The Software-Defined Vehicle: Enabling and Delivering its Benefits
February 2021
Pioneered by Tesla, the concept of the “software-defined vehicle" will quickly become mainstream. It is already deeply transforming the industry. Benefits include the easy deployment of new features, constant user experience improvement, the creation of new revenue streams, hardware standardization, easier servicing, cheaper recall campaigns, and less of a need for model refresh.
How is the industry making this strategic transition? How much software development will OEMs and Tier 1 suppliers undertake in-house? What role will tech companies play?
Benefits of Software-Defined Vehicle and Rapidly Changing Hardware Paradigm
Tesla showed the way to the software-defined vehicle. Introduced in 2012, the Model S continues to benefit from regular upgrades and updates, ranging from range, to braking performance, user interface or infotainment. Besides all this, all Teslas are fitted with the hardware (sensors and compute) reportedly adequate for tomorrow’s Level 4 autonomous driving, whereas the software (SW) is regularly updated to incrementally bring it closer to real autonomy.
This approach will also enable new business models leading to recurring revenue. We will buy a package for a track day (e.g. power, handling…), heated seats for a ski outing, or extra range and autonomous driving features for long trips. Software will be sold on the OEM’s “store” yet could be designed by third parties — Apple takes a 30% cut on its app store. The extra cost of deploying all hardware (HW) across all vehicles will have to be compensated by lower diversity management cost, reduced inventory all the way to dealers and above all, new recurring revenue.
This revolution on the SW side is accompanied by a deep transformation of the electronic architecture. The industry is progressively — and faster than expected — shifting from dozens of function-specific ECUs per vehicle — which cannot be updated — to half a dozen domain computers that are SW agnostic and updatable over-the-air. The decoupling of SW vs. HW (a.k.a. HW abstraction) enables the latter to be commoditized, though at automotive grade. It also allows SW players to enter the automotive market, whether they be tech giants or startups.
In addition, efficient SW development will lead to shorter overall development / deployment cycles. OEMs will be able to avoid creating very detailed specification for the complete feature set. Rather, they will frequently launch new SW versions to improve existing features, deliver new ones, or fix bugs, as Tesla already does. Customer feedback loops will become critical to test beta versions on select groups and continuously improve customer experience.
Software Development In-house or Outsourced, a Strategic Decision
Once the HW is abstracted, SW development comes down to two distinctive blocks. With its modular structure, the middleware layer (incl. operating system) interfaces SW-agnostic HW with applications, and can be commoditized (e.g. with AUTOSAR) as it is transparent to users. Conversely, applications are service specific. They can contribute significantly to brand differentiation as they are often related to user experience. Altogether, the software stack is expected to reach 300 to 500 million lines of code for automated vehicles.
OEMs are starting to source compute HW (high-performance computers) and SW separately — the market for software-intensive electronic systems is expected to have a 15% CAGR between now and 2030 according to Bosch. To develop their SW, OEMs face a major strategic decision, which can be brought down to three options:
-
develop in-house a significant portion of the SW needed across most (or all) domains, building a broad range of expertise,
-
develop in-house SW for one or two deeply differentiating domains and specify, outsource development, integrate and validate the balance of their SW needs,
-
outsource all SW development and services to specialized outfits, while keeping feature specification, integration and validation in-house.
Option 1: Turning OEMs into Software Powerhouses
This “vertical integration” option secures the highest level of independence — and likely higher long-term financial performance. However, it requires a deep cultural shift, significant agility and massive SW development capabilities. A few OEMs have selected this path and started to deploy the necessary organization.
A significant volume base is needed for incumbents to justify engaging such an initiative given the need to amortize high, fixed-development costs. This is most adequate for companies like the VW Group or Toyota, which respectively launched the Car.Software and Woven Planet / Arene initiatives. Conversely, such an approach can be more easily justified for emerging OEMs (e.g. Rivian, Lucid) despite their small size as SW development is integral to the way the business is built from the start.
Created in Jan 2020, VW Group’s Car.Software business unit is charged with the goal of eventually developing in-house 60% of the group’s SW needs by 2025 vs. 10% today. This entity is supposed to grow to 10,000 SW experts by 2025. However, the strategy proved more difficult to implement than expected. VW ID.3, the group’s critical first mass EV, was the first to host the new OS. Its launch in 2020 was delayed by several months due to SW issues.
