Anders Bell is Chief Engineering & Technology Officer at Volvo.Volvo
Volvo is fully committed to partnerships with Nvidia or Qualcomm. This aims to make the Software-Defined Vehicle a reality quickly. Anders Bell, Head of Development, explains in an interview how the development is expected to proceed in the future.
Mr. Bell, how does the development of the Superset differ from traditional vehicle development approaches, and what challenges had to be overcome?
Of course, we still do a large part of traditional vehicle development because we continue to build cars. But the crucial difference lies in the introduction of what is commonly referred to as a "software-defined vehicle." It's a term I personally don't particularly like, but it describes a concept that most people can understand. This change brings a completely new dimension to vehicle development and fundamentally changes the considerations of how to approach development. We still have physical platforms and vehicle architectures. For example, the step from SPA-2 to SPA-3 is a significant advancement in vehicle architecture. The scalability of SPA-3 is much greater than that of SPA-2, which opens up new possibilities. We continue to develop a roadmap for propulsion and energy solutions that span architectures. This concerns drive units, power electronics, and batteries. This roadmap is developed jointly for SPA-2 and SPA-3. However, the crucial difference lies in the computing electronics and particularly in the software.
Can you please elaborate on that?
While there has always been computing electronics in the automotive industry, a major transformation is taking place: We are moving from closed black-box systems to open systems. The last component we fully control is the integration and development of the software stack that drives all vehicles. This means a huge change. In the past, we operated platforms for many years, then started anew with a blank sheet of paper and developed a new architecture. That world is a thing of the past. Now we are in a software-defined world, where every consideration begins with the software stack. Everything runs under a central software stack that serves as the basis for all architectures. This means that platforms and architectures continue to exist but are overlaid by a central software master. This central master allows us to make significant progress in vehicle architecture and develop iteratively without having to start from scratch at every step.
Advertisement
How do you ensure that the continuous expansion and improvement of the Superset remains flexible and future-proof in the long term?
This must be integrated into the design from the beginning. It is not a static system, but an ongoing development. We are moving towards a 'One-Tech-Stack' approach. However, this does not mean that the tech stack remains static. Quite the opposite: it must continuously evolve and adapt. Every single component - whether it is drive and energy, computing electronics, or mechanical hardware - must be continuously developed. It is essential that this happens under clearly defined and well-managed interfaces, both mechanically and electronically. A good example is successful technology companies like Apple. There, all products - from iPhones to Macbooks to Vision Pro systems - are integrated into a single, closely connected ecosystem. We apply a similar principle to vehicles. All components are housed within the shell of a vehicle and are connected to each other as well as to the cloud. Through this connection, we can collect, store, analyze data, and replay new software instructions to the vehicles via the air interface. The car thus becomes a dynamic part of a larger, constantly evolving system that continuously improves.
Volvo wants all models to benefit from the developments of the Superset. How does this knowledge transfer work in practice?
The vehicles are all connected to each other. If someone asks me what the biggest technological leap in the automotive industry of the last decades was, I would say: telematics. The introduction of telematics with full read and write access to the fleet was a significant advancement. This is not an original automotive technology, but its implementation in this area has revolutionized our way of working. It allows us to understand how vehicles are used and driven. We can analyze data in real-time and send new instructions directly to the fleet via the OTA interface. Of course, there are also things that we cannot change through software. These then flow directly into the hardware development of the next generation. Through these real-time insights from actual usage, we can significantly improve our understanding of the vehicles.
Are there already measurable improvements in quality or development times achieved through the Superset?
An example is the over-the-air update we recently announced. It will equip 2.5 million vehicles with an HMI upgrade that significantly improves the infotainment systems. This update demonstrates how we can continuously improve products by analyzing real-time data and utilizing customer feedback. It is not only a significant technical achievement but also proof of how all models are based on the central software stack of the Superset and can benefit from its developments. The special aspect is that we can gain insights from the actual use of the vehicles and directly incorporate them into product development. Customers who bought their vehicles three years ago receive a significant update to their infotainment system, which goes far beyond what they originally had. This process, which we call One UX, is based on all vehicles using the same software master, which is also integrated into new vehicles like the EX90. These are not separate development strands but a consistent software stack that is continuously updated and optimized.
