As vehicles become increasingly software-defined, the challenge is no longer adding more AI features- it’s creating an architecture that allows them to work together.
In this interview, Natalie Stormanns, Head of Infotainment and Digital Experience – Associated Partner at MHP – A Porsche Company, explores why unified AI stacks, centralized compute, and integrated vehicle and cabin data are becoming essential to delivering safer, more intelligent, and more personalised driving experiences.
1. What architectural barriers are preventing OEMs from achieving a truly unified AI stack today?
The biggest barrier is the legacy domain-centric vehicle architecture that has evolved over decades. Infotainment, ADAS, connectivity, body electronics, and cloud services have largely been developed independently- each with its own toolchains, data models, suppliers, middleware, and lifecycle management.
This leads to three key challenges:Â
- Fragmented AI stacks: Separate AI stacks for ADAS, Smart Cabin, and cloud-based services prevent consistency and scalability.Â
- Inconsistent data foundation: Vehicle, cabin, and cloud data are not harmonized, even though cross-domain data is the basis for AI value creation.Â
- Safety, middleware, and platform dependencies: OEMs must manage mixed-criticality workloads, proprietary middleware, SoC ecosystems, and supplier IP boundaries.Â
In addition, many organizations lack a clear AI target operating model – including stack ownership, model lifecycle management, validation, use-case prioritization, and edge/cloud orchestration.Â
 In short: A unified AI stack is not primarily a technology challenge- it is an architecture and governance challenge. Success requires a holistic approach that combines platform architecture, data strategy, organizational ownership, and safety-by-design principles.Â
2. How can automakers balance centralized compute with the differing safety requirements of infotainment and ADAS workloads?
The core challenge is balancing:Â
- Safety-critical systems such as ADASÂ
- User experience-driven systems such as infotainment and Smart Cabin AIÂ
A key solution is logical separation combined with physical consolidation:Â
- Central high-performance compute platforms enable shared compute resourcesÂ
- Virtualization, partitioning, and safety mechanisms ensure strict separation of safety levelsÂ
Currently, two main approaches can be observed:Â
- Single-compute approach: Shared HPC for ADAS and cabin workloads with strong isolation, deterministic resource allocation, and safety supervisionÂ
- Dual-compute approach: Separate systems optimized for specific requirements, especially where ADAS safety and infotainment flexibility are difficult to combineÂ
In the long term, architectures will likely converge toward:Â
- Strict isolation of safety-critical workloadsÂ
- More dynamic scaling of non-safety AI workloadsÂ
- Controlled sharing of GPU/NPU resources under clear safety and resource constraintsÂ
The real differentiator is not centralized compute itself, but the software architecture that orchestrates it. Hypervisors, middleware, resource scheduling, data brokers, and lifecycle management collectively determine whether consolidation can be achieved without compromising safety.Â
3. What opportunities emerge when cabin data and vehicle data can be analysed together in real time?
Combining vehicle and cabin data elevates AI from reactive features to a context-aware, proactive system.Â
Key opportunities include:Â
 Holistic context intelligence:Â
- Vehicle state, road context, and driver state can improve real-time decisions, for example in handover or warning strategies.
Personalized user experiences:
- User preferences, occupancy, driving context, and interaction behaviour enable adaptive HMI, comfort, media, and climate functions.
Enhanced safety:
- Vehicle dynamics, environmental perception, driver monitoring, seatbelt status, and occupant position can enable earlier and more targeted intervention.Â
Energy and operational optimization:
- Cabin occupancy, HVAC demand, battery state, route, and driving behaviour can improve range, comfort, and maintenance recommendations.Â
New business models:Â
- Context-aware digital services and predictive services become more relevant when based on real vehicle and user context.
The fundamental shift is that the vehicle evolves from a collection of independent functions into an intelligent, continuously learning system whose capabilities improve through governed data integration, AI model evolution, and OTA updates.Â
4. Which AI use cases are most likely to deliver measurable customer value in the next generation of software-defined vehicles?
The most successful use cases are those that deliver direct, visible, and measurable customer value, not just technological sophistication.Â
Four key categories stand out:Â
Predictive & data-driven services
- Predictive maintenanceÂ
- Vehicle health monitoringÂ
- Service and uptime optimizationÂ
→ Measurable value through fewer breakdowns, reduced workshop visits, and lower operating cost.
Intelligent assistants
- GenAI-powered voice assistants
- Vehicle-context-aware troubleshooting and function guidance
- Proactive interaction based on route, vehicle state, and user intentÂ
→ High usage frequency and strong UX impact if deeply integrated into vehicle functions.
Safety & monitoring
- Driver state detection intervention
- Occupant monitoringÂ
- Integration with ADAS for proactive safety
→ Direct impact on safety, trust, and regulatory or rating performanceÂ
Personalization & Smart Cabin
- Adaptive UI/UXÂ recommendations
- Individualized comfort, media, climate, and interaction behaviourÂ
- Context-based automationÂ
→ Strong brand differentiation when reliable, subtle, and genuinely usefulÂ
 Ultimately, customers do not buy AI- they buy safer driving, greater convenience, lower operating costs, and better experiences. The winning AI use cases will therefore be those that solve real customer problems rather than simply demonstrating advanced technology.
Interested in exterior vehicle sensing technology?
With a pass to InCabin Europe, you’ll also get full access to our
co-located sister event, AutoSens. Find out more here >>
Want more inspiration? Check out more InCabin interviews here ⬇
