A Primer: Software Defined Vehicles

Automakers have been touting a new era of Software-Defined-Vehicles (SDVs) for several years. Tesla is widely considered the pioneer of this movement, having developed its vehicles from scratch with an intent to sell products that are as much a computing platform as a physical machine. According to Goldman Sachs, Tesla is already selling $3B per year of subscriptions, and software sales could account for up to 20% of their total revenue by 2030. 

Other OEM’s are looking to stake their claims with their own SDV architectures. GM has Ultifi; Volkswagen has Cariad; and Toyota has Woven. Fresh off a week at CES 2024, it’s clear to us at Viaduct that the SDV hype is only getting louder, and more and more companies are piling into the action. Tier 1 Suppliers such as Aptiv, Bosch, and Continental, just to name a few, are offering branded versions of SDV Architectures (SDVAs), while Blackberry, Google, Apple and Amazon are all jockeying for their piece of turf in this new technology stack. Lastly, there is no shortage of industry groups publishing ideas on reference architectures and standards. 

And there is almost no end to the content from media and consultants about how this SDV revolution is transforming consumer expectations, business models, and automotive value chains. With all this collective buzz we thought it would be useful to step back and share a plain-speak primer on the Software Defined Vehicle. If you are new to SDVs, you’ll find this a helpful introduction. If you are an SDV expert, you might find this an interesting synthesis of stuff you already know. (Stay tuned, though, because in later blog posts we will share our insights on the future of SDV.)

Part I: The SDV Layer Cake

The business and technology landscape for software defined vehicles is not easy to wrap your head around. That’s because it’s not really a single thing and it doesn’t really have a beginning (or an end). The term “software defined vehicle” is relatively new, coming into vogue just in the last 10 years or so (according to Blackberry). Many of the concepts have been around for much longer, though. Initial SDV concepts date back to early efforts by the industry to define more modern approaches to on-vehicle E/E system (electrical/electronic system) architectures defined by AutoSAR in 2002 and connected vehicle standards like OBDII and J-1939 from the 90’s. 

To lay this out in the simplest terms possible, we can borrow some existing work from the strategy pundits that analyze this space (i.e. the strategy consulting firms). BCG, with some help from the World Economic Forum, describes software defined vehicles as a 6 layer cake starting with the E/E systems on the vehicle and extending up through to the broader “mobility ecosystem” that a vehicle might interact with. 

Source: “Unlocking Safety and Innovation in Vehicle Software”, June, 2023, World Economic Forum with Boston Consulting Group.

Layers 1-2: Vehicle and Computational Hardware

Moving down the layer cake, vehicle OEMs and vendors of hardware components dominate the landscape and are competing aggressively to outperform each other on speed, energy consumption, and cost. Traditional E/E System and component vendors are trying to shape the Vehicle Software platform standards in order to position themselves to deliver critical platform functionality in the emerging marketplace. The OEMs see the strategic importance of Vehicle Software platforms and are trying to develop these core competencies in-house.

Layer 3: Vehicle Software Platform

The Vehicle Software platform is the heart and soul of the system, and appropriately, where most of the SDV discussion has historically been focused. This is the layer with arguably the most interdependence between hardware and software, the most risk related to security and data privacy, and the area most foreign to the legacy players in the industry. As a result, this is also the layer with some of the biggest challenges that need to be addressed. 

There are at least 5 different formal industry collaborations focused on this layer, the most prominent being AUTOSAR and SOAFEE. Other organizations of note include eSync Alliance, COVESA, and Eclipse. Membership in these groups includes OEMs, legacy Tier 1 and Tier 2 suppliers, new suppliers of ADAS components, mobile platforms (e.g. Blackberry), and IT/software vendors looking to stake out positions in the industry. With significant overlap in membership across these groups, each has its particular focus in one or more problem areas within the Vehicle Software platform and extending to layers above and below. 

Layers 4-6: From Data to Smart Mobility

Moving up the layer cake from Vehicle Software platforms, IT/software vendors, cloud platforms, and the traditional mobility players such as fleet management companies, rental car companies, and government agencies are anticipating and trying to project and protect their role in the new world that is emerging because of SDVs. OEMs are also big players in these spaces, as they attempt to convert their traditional strengths in manufacturing and distribution into digital platforms for differentiated services and subscription revenues.  

Part II: SDV Applications

SDVs are supposed to unlock a range of use cases through applications that will deliver improved safety, higher performance, more functionality, and lower costs. These applications can be thought of as vertical slices of the layer cake, since each depends on functionality that permeates every layer of the stack.

