market.

2. NVIDIA's Ambition: NVIDIA attempted to acquire Arm for $54 billion. Goldberg believes the motivation was not entirely clear, but CEO Jensen Huang's "superpower" lies in making big bets. He may have been dissatisfied with the pace of Arm's development in data center CPUs at the time and hoped to accelerate its own CPU development through the acquisition.

3. "Awakening" Effect: After the failed acquisition, SoftBank recognized that its "sleep" had caused it to miss opportunities, so it introduced a new CEO, Rene Haas. Haas implemented a comprehensive overhaul of the company's pricing, products, and marketing, achieving a "revival".

• Current Status and Future: Arm is systematically enhancing its value capture capability. Taking NVIDIA as an example, for its AI system priced at $100,000, Arm may only earn $1-2 in royalties, implying huge room for price increases. Goldberg points out that Arm is locking in more customers by lowering upfront license fees and raising later-stage royalties, gradually increasing the average royalty from $0.07 to above $0.12.

4. Financial Model and Moat: Competition and Risk Under High Leverage

Jay Goldberg likens Arm's financial model to that of a "software company" and points out that its biggest moat is extremely high switching costs, while it also faces challenges from RISC-V and structural changes in the AI market.

• Financial Model: Arm boasts >90% gross margin and 40-50% operating margin. This is because its core cost is upfront R&D investment, while the marginal cost of each additional chip royalty is nearly zero, creating strong operating leverage.
• Moat: Switching Costs: Goldberg emphasizes that replacing an ISA is extremely difficult. He cites Apple's transition from Intel x86 to its own M-series (Arm) processors, noting that Apple spent years and enormous capital on software compatibility development. This demonstrates that once customers adopt the Arm architecture, they become deeply locked in, with switching costs measured in "decades".
• Risks and Competition:
• RISC-V Threat: As an open-source ISA, RISC-V is a potential threat, but it is currently only suitable for embedded, low-value IoT devices and is not yet ready for smartphone or data center workloads. Rene Haas has significantly weakened RISC-V's appeal by adjusting pricing strategies.
• Structural Changes in the AI Market: This is Arm's "bear case". If AI systems become far less dependent on CPUs (e.g., the GPU-to-CPU ratio shifts from the current 2:1 or 8:1 to 100:1), then Arm's total addressable market (TAM) will be much smaller than expected, and growth may hit a ceiling.
• Customer Relationships: Arm is engaged in litigation with major customers like Qualcomm, which could affect their partnership, but customers have almost no choice due to prohibitive switching costs.

Mentioned Positions

Position	Guest Attitude (Bullish/Risk Warning/Neutral)	Key Data
NVIDIA	Neutral (Related party, also a customer)	In NVIDIA's AI system (priced at $100,000), Arm earns approximately $1-2 in royalties.
Qualcomm	Risk Warning	Currently in litigation with Arm; one of Arm's largest customers, but relations are tense.
Apple	Bullish (Historical success case)	Owns the M-series chips; Apple's migration from x86 to Arm demonstrates switching costs.
Amazon / Microsoft / Google / Meta / Alibaba / ByteDance / Tencent	Bullish (Data center growth point)	These "Super 7" companies have all designed self-developed data center CPUs based on Arm to replace expensive Intel/AMD chips.
Intel / AMD	Risk Warning (Competitors)	Owners of the x86 architecture; main competitors to Arm in data centers and PCs.
RISC-V Ecosystem	Risk Warning (Emerging competitor)	Has some share in embedded, low-value IoT devices, but does not yet pose a material threat to Arm.

Key Takeaways

1. Arm's business model is a "blueprint" rather than a "finished house" (Jay Goldberg): It standardizes the common parts of chip design (e.g., basic mathematical operations) and licenses them to customers, allowing customers to focus on differentiating features (e.g., graphics, AI processing). This is the cornerstone of its business model's success.

2. The true advantage of the RISC architecture is "flexibility", not just "low power consumption" (Jay Goldberg): Low power consumption enabled its success in the mobile phone market, but flexibility and its open licensing model allow Arm to have a vast ecosystem, which is key to competing against the closed x86 ecosystem.

3. NVIDIA's failed acquisition of Arm was SoftBank's "awakening" (Jay Goldberg): The failed acquisition prompted SoftBank to realize that Arm's value was severely underestimated, leading to the appointment of new CEO Rene Haas, who implemented a comprehensive overhaul of the company's strategy, pricing, and products.

4. Arm's "moat" is a switching cost measured in "decades" (Jay Goldberg): Changing the core ISA means almost rewriting the entire software ecosystem. Apple's migration from x86 to Arm is the most vivid proof. This creates a powerful customer lock-in effect.

5. RISC-V is not currently an "existential threat" to Arm (Jay Goldberg): Although RISC-V is open-source and free, it is not yet mature enough to handle high-end workloads in smartphones or data centers, and Arm has already significantly weakened its initial appeal by adjusting pricing strategies.

6. Arm's growth engine is "value capture" rather than just "market expansion" (Jay Goldberg): The core logic is that the royalty collected per chip, currently at 7-9 cents, has huge room for improvement (e.g., to 12 cents). This "built-in growth" is a key support for its valuation.

7. Arm's "bear case" is the "de-dependency" on CPUs in AI systems (Jay Goldberg): If future AI computing systems have an extremely imbalanced GPU-to-CPU ratio (e.g., 100:1), Arm's market space will shrink significantly. This is the most noteworthy risk in its growth path.