The age of the AI-defined vehicle: Arm launches Zena CSS

At embedded world 2025 Paul Williamson, SVP and GM, IoT Line of Business at Arm spoke to IoT Insider about the company’s outlook five years in the future - to understand what technologies will grow in adoption and where to place their bets. Arm once again demonstrated their prescience and finger on the figurative pulse with the Arm Zena CSS platform.

The solution, which was launched on the 4th June 2025, is designed to not only help vehicle manufacturers to speed up development cycles but also to allow them to remain competitive in an era increasingly being shaped by the software-defined vehicle (SDV). 

What differentiates SDVs from conventional cars is the inclusion of software that can be regularly updated and bring new functionality to the car to improve the driver experience; supporting infotainment systems and voice assistants, for example.

Figures mentioned in the briefing quoted Arm Zena CSS as reducing development time by up to 12 months, and reducing engineering resources by 20%. By accelerating time-to-market, Arm has launched a solution it believes will keep automotive manufacturers competitive and allow them to differentiate themselves.

The automotive industry, which was previously static and slow-moving, has sped up with the advent of SDVs and AI, said Dipti Vachani, Senior Vice President and General Manager, Automotive Line of Business at Arm in the briefing.

“You’re already seeing this transition into more centralised zonal compute, and this centralised zonal compute is very necessary for the software-defined vehicle,” said Vachani. “This software-defined vehicle now has connectivity to the Cloud, and software is deployed in real time to the car.”

AI-defined vehicles are building off of the success of SDVs, Vachani explained, and what this could look like involves three distinct use cases: advanced driver assistance systems (ADAS) such as acknowledging it is now driving through a city - and responding accordingly to ensure it is aware of potential hazards like cars, pedestrians and cyclists; in-vehicle infotainment to improve voice and app interaction for the driver;and vehicle system control, where the car will adjust its energy based on the load it is carrying.

Leaning into the ecosystem

Arm’s biggest strength is its ecosystem, said Vachani, and by being in touch with its customers - from vehicle manufacturers to chip vendors - it is able to listen to what the concerns are revolving around the AI-defined vehicle.

“A lot of my conversations revolve around, ‘How do I get the flexibility to innovate now that we have this AI-defined vehicle, the car is behaving completely differently? What does my brand stand for? What experiences do I want my customers to have?” said Vachani. “That changes based on which OEM you’re speaking to, and how they define themselves in the market.”

Another question which pops up is about scale. “Because of the explosion in compute and the explosion of software, scale matters,” added Vachani. 

Then, because of the pace with which the automotive industry is moving, keeping up matters; hence the reduction in time to market.

These conversations culminated in the launch of the Arm Zena CSS platform: continuing a journey started in 2024 with its automotive-enhanced IP and Arm v9 architecture delivering high performance and low power. At the time of this announcement, the company said it would integrate the IP into a compute subsystem, and Arm Zena CSS is the result.

Zena is made up of 16 Armv9-based Cortex A720AE cores which are pre-validated and pre-verified in response to increased demand for safety from the automotive industry, allowing its customers - including OEMs and silicon makers - to design and scale quicker. 

It also features a Cortex-R82AE powered safety island to increase the security and safety of the solution, which is of the utmost importance to the industry.

One of the biggest assets is the platform being standardised, which means automotive customers and silicon providers can reuse architecture and software - without having to build custom hardware and software stacks.

This standardisation started with the establishment of the SOAFEE (Scalable Open Architecture for Embedded Edge) initiative - which celebrated its three-year milestone in September 2024 and now registers 150 members. “In this ecosystem, we started creating a standardised way to implement the car and the Cloud … that is magic in allowing the whole software ecosystem to create software that anyone can plug and play to their vehicles today,” said Vachani.

Red Hat, who is a member of the SOAFEE initiative, have deployed a standard over-the-air communication that Arm’s customers can take advantage of, for example. Other examples include Mapbox, providing navigation with integrated voice technology; Strad Vision for AI-powered perception supporting ADAS and autonomous driving; and more.

The key takeaways of the platform were saving time and money for Arm customers; fast-tracking software development using pre-silicon virtual prototyping; and facilitating the differentiation of brands.

“The AI-defined vehicle will be built on Arm,” Vachani concluded.