As the automotive industry seeks to
move from SDV vision to series implementation, the pressure is shifting
from technology availability to integration capability, governance and
long-term platform discipline. Open software foundations, scalable ecosystems
and consistent operating models are becoming decisive factors in turning software-defined vehicle strategies into industrial
reality.
Dr Thomas Irawan, President of ETAS and Chairman of the ETAS
Board of Management, will address these questions together with Dr Christian
Salzmann of BMW at the Automobil-Elektronik Kongress in
Ludwigsburg. In their joint presentation on Eclipse
S-CORE, the focus is on speed, efficiency and community in open source,
as well as on the path from shared software foundations to deployable
automotive platforms.
Ahead of the event, we spoke with Dr Irawan about the
operating model, platform decisions and ecosystem
alignment needed to scale SDV and AI strategies.
Looking ahead three to five years, what will be the
biggest bottleneck in turning SDV and AI strategies into scalable,
industrialized vehicle platforms?
The biggest bottleneck will not be technology availability. The core technologies for software-defined and AI-enabled
vehicles are largely already there. The real bottleneck will be the industry’s
ability to align its operating model and ecosystem to industrialize these
technologies at scale. Over the next three to five years, success will
depend less on who has the most advanced individual AI use cases and more on
who can bring architecture, organization, and partners into one scalable
execution model. In other words, the challenge is not innovation itself, but
coordinated industrialization.
That became very tangible to me during my recent
trip to China. What stood out was not only the pace of development, but also a
different architectural logic and approach. In many Western automotive
programs, the industry is still moving from a hardware-centric legacy toward
the software-defined vehicle, with AI often added as a further, mainly feature-based
layer on top. In China, some leading Chinese OEMs are already approaching the
vehicle more from an AI-native perspective. There, we increasingly see AI
requirements influencing software architecture decisions, with the software
architecture then defining the hardware. I would not frame this as a question
of one side being right and the other being wrong. It is largely a reflection
of different starting points and different legacy constraints.
But it does
highlight a decisive shift: future competitiveness will depend less on adding
intelligence to existing systems and more on designing hardware, software,
data, and AI as one coherent architecture from the outset. And that is exactly
where the real bottleneck lies. To succeed in that environment, companies need
more than strong individual technologies. They need a much more
software-centric operating model, earlier cross-functional decision-making,
faster integration loops, and much tighter collaboration across OEMs,
suppliers, semiconductor players, cloud providers, and infrastructure partners.
In the SDV era, ecosystem performance has become a core competitive capability.
If the ecosystem is misaligned, scalability breaks down – in speed, cost,
quality, and lifecycle management. If it is aligned, companies can turn
innovation into repeatable, industrialized platforms. So my view is very clear:
the biggest bottleneck in the next three to five years will be operating model
and ecosystem alignment – the ability to align architecture, governance,
development processes, and partners around a common software and AI foundation.
Which decision being made today will most strongly
determine where value is created in the future automotive ecosystem?
The defining decision is this: how an OEM defines where to
differentiate and where to standardize. The strategic question is no longer
simply who owns the software. The real question is who controls the parts of
the vehicle experience that truly define the brand. That is where future value
will be created. The winners will not be the companies that try to build every
software layer themselves. They will be the ones that standardize the
non-differentiating base and focus their own resources on what makes the
vehicle unique – from driving behavior and automated functions to user
experience, comfort, and AI-enabled services. This is why open,
production-ready foundations matter so much.
If OEMs stop reinventing the basic
software stack – middleware, operating system layers, communication frameworks,
hardware abstraction – they can direct scarce engineering capacity toward the
functions the customer sees and feels. So, in my view, OEMs should not try to
be coders of everything. They need to be architects of the whole. They should
leverage strong partners and open ecosystems for the commodity layers, while
retaining control over the functional logic that defines the character of the
vehicle. In simple terms: standardize the base, protect the differentiation.
That decision will determine where value sits in the future automotive
ecosystem.
Where do current approaches to SDVs and next-generation
E/E architectures still fall short in real-world programs?
Current approaches to SDVs and next-generation
E/E architectures do not fall short in ambition. They fall short in how
they are integrated, industrialized, and sustained over time. One major
weakness in real-world programs is still late integration. Hardware, base
software, middleware, applications, and cloud-related components are often
developed in parallel, but not integrated early enough against a stable
architectural target. As a result, critical dependencies, performance issues,
and interface conflicts surface only late in the program, when they are far
more expensive and disruptive to resolve. That creates delays, rework, and
significant execution risk.