Let's talk about the importance of real-time data. What role does data analysis play in the development cycle, and how is real-time feedback integrated into product development?
The topic of safety is particularly close to our hearts. It is a fundamental part of our work - in our offices, in our thoughts, and in every step of our development processes. A concrete example of the use of real-time data is the continuous improvement of the ADAS stack. Through real-time analysis of vehicle data that we gain from use in real-world operations, we can directly identify incidents, challenges, and risks. These insights are immediately fed back to our development teams in a closed loop, allowing them to quickly improve the systems. We can implement these improvements thanks to the generous consent of our customers who provide us with their data. This enables us to roll out code to vehicles daily to update systems or add new features. This is a tremendous advancement over traditional development approaches, where data often had to be physically retrieved from vehicles. With the Superset architecture, especially in the EX90 and the models built upon it, this process is fully integrated. But it's not just about active safety - that is, avoiding accidents. We also use this data in passive safety.
To what extent?
Whenever an accident is reported, we analyze the data to understand the impact and develop measures. Since the 1970s, we have been collecting information on accidents, but back then we had to physically go on-site with measuring devices. Today, we can obtain this information much more precisely and quickly through the technology of the Superset. These insights flow into both future hardware developments and software updates that we continuously provide. This closed loop - from data collection to implementation - is a central part of our work. It affects not only safety but really every aspect of the vehicle: how it is used, charged, and driven.
How does the scalability of the SPA-3 platform specifically differ from the SPA-2 platform, and what new digital possibilities have been realized as a result?
The SPA-2 platform was a major advancement in vehicle development, bringing us fantastic vehicles that were the best we had ever developed at the time. But with the introduction of the SPA-3 platform, we are taking a decisive step further. Especially with the EX90, we can show how much potential this new architecture holds. The SPA-2 platform was originally developed for a limited number of vehicle types, which meant its scalability was restricted. The SPA-3 platform, on the other hand, was designed from the outset to be fully scalable - from B-class vehicles to F-class. This comprehensive approach allows us to develop a wide variety of vehicle types on a single architecture without compromising on efficiency or performance.
What advantage does it offer that the SPA-3 platform was developed exclusively for battery electric vehicles?
This is a crucial advantage because it allowed us to eliminate all the limitations associated with internal combustion engines. This gives us the freedom to tailor the architecture specifically and without compromise to the requirements of a BEV. For the engineering team, this means we can optimize the platform for maximum efficiency. Without the complexity that comes with the internal combustion engine, we were able to create a scalable platform that is cost-efficient, easy to build, and competitive. Additionally, we benefit from a unified software architecture. Both the SPA-2 and SPA-3 platforms are based on the same central software master, which means we don't have to start from scratch with each new platform. The digital possibilities created by this are enormous - from advanced ADAS systems to seamless over-the-air updates.
You mentioned cost and efficiency advantages: How long does it take to develop a model on the new platform compared to previous development times?
With the introduction of the SPA-3 platform and the Superset architecture, we have fundamentally changed our development processes. We are now in an environment that is very stable and efficient, and that is a crucial advantage. Previously, technological advances were often tied to the introduction of new models. This meant that every new technology had to go through a complete development cycle before it could be implemented in a vehicle. With the SPA-3 platform and the Superset, that is no longer the case. We can advance technologies in parallel with vehicle developments, which gives us enormous flexibility.
How fast?
The goal is not necessarily to drastically reduce development time in months - that could always be achieved with enough resources and money. Rather, it is about establishing a stable and predictable development rhythm. Our goal is to introduce a new vehicle model every year, update another model, and provide at least four software updates. This rhythm ensures that our products remain fresh and integrate the latest technology - both at launch and throughout the product's lifecycle. With this approach, we can not only respond faster to market demands but also ensure that our customers always benefit from the latest developments. The vehicle launch machine is organized to operate predictably and efficiently, optimizing both development and production times.
