Arm Projects $25 Billion Revenue by 2031 as AI Inference Chip Market Nears $50 Billion

TL;DR: Arm Holdings forecasts $25 billion in revenue by fiscal 2031, driven by surging demand for AI inference chips. The inference‑focused AI chip market is expected to hit $50 billion this year, with data‑center workloads set to rise from 21 GW to 93 GW by 2030.

Arm designs energy‑efficient processor cores that other companies license to build chips. AI inference runs a trained model to produce outputs, requiring far less compute than training the model. Because inference can be handled by CPUs as well as accelerators, firms favor architectures that deliver performance per watt.

Deloitte estimates the inference‑focused AI chip market will reach $50 billion in the current year. McKinsey projects AI inference workloads in data centers will grow from about 21 gigawatts to 93 gigawatts by 2030, a compound annual growth rate of 35 %. Arm says it will generate $25 billion in revenue by fiscal 2031, more than five times its recent trailing‑twelve‑month figure of $4.7 billion.

Arm earns an upfront licensing fee when a partner signs a deal and a royalty on each chip sold using its architecture. Royalty rates for the Armv9 AI‑focused design are roughly double those of the previous Armv8 generation, boosting revenue per unit. The company’s fiscal year ends in March, so fiscal 2031 covers the period ending March 2031.

The outlook suggests Arm’s licensing and royalty businesses will expand as more chipmakers adopt its Armv9 architecture for inference tasks. Higher royalty rates on the new design and Arm’s move into selling its own silicon could boost top‑line growth. Nvidia, AMD, Broadcom and Intel are also developing inference‑oriented processors, aiming to capture share of the growing market.

Investors should watch for new design wins, shifts in royalty rates, and how quickly Arm’s proprietary CPU gains traction in data‑center and edge markets.