Arm Seeks Larger Stake in Chip Market Amid AI Revolution

{{IMAGE:1}}

In the architecture of modern computing, Arm Holdings occupies a paradoxical position. The British-origin, SoftBank-controlled company's processor designs power over 99% of smartphones and countless Internet of Things devices, yet it manufactures no physical chips itself. Instead, Arm operates on an intellectual property licensing model: Companies pay upfront fees for architecture blueprints and ongoing royalties—typically 1-3% of chip value—for the right to produce Arm-compatible processors. This approach has shipped over 300 billion chips to date, including 30 billion units in 2025 alone.

{{IMAGE:2}} Rene Haas, Arm's CEO, has signaled strategic shifts to capture more value from AI-driven chip demand. (Photograph: Rodrigo Reyes Marin/ZUMA Press Wire/Eyevine)

The rise of artificial intelligence is disrupting this decades-old formula. AI workloads require specialized processor configurations that extend beyond Arm's traditional CPU designs into adjacent domains like neural processing units (NPUs) and high-bandwidth memory systems. As companies like Nvidia, Amazon, and Google design custom AI chips, Arm risks being relegated to providing commodity CPU cores while others capture premium value from integrated AI subsystems.

According to recent financial disclosures, Arm's royalty rates haven't significantly increased despite chips growing more complex and expensive. Flagship smartphone processors now cost over $150 compared to $20 a decade ago, yet Arm's take remains proportionally small. This discrepancy becomes more pronounced in AI server chips costing thousands of dollars, where Arm's architecture often serves as the control hub coordinating specialized accelerators.

CEO Rene Haas has signaled strategic shifts toward "value-based pricing" that better reflects chip complexity. Industry analysts suggest this could mean tiered royalty structures where customers pay higher percentages for chips incorporating Arm's newer AI-enhanced designs like the Neoverse Compute Subsystems. The company is also expanding its Total Design ecosystem to help partners integrate Arm IP into custom AI chips faster, potentially capturing more design service revenue.

Recent financials reveal the urgency: While Arm's Q4 2025 licensing revenue grew 18% year-over-year to $454 million, royalty revenue increased just 5% to $524 million—lagging behind the 35% growth in AI chip sales industry-wide. With SoftBank reportedly planning to sell additional Arm shares later this year, pressure mounts to demonstrate how the company will capture more value from the $50 billion AI chip market it helps enable.

{{IMAGE:4}}

The transition involves delicate negotiations with partners like Qualcomm and Apple who've built empires atop Arm's foundations. Some are pushing back against proposed royalty increases, arguing their own engineering investments create much of the value. Meanwhile, RISC-V open-source architecture looms as a potential alternative, though it currently lacks Arm's extensive software ecosystem.

As Haas navigates these tensions, his strategy appears twofold: First, position Arm as the essential control layer orchestrating diverse AI accelerators through technologies like the Compute Subsystem for AI. Second, expand into new markets like automotive and AI PCs where integrated designs command higher margins. Success would redefine Arm from a behind-the-scenes enabler to a central value-capturer in the AI hardware stack—a transformation as complex as the chips it designs.