How exactly does the Superset help reduce production costs?
With the SPA-3 platform, we have focused from the beginning on optimizing production processes. An example of this is the introduction of megacasting in the rear area of the vehicle. This process allows us to produce large parts of the vehicle structure in a single cast, which reduces complexity, lowers production costs, and at the same time makes logistics more efficient. Another example is our integrated battery pack. This was designed to serve as a structural element of the vehicle without compromising maintainability. This not only saves weight but also costs and allows for more efficient packaging. As already mentioned, a fundamental advantage of the SPA-3 platform is that it was developed exclusively for battery electric vehicles. This has allowed us to align the platform from the outset for maximum cost efficiency and production friendliness. These optimizations are not only beneficial for us as manufacturers but also for customers, as they ultimately translate into more competitive prices and more efficient products.
What role do partners like Nvidia or Qualcomm play in the development of the Superset and the SPA-3 platform?
Anders Bell
Anders Bell is a Volvo veteran. He was only not working for the Swedish OEM between 2016 and 2021. During this time, Bell worked for Tesla as Senior Director of Engineering in Palo Alto and Berlin. Since his return, Bell has risen to Chief Engineering and Technology Officer. Since November 2024, the Swede has been a member of the board of Volvo Cars.
Nvidia and Qualcomm are undoubtedly among our most important partners, alongside many others, including traditional Tier-1 suppliers like Bosch, who also play a central role in the Superset. Regarding Nvidia, it is important to understand that the application of their SoCs and their Drive OS is an integral part of our architecture. It is not a system that is simply placed on top of a traditional domain-based architecture to control ADAS functions. Instead, it is a central part that enables the control, management, and driving of the vehicle. So, it is a true core computer system. That's why NVIDIA is an extremely important partner for us. We have already announced that we will continue to invest in this technology and our partnership. So far, we are using Orin with 254 teraflops, and with the SPA-3, we will switch to Thor, the new SoC from Nvidia with 1,000 teraflops, which shows the technological perspective we are pursuing for these fundamental technologies. Similarly significant is Qualcomm, which is a central hub for everything related to Android.
What specific advantages does this bring?
Through close collaboration with Qualcomm, we are able to stay at the forefront of Android Automotive OS and Google Automotive Services. Both Nvidia and Qualcomm are therefore crucial for our technological strategy. Of course, there are other partners as well, but with Nvidia and Qualcomm, we are closely aligned regarding their development roadmaps and the timing of our product launches. This does not mean that we always have to introduce the latest and most powerful generation of each technology. Rather, it is about keeping a common pace with the advancement of their silicon, functional, and production capabilities. As I like to put it: We must keep up with the speed of technology. Our development processes must progress at the same pace as those of the leading chip manufacturers worldwide.
What are your long-term goals for the Superset and the integration of advanced technologies like AI?
This is one of the great advantages of our partnership with Nvidia, which is among the leading companies worldwide in their field. This applies to both infotainment and connectivity as well as vehicle technology, where Nvidia is particularly strong. This partnership enables us to implement powerful edge applications - that is, applications directly in the vehicle. Additionally, through our collaboration with Qualcomm and Google, we have developed very strong capabilities to provide off-board applications or less latency-critical applications. I cannot answer in detail today what we will specifically use these capabilities for. However, it is important that we now have a platform that allows us to continuously provide increasingly powerful AI applications. This applies not only to today's use but is also an integral part of our future roadmaps. Another crucial point is the expansion of vehicle functions into the cloud. Our data infrastructure Odin, which is also based on Nvidia technology, plays a central role here and is closely integrated into our closed development cycle. The EX90 is our first proof point on this new Superset tech stack - the first vehicle on the road based on this platform. With it, the actual journey now begins.
