Automotive OEMs Focus On SDVs, Zonal Architectures

Giant automotive OEMs are re-evaluating how quickly to move to advanced technologies and software-driven designs amid crushing financial pressure from low-cost EVs developed in other markets such as China.

U.S., European, and Japanese OEMs have been struggling for the past half-decade or so to figure out which is the best approach to developing EVs, undergoing multiple shifts in both hardware and software architectures. What they have discovered is that developing these technologies and shifting direction is harder than expected, which accounts for different rates of adoption by region and by company. So, while investment in Advanced Driver Assistance Systems (ADAS) is expected to continue through 2030, it will decelerate in the coming years, according to research conducted by Siemens. In-Vehicle Infotainment (IVI) investment already has peaked and is also decelerating, suggesting that OEMs view these as largely solved problems, allowing them to shift focus to more transformative technologies.

What comes next is a big push toward software-defined vehicles (SDVs) and zonal architectures. In Europe and North America, investment in these areas is growing exponentially. OEMs now realize that incremental improvements to ADAS and IVI won’t achieve the desired results, and a more fundamental shift to zonal architectures is necessary, according to David Fritz, vice president for hybrid and virtual systems at Siemens Digital Industries Software.

However, the transition isn’t uniform across all regions. In China, for example, many new entrants started with zonal architectures or a clear path towards them, so their investment in SDV is comparatively lower.

“What is happening under the hood is that ADAS and IVI are pretty much considered solved problems, so the amount of investment going forward decelerates based on that assumption,” Fritz observed. “But in Europe and in North America, the investment in SDV and zonal is exponential. They are learning what we’ve been saying for a very long time — you need a zonal architecture, you need central compute, you need all the things on the edge working properly, and then you can implement SDV, and you’ve got a full CI/CD flow, and all those other things fall from that zonal architecture. They are giving up on the fact that they thought they could incrementally get there by investing in ADAS and IVI, and we can’t get there. We have to go to a zonal architecture so they’re just starting to invest in that. And the investment grows exponentially.”

And even though China is investing slower in SDV, it is following a traditional path. “I went to my old buddies at Xiaomi that I knew from my Qualcomm days and asked what they were doing for their cars, Fritz noted. “They said, ‘We’re just doing the same thing we did with smart phones.’ When I asked them to draw it out, they drew a zonal architecture with SDV. So, I talked to several companies there, and they said the same thing. The investment in China goes to other parts, like ADAS and IVI, because it’s flashy. They started with a zonal architecture, or at least a path to get to it, so their investment is lower.”

Wayne Lyons, senior director, Automotive Segment at AMD is also observing that OEMs and Tier 1s are very much focused on making the transition from domain-based architectures to zonal architectures. “While just a few automakers have production vehicles on the road that utilize a zonal-based architecture today, enabling this transition is a key priority in the automotive industry. The shift to a zonal architecture also coincides with the Software-Defined-Vehicle – the two topologies complement one another.”

While there is not a 100% correlation between SDV and zonal architectures, zonal architectures greatly simplify many complex tasks required for effective SDV implementation. To date, Fritz said he has yet to see a successful SDV implementation that isn’t zonal, as the two technologies are very synergistic.

Lyons agrees zonal architectures and SDVs go hand in hand, and explained, “Zonal architectures streamline vehicle electronics by centralizing processing power into fewer, more capable controllers that manage different zones of the vehicle. This setup reduces the complexity of wiring, increases efficiency, and creates a robust platform for real-time data management. In parallel, SDVs rely on advanced software that can be updated over the air, delivering new features and services long after the vehicle leaves the factory. For SDVs to thrive, vehicles need the adaptable, high-performance computing backbone that zonal architectures provide.”

Money matters
The financial pressure from new entrants is influencing traditional OEMs’ technology adoption strategies. In China, BEVs sell for a fraction of what a comparable vehicle sells for in Europe and the United States. A BYD Seagull, for example, has a list price of about $12,000 in China.

Traditional OEMs recognize the need to adapt quickly to remain competitive. Some are creating separate divisions for autonomous solutions to facilitate faster implementations of new technologies without being hindered by existing organizational structures.

“One concern is that the Chinese OEMs are very successful in China, which is putting pressure on OEMs like VW, and the Japanese and U.S. OEMs, which used to sell a lot of vehicles in China,” said Judy Curran, senior chief technologist for automotive at Ansys. “So they’re losing market share globally, and that is putting pressure on them because their overhead and staffing always gets spread across how many vehicles are sold. Another concern is that the total industry volume has been about flat for the last four or five years. As some OEMs grow, others have to shrink. There’s also a concern about the Chinese OEMs gaining market share in Europe. There are new tariffs coming, but they’re also purchasing plants, which adds pressure. In the U.S. we have tariffs, but in Mexico there’s a lot of presence of [Chinese vehicles]. So there are a growing number of OEMs — Chinese, for sure, but also West Coast OEMs — that are putting pressure on the volume. Everybody is feeling that financial strain.”