A second issue is fragmented stacks. Many current
SDV approaches are still built from a patchwork of vendor-specific solutions,
hardware-bound software layers, and partially incompatible frameworks. That may
work at prototype level, but it becomes a serious obstacle when companies try
to scale across platforms, vehicle lines, and generations. Fragmentation
reduces reuse, increases validation effort, and makes the overall stack harder
to evolve in a consistent way. And that leads directly to the third gap:
lifecycle burden. In many cases, the stack is designed primarily for SOP, but
not sufficiently for the full lifecycle that follows.
Once software has to be
maintained, updated, secured, certified, and supported over many years,
architectural fragmentation turns into a major cost and complexity driver. What
looks manageable at launch can become extremely burdensome over a 10- to
15-year vehicle lifecycle. So my view is this: current approaches often fall
short because integration happens too late, the technology stack remains too
fragmented, and the long-term lifecycle burden is still underestimated. The
real challenge is not defining the SDV vision – it is building a stack that can
be integrated early, scaled consistently, and managed efficiently over time.
What defines a scalable software platform for SDVs in
practice?
In practice, a scalable software platform for SDVs must
enable consolidation without compromising safety. That is the real benchmark.
OEMs cannot continue to manage software complexity by simply adding more ECUs,
more interfaces, and more wiring. A scalable platform must allow diverse
functions to run on centralized compute in a controlled, secure, and
maintainable way. From the ETAS perspective, that requires three things.
First,
strong isolation. If safety-critical functions such as braking, steering, or ADAS
coexist with infotainment and AI-driven services on the same hardware,
separation is non-negotiable.
Second, intelligent middleware. Middleware is the backbone that connects applications,
services, and communication across domains. It enables safe interaction,
supports the safety concept, and creates the basis for reuse and portability.
Third, deterministic connectivity. In zonal and centralized architectures,
scalability depends on predictable communication behavior, especially where
safety-relevant data and timing requirements are involved. The real goal is
safe consolidation at industrial scale. If that is achieved, OEMs can reduce
complexity, improve reuse, accelerate updates in non-safety domains, and build
a platform that remains manageable over the full vehicle lifecycle.
That is
exactly what we are building at ETAS: the production-ready software foundation
that turns the SDV vision into a safety-certifiable reality.
Where do current software approaches still struggle to
deliver reuse and long-term maintainability?
They still struggle wherever software is not treated as a
long-term industrial asset. The first barrier is hardware dependency. Too much
software remains too closely tied to specific silicon or platform decisions. As
soon as the hardware changes, reuse breaks down and teams are forced to port or
rebuild far more than necessary. The second barrier is lifecycle discipline. In
automotive, software must remain maintainable for well over a decade. That
requires reproducible toolchains, robust version management, and update
capability long after the original development environment has changed. The
third issue is limited transparency.
When there is insufficient transparency
across interfaces and dependencies, integration slows down, flexibility is
reduced, and OEMs become dependent on external roadmaps even for minor changes.
And finally, many programs still accumulate technical debt because they are
optimized for launch timing rather than modularity and long-term evolution.
That reduces reuse across vehicle lines and makes systems harder and more
expensive to maintain over time.
So the core issue is this: the industry has
made strong progress in adding more software, but not enough progress in
industrializing software. Only with open, standardized foundations and stronger
hardware-software decoupling can software become truly reusable, maintainable,
and scalable across generations.
What has been the biggest unexpected challenge in turning
S-CORE into an industrial-grade platform?
The biggest challenge – and perhaps the one many
underestimated – was not building an open foundation fast enough. It was
turning that open foundation into something the automotive industry can
actually deploy in series programs at scale. Open
collaboration creates speed. But in automotive, speed alone is not enough. OEMs
need traceability, stable versioning, integration quality, safety and
cybersecurity readiness, and long-term maintainability. That is what
makes the difference between community code and an industrial platform.
That is
also where ETAS adds value. We see S-CORE as a shared open foundation for the
industry. But customers need more than access to open-source components. They
need a production-grade distribution that is consistently packaged, traceably
versioned, integrated into industrial toolchains, and aligned with modern
development and validation processes. This is why we are bringing S-CORE into
the ETAS Vehicle Software Platform Suite.
Our role is not only to contribute to
the open ecosystem, but to industrialize it – and to help customers turn an
open middleware foundation into something deployable, maintainable, and
scalable across the full vehicle lifecycle. The biggest challenge in one
sentence would be this: transforming a promising open initiative into a trusted
industrial-grade platform for automotive series production.
