Arm makes history by smashing more records in blockbusting Q4

It crossed the $1 billion revenue milestone for the first time in its 35-year corporate history, achieving record licensing revenue and smashing the royalty amount it receives per chip.

Looking at the split of returns, royalty revenue exceeded $600 million for the quarter. The full-year revenue figure topped $4bn and royalty revenue $2bn for the first time, Haas revealed.

He said revenue poured in from multiple markets including data centre, automotive, smartphones and IoT.

Haas said the earnings power allowed the business to invest record amounts in R & D. He said that AI growth from the cloud to the edge was creating demand for more energy-efficient compute.

New AI hardware, the rise of AI agents, and the emergence of smaller, more efficient language models are unlocking new AI use-cases in data centres.

Haas said: “Time and again we have seen AI workloads – especially inference – migrate from the cloud to the edge, driven by the need for faster response times, enhanced privacy, and reduced energy consumption.

“Arm is uniquely positioned to lead this shift, with the most pervasive compute platform in the world – already enabling AI inference at scale across cloud and edge environments,”

During the quarter, NVIDIA announced that their Armv9-based Grace Blackwell superchip was now in full production. The computational power and efficiency of the Arm Neoverse-based Grace CPU is integral to meeting this high demand, according to Haas.

He said that iIn addition, major hyperscalers were pairing Armv9-based custom silicon with their own accelerators to run AI workloads, including Amazon Web Services (AWS), Google Cloud and Microsoft Azure.

With more AI software being written first for Arm-based chips, he said Arm expects that close to 50 per cent of all new server chips shipped to top hyperscalers this year will be Arm-based.

NVIDIA also announced that its AI-desktop, DGX Spark, powered by the Grace Blackwell superchip with Armv9 CPUs, is gaining traction - reinforcing strong demand for Arm-based AI infrastructure among developers and enterprises building next-generation AI models.

Also during the quarter, Google and Microsoft both announced the rapid expansion of their Arm-based data centre chips. Google confirmed that Axion is now available in 10 regions and is being used by approximately 40 of its top 100 customers, including Spotify. It offers up to 65 per cent better price performance than current generation chips based on x86.

Microsoft expanded the range of software support on their Arm-based Cobalt 100 chips and are running massive workloads from customers spanning Databricks, Siemens, and Snowflake, as well as internal workloads including Microsoft Teams and Copilot.

Haas said that Arm would continue “to invest relentlessly’ in long-term innovation to capture the wide range of opportunities in front of us.”