Heavy investment in EVs ratchets only adds to that strain. “Building battery plants and assembly plants is not inexpensive,” Curran said. “But the consumer isn’t purchasing [EVs] at the rate U.S. and European OEMs hoped, and for which they now have invested. So they can’t get the payback on their investment. We see these very distressing headlines with Volkswagen talking about shutting down plants in Germany, which is unheard of. Then you have Stellantis’ announcement that their ‘days on lot’ is much higher, so their stock dropped double-digit on the day they made that announcement. The combination of the EVs and the volume are stressing these OEMs. And then, BMW had the issue with Continental and the brake recall, affecting 1.5 million vehicles. Across the board, there is financial stress.”

Curran believes this is an ideal time for the automotive ecosystem to adopt tools such as simulation software, because the big trend is reducing cost, which is a combination of the cost of the components in the vehicle and the cost of engineering the vehicle. “Whatever you spend to engineer the vehicle has to be spread across the project.”

The financial pressures also are impacting headcount and technology development within the automotive ecosystem.

“There are layoffs in the industry,” she said. “General Motors said it was laying off 1000 software engineers. It was supposed to be all about the software-defined vehicle, so they need to look at their overhead and their expenses and try to decide where to reduce costs.

OEMs also want to reduce the cost of building and testing vehicle prototypes, and this is where simulation can play a critical role. Adding AI can help leverage learnings from past designs, as well. And both will be essential in a broad shift toward zonal architectures and SDV, which are necessary for long-term competitiveness.

Traditional OEMs aim to reduce the number of ECUs from about 100 to 30 to 40, which would result in significant cost savings and greatly simplified networks. No one doubts the transition will happen. The question now is when that will happen.

Not every vehicle model can be updated simultaneously, and OEMs often take a staged approach, implementing partial steps towards full zonal architectures across their fleet. While there is pressure to move quickly, OEMs are cautious about over-promising and under-delivering, especially given past hype cycles around autonomous vehicles. They are striving to balance innovation with realistic timelines and deliverable promises.

“Let’s not forget that most vehicles today still rely on the traditional distributed and domain-based E/E architecture that was designed to optimize costs, with cost as the driving factor,” noted Adiel Bahrouch, director of business development for silicon IP at Rambus. “This includes the selection of the sensor sets, the selection of compute performance, the memory storage. Everything is selected in a way to have a cost-effective solution, where the hardware and the software are very tightly coupled. So it’s a cost-effective strategy, but it’s almost impossible to add updates or upgrades, and definitely not for in-field vehicles post-production. Now we’re seeing a move to the next generation centralized e-architecture, where the focus is to decouple the hardware and software, which allows feature upgrades and so forth. It will come with some challenges, and obviously the cost will increase, since to make it happen the underlying hardware must be oversized in a way — over-specified. But since 60% or 70% of vehicles nowadays still rely on the traditional distributed, and maybe even domain-based e architecture, this legacy will always be there. It’s going to be tough to make that transition to the architecture where the hardware and software are decoupled and software updates will be easy to manage throughout the lifecycle of the vehicle.”

Adopting new technologies requires OEMs to have a high level of expertise in house, and with automotive software marketplaces like SDVerse emerging, the weight of the task is causing some projects to falter.

“When VW or Toyota struggle, then the whole industry is going to struggle, because those are the big guys in the industry with the highest volume,” said Curran, pointing to a recent VW-Rivian announcement. “VW was on the forefront in saying they were putting all their software engineers together from their 10 brands under the CARIAD brand. They’re going to do this new architecture and do all the software, and all the other OEMs said, ‘We’ve got to do that too.’ But then CARIAD completely failed. They could not deliver those first vehicles. Those vehicles were delayed — and they were big-brand vehicles like Porsche, so there was a lot of backlash. People lost their jobs. Then they got a new CEO, who was setting them up for round two of this goal of doing this architecture and the software. And then the team was surprised to read in the press that they’re going to give $5 billion to Rivian, and part of that deal is to co-develop, but also get their architecture. So here’s VW saying it’s too hard to come up with an architecture. ‘We’re an 86 year old company, and we have to undo all this. All these brands are too complicated. We’ll just say this is the one architecture, and Rivian is able to do it because they are ground-up.’ So, they started with it. The real question is how are they going to go about doing it? A) It’s not easy. B) It’s expensive. So the speed at which it happens might go longer. Maybe they don’t go from where they are today to this end architecture overnight. Maybe they do a couple steps along the way, and morph their way into it.”

