Arm powers Q3 revenue surge while chips shipped hit 280.3 billion

09 Feb, 2024
Tony Quested
Revenues in Cambridge-based superchip architect Arm were 14 per cent up to $824 million in its third quarter – the company’s second quarter on NASDAQ, the US technology exchange.
Arm CEO Rene Haas. Credit – Arm.

Royalty revenue of $470m in Q3 was up 11 per cent year-on-year, driven by the semiconductor industry recovery and the rapidly growing penetration if Armv9-based chips which typically command a higher royalty rate.

Arm also saw better than expected licence revenue of $354m in the third quarter – up 18 per cent year-on-year; it said a surge in bookings was due to hefty demand for more advanced Arm CPUs as companies increased their investment in Artificial Intelligence across all end-markets. Q3 non-GAAP operating profit increased 17 per cent year-on-year to $338m.

Customers reported that they had shipped 7.7 billion Arm-based chips for the September quarter shipping period. This takes the cumulative number of Arm-based chips reported as shipped to 280.3 billion.

CEO Rene Haas said: “Arm delivered another quarter of record revenues driven by continued adoption of the world’s most pervasive compute platform.

“More customers moving to higher-value Armv9 technology combined with market share gains in cloud server and automotive resulted in strong royalty growth. The AI wave drove licensing growth as these new devices require Arm’s performant and power-efficient compute platform.”

Haas said Arm was “only at the beginning of its growth potential as its strategy promoted multiple long-term expansion drivers. And that growth will be driven by royalty revenue”, he told shareholders. “We expect the demand for compute to continue to grow as the complexity and performance requirements of electronic devices increase.”

“Additionally we expect that Arm will continue to gain market share, especially in key end-applications such as automotive and cloud servers, where the virtuous cycle of more Arm-based chips going into more products drives more software development, which in turn generates more demand for Arm technology.

“This is compounded with increasing performance requirements that continue the trend of higher royalty rates as end products require more Arm CPU’s per chip and more advanced Arm technology, such as Armv9.

“Chips based on Armv9 technology now contribute around 15 per cent of Arm’s royalty revenue, up from around 10 per cent last quarter.”

The company says it is seeing the demand for Arm technology to enable AI everywhere, “from the cloud to edge devices in your hand. Generative AI and Large Language Models (LLM) need very high-performance processors. These processors need to operate within constrained energy and thermal budgets making a power-efficient computing platform essential.

“The most demanding AI applications are already running on Arm today; for example, NVIDIA recently announced that dozens of new AI supercomputers are coming online incorporating their Arm-based GH200 Grace Hopper Superchip.

“Key players like Dell Technologies, Hewlett Packard Enterprise (HPE), Lenovo, Quanta and Supermicro are among those using the GH200 to tackle some of the world’s most challenging problems.

“At the edge, Google recently revealed their LLM, Gemini Nano, running on the latest Arm-based Pixel 8 smartphone, and Samsung, Vivo and Xiaomi all have announced new Arm-based smartphones that demonstrate generative AI and LLM capabilities.

“Some of the features include live call translation, intuitive search features such as circle to search, and advanced photo and video capabilities.

“Growth will be driven by Compute Subsystems. The growth of AI is putting more demand on scarce design resources, creating the opportunity for Arm to do more.

“Complex chips are becoming more difficult and taking longer to design, compounded by chip manufacturing cycles taking longer as well. This means chip designers must do more in less time to hit their product launch window.

“Our solution to this is Arm Compute Subsystems (CSS) which are integrated and verified configurations of Arm technology platforms targeting specific end markets and use cases.

“We are already seeing proof points of this strategy in the infrastructure market. One Arm Neoverse CSS customer successfully went from concept to working silicon of an advanced server chip in just 13 months, reducing their development time by 50 per cent, and another reported saving 80 years of engineering effort resulting in over $20 million savings in cost.”

During the quarter, Microsoft announced its first cloud server chip, the Microsoft Cobalt 100, which is based on Arm Neoverse CSS.

Growth will also be driven by Arm’s unique ecosystem of software and design partners, shareholders were told. Arm already has the world’s largest compute ecosystem with more than 15 million software developers.

This unprecedented number of developers leads to seamless design choice by OEMs selecting Arm CPUs as the platform of choice. As more applications move to AI, Arm is investing in the software ecosystem needed to bring AI to the billions of Arm-based devices.