Option 2 : Selectivity and Agility to Deliver Software
OEMs will likely select strategic domains for which they will develop SW in-house. The electric powertrain / battery management system is a likely target for many OEMs. For other domains such as ADAS / autonomous driving or infotainment, they will likely outsource SW development, yet still source the HW separately.
Companies pursuing this option must rely on a network of partnerships with a range of SW providers. These likely offer domain-specific expertise such as electrification, autonomous driving or infotainment, or function-specific expertise such as connectivity, cybersecurity, edge computing or cloud services. Nevertheless, in-house SW expertise remains critical here, at least to specify solutions, design HW and SW architectures, integrate and validate.
I suspect Renault has selected this option and decided to focus their SW development efforts on electrification given their early start (battery EV Zoe was launched in 2012). Besides, they were the first OEM to partner with Google Automotive Services’ (GAS) for embedded infotainment — GAS has since been adopted by Polestar and Ford. Renault also recently announced “Software République,” an ecosystem-type initiative aimed at providing the OEM access to core knowhow and expertise, such as big data, AI or cybersecurity.
Option 3: Full Software Development Outsourcing
This option will likely be relevant for small OEMs with limited ability to build SW development expertise in-house. Their role may be constrained to specifying feature sets as well as integrating and validating full systems. They may also be prime adopters of the following alternative.
Coopetition: Cross-Industry Licensing
An alternative is emerging that applies to all options: licensing among OEMs in a coopetitive mode. The VW Group has announced they could make their car OS available to competitors. Likewise, Tesla recently said it is open to licensing its autonomous driving tech to other OEMs.
Such an approach favors licensors in two ways. One, it can keep competition at arm’s length and preserve a technology/feature leadership, assuming pricing deters other OEMs from creating their own SW teams. Two, licensing income goes almost entirely to licensors’ bottom line as SW development is again largely a fixed-cost business.
Tier 1 Suppliers Are Also Adapting
As OEMs look outside for their SW development, Tier 1 suppliers must decide whether to capture this fast growing business or let it go to SW-native companies. Building on their unique relationship with OEMs and on existing SW expertise, some suppliers have made a decision to embark full steam in SW development.
Bosch is pulling together 17,000 associates working on compute HW and SW in a new entity called Cross Domain Computing Solutions. The objective is bring solutions faster and more efficiently to market.
Similarly, ZF will establish a Global Software Center in 2021, which will be responsible for developing SW systems for future architectures at group level, first focusing on middleware. However, the group does not intend to centralize SW development. Continental can rely in part on its Elektrobit entity (acquired in 2015). With its staff of 3500, the company develops middleware, applications and cloud architecture, and provides consulting services.
Tech Companies Become Strategic Partners for OEMs
Since January 2021, the VW Group, Bosch, GM/Cruise and Ford all announced partnerships with either Microsoft or Google. In all cases, the automotive player intends to leverage the tech giant’s cloud and edge computing services and its AI/analytics expertise.
Microsoft’s partnership with Bosch aims to enable wireless SW updates combining the Azure cloud and Bosch SW modules. The solution will be used first in vehicle prototypes by the end of 2021. VW Group’s collaboration with Microsoft is intended to build a cloud-based Automated Driving Platform on Azure in order to accelerate the emergence of autonomous driving. GM’s and Cruise’s relationship with Microsoft also aimsto focus on autonomous vehicles, leveraging Azure and the edge computing platform. Microsoft participated in a $2B round raised by Cruise.
On the Google side, Ford’s deal will enable cloud-based data analytics to create new business opportunities. Google’s inroads in automotive is not limited to cloud. As mentioned earlier, Google Automotive Services has signed infotainment-related deals with Renault, Polestar and Ford, leading to embedded assistant and navigation services, and more.
Automotive SW-focused companies will also play in increasing role. They include Elektrobit (part of Continental), ETAS (part of Bosch), Wind River or promising startups such as Silicon Valley-based Sonatus.
Marc Amblard
Managing Director, Orsay Consulting
Feel free to comment or like this article on LinkedIn. Thanks!