Different organizations might slice the cake in different ways depending on their objectives, but a 5-slice cake composed of ADAS, Infotainment, Powertrain, Chassis, and Body seems to work pretty well for a couple of different reasons. First, these slices align with how the OEMs divide up “domains” from a system engineering perspective. Each domain puts different requirements on the vehicle hardware and software platforms so it’s common to consider them separately in the design process. The slices also align well with how drivers experience the vehicle and how they judge their satisfaction with the product. This user-in perspective is useful for evaluating the relative value of different SDV applications. 

As we did in part I of this series, we borrow again from one of the strategy pundits, namely Mckinsey, to illustrate and analyze the 5 slices. Per a paper published by Mckinsey and the Global Semiconductor Alliance (GSA) in 2023, the landscape can be divided as shown in the illustration:

Each application space (or domain) has different computational, security, and communication requirements that are dictated by functionality being offered. In the Mckinsey figure there is a pretty clear dividing line between the 3 application slices on the left and the two on the right. The ones on the right represent what you might call the “marquee apps” of an SDV platform, ADAS/AD and Infotainment. These are the ones that capture the imagination of the marketplace and also, incidentally, have the most challenging technical requirements. The 3 application slices on the left you might call the “core apps” of an SDV platform. These are not as sexy, but they are critical to the vehicle’s core purpose, which is to provide a source of transportation. Their technical requirements are more within reach of current and emerging SDV architectures. 

For the rest of this post we’ll briefly look at the challenges and opportunities inherent in each of these application “slices”.

SDV “Marquee Apps” - ADAS/AD and Infotainment

For many people autonomous driving is synonymous with the Software Defined Vehicle revolution, not just because it captures the imagination, but because it has the potential to completely upend how we get from point A to point B and who will be in control of the journey. But while ADAS/AD is making steady forward progress, it still has quite a ways to go. According to Deloitte: 

“Self-driving vehicles may be taking longer to make it to the streets than many had originally forecasted…. by 2035, we could see advanced autonomous technology and further pilots around autonomous services on their way…[but], they must still overcome a variety of barriers before they can safely share city blocks with human drivers.” (Source: The future of automotive mobility to 2035: What might mobility providers’ role be in tomorrow’s value chain? February, 2023, Deloitte). 

Nearly as compelling as ADAS is the promise of turning our vehicles into entirely new information and entertainment platforms (aka infotainment or IVI) to compete with mobile phones and in-home streaming platforms. The vehicle infotainment space receives a lot of attention from the popular press because of the potential for cross-pollination with non-automotive media platforms, which could create enormous new opportunities and competitive threats. From both a technology and a competitive perspective, this is not a domain where automotive companies have thrived, so it makes sense that a lot of energy is going into thinking about this solution space from both inside and outside the traditional automotive value chain.

SDV “Core Apps” - Body, PowerTrain, Chassis

Many SDV Core apps are already starting to appear in our daily lives. Every commercial and retail OEM is working to deliver a more personalized and automated in-cab experience, which is enabled by more modern SDV platforms. Biometrics can be used to unlock the vehicle and identify the passengers; seat positions and temperature settings adjust for the specific person in the driver seat; and mobile apps allow people to unlock doors and preheat cabins before they get near the car. 

Similar applications are also becoming visible in the Chassis and Powertrain areas, but these features are not coming as quickly and are often not as evident to the typical driver. These apps are deeply embedded in the vehicle platform, with complex and mission-critical electrical, chemical and mechanical system interactions. Historically, the software that has been used to control these systems has been very tightly coupled to hardware, so personalization or dynamically updating functionality has not really been possible. 

With the advent of EV propulsion, the model for Powertrain and Chassis systems is getting flipped on its head. EV battery management systems demand more dynamic, adaptable software-based controls, so the hardware-constrained model of the past is now giving way to software-defined powertrains. The same model is spilling over to the Chassis applications as well. As an example, on some vehicles it is now possible for a driver to choose whether they want their vehicle to handle more like a smooth-riding sedan or a formula 1 race car. 

OK, Great, I get it!  What’s Next?

In this blog, we have established a reasonably clear framework for understanding what SDVs are and why anyone should care. The framework is described by a horizontal layer cake, cut into 5 chunky vertical slices. It looks something like this:

In future blog posts, we will rely on this framework to start discussing some key questions about the current state and future of the SDV revolution. So please, stay tuned …

More articles