Others agree that OEMs are having a tough time with this transition. “The pendulum is still swinging,” said Marc Serughetti, vice president of product management and applications engineering at Synopsys. “Two years ago you had OEMs saying, ‘I’m going to do everything. I’m going to do semiconductor. I’m going to do software. I need to own the software.’ What we’re seeing now is maybe that pendulum swinging back a little bit the other way, and the OEMs are now asking, ‘How am I going to do this? What needs to change?’ It’s really a digital transformation that needs to happen for them. It just doesn’t happen in one day. You need SoCs that are more powerful, that have more capabilities. We talked about AI and all those things coming in. You need a software development infrastructure that is in place, and to me there are a lot of technology aspects, but there are also methodology aspects and organizational aspects. That needs to evolve. So that transformation cannot just happen. It’s not just a technology transformation. It is an organizational and methodology transformation. The tooling that goes with that asks, ‘How do you start integrating all that enterprise software?’ We’re still in an embedded world within the car. How do you start looking at the embedded aspect and the enterprise software development aspect and bring those things together? These are things that have been solved in other industries. We must figure out how to apply this to this specific domain of the car — an embedded system that’s distributed, that has safety and security requirements all together.”

That also requires a corporate structure to allow for change. “Some of these companies in Detroit go out and hire a new CEO, and they go out and tap into Apple or Google or another company,” said Siemens’ Fritz. “Then, six months later, they get rid of that person and do another reorganization. What they can’t seem to grasp is there is an approach that works here, and it’s been well known for decades. But somehow it has been forgotten here in North America. You create a separate division, you isolate it from all other influences, let them make their own autonomous solutions. When they start to be successful, you spin them out as a wholly owned subsidiary, keep them away from all the felgercarb that these other guys have to deal with. Once they’re successful enough, you roll them back in and they take over. That happens in Japan on a regular basis. It’s an old way of doing this.”

Challenges and industry movement
While the move to zonal architectures and SDVs is promising, it presents challenges, AMD’s Lyons said. “Automakers must navigate integration with legacy systems, develop advanced software capabilities, and address cybersecurity concerns as connectivity increases. Moreover, the semiconductor supply chain still poses challenges for accessing advanced chips. Overall system cost must be addressed as well, balancing the need for feature-rich hardware that can enable the underlying requirements for an effective centralized architecture.”

Despite these hurdles, automakers and Tier One suppliers are making significant progress. “Leading OEMs are partnering with tech innovators like AMD to bring zonal architecture and SDV technologies to life,” Lyons added. “Based on recent market data, the shift to a more centralized zonal architecture was expected to move much more quickly. Due to the challenges indicated, however, the majority of vehicles will continue to be distributed and/or domain-based through the end of the decade. Zonal architectures will become more common beyond 2030.”

Robert Schweiger, group director automotive solutions at Cadence agreed. “For most OEMs, the transition to a zonal architecture is an evolutionary process that requires various key components. Requirements include the introduction of new higher-resolution sensors and high-performance SoCs while simultaneously upgrading the in-vehicle network to higher speed grades. There are several essential elements in this transition. First, a high-speed link between the central compute unit and zonal controllers is required, ideally using a 10G automotive Ethernet MAC interface in the zonal controller and the central compute SoCs. This is available today. A 10G-Base-T1 Ethernet PHY as a separate chip is also needed and is currently available as test samples from various providers but not yet available in volume production. Finally, a high-speed SerDes such as an MIPI A-PHY or ASA-ML SerDes is required. While an 8Gbit/s MIPI A-PHY is currently available, a 10Gbit/s ASA-ML SerDes is available as test chips but not yet available in production volume. Hence, OEMs must upgrade their electrical/electronic architecture, or in-vehicle network, as these PHYs and SerDes devices become available.”

Conclusion
While financial pressures are prompting re-evaluation, automotive OEMs recognize the necessity of moving toward advanced technologies like SDV and zonal architectures to remain competitive in the long term. The challenge lies in managing the transition effectively while balancing immediate financial constraints. And in a risk-adverse industry populated by public companies, this is no small feat.

“The industry in general is risk-averse,” said Ansys’ Curran. “I worked in that for 30-some years, and I totally understand it because there’s no greater pressure than sitting at your desk knowing that assembly plants are running and every minute a vehicle is going out the door and, on a train, and that’s being sold to a customer. It’s their first or second most expensive purchase, so you want to do a good job. You do not want the feeling that that car is not good. It is a risk-adverse industry because of the speed at which the vehicle is being built and the expense of the item. It’s not like a phone, which is only $1,000 every few years. It’s expensive, but not $50,000, and it doesn’t kill you. Cars can kill you. And because of all that, it is risk-averse. The automotive industry is making decisions, but they’re not going very fast. They are going to take the risks, but it’s not going to be overnight.”

Related Reading
Balancing Programmability And Performance In Cars
Customization and adaptability are essential for software-defined vehicles, but there’s a price to pay for that flexibility.
Moving Software-Defined Vehicles Forward
Automotive OEMs must undergo a change of mindset and a willingness to embrace open standards for SDVs to become a reality.
Toward A Software-Defined Hardware World
New approaches to software-defined hardware involve a rethinking of model-based systems engineering.