Arm's Zena CSS aiming to speed up automotive AI

British compute platform company Arm is positioning its new Zena CSS as a flexible computing foundation for the next generation of AI-powered vehicles, providing automakers and chipmakers a way to accelerate development while maintaining flexibility for differentiation.

The platform, which it claims can enable automakers to launch new vehicle models at least one year faster, comes at a time when the car industry is accelerating its evolution from mechanical products to AI-driven ones.

Suraj Gajendra, vice-president for Product and Solutions, Automotive Line of Business at Arm, said the platform has attracted more than 10 global partners, including automakers and chip designers.

"Some of them have licensed the technology, while others are in advanced stages of engagement," said Gajendra in an interview on Tuesday without disclosing their names.

Gajendra said the new Zena CSS delivers high-performance computing capacity to support applications ranging from in-car AI assistants to advanced driver-assistance systems and vehicle control applications, which are must-haves for new vehicles these days, especially in China.

Long before the launch of the new platform, Arm had been providing solutions for the auto industry. Most big names in the sector are its customers, including BMW, Toyota, Geely and Nio.

In the last five years, Arm-based chip shipments to the automotive market have tripled, statistics from the company showed.

Besides raw performance, the Zena platform allows partners to optimize designs according to their own needs, unlike traditional chip designs that lock customers into specific layouts or manufacturing processes.

"They can customize features, integrate with other components, or target different markets without being constrained by rigid specifications," said Gajendra.

The platform also anticipates the growing trend of modular chip designs, giving partners the ability to combine or adapt Zena CSS units according to their product strategies.

This approach allows manufacturers to scale computing power as AI demands increase, while still retaining flexibility to innovate in areas such as AI accelerators, Input-Output interfaces, or other features.

"Our core goal is to standardize a small but critical set of components — the CPU cores — while leaving the rest of the design to partners for differentiation," said Gajendra.

"Whether it's a carmaker developing its own chips or a traditional chip partner building a SoC (system on chip), they can innovate in accelerators, custom logic, or overall system architecture. This approach helps reduce development complexity while enabling creativity and flexibility," he added.

With vehicles expected to host more advanced AI workloads, from predictive maintenance and driver monitoring to adaptive infotainment systems, the demand for integrated, high-performance platforms is rapidly growing.

"The transition to software-defined vehicles demands a shift in how we approach compute architectures," said Magnus Ostberg, chief software officer at Mercedes-Benz.

He said standardized, pre-verified compute subsystems like Arm Zena CSS can significantly accelerate development timelines and reduce complexity across the industry.

"At Mercedes-Benz, we recognize the value of collaborative platforms that drive the entire ecosystem forward," said Ostberg.