Momenta, XHEART and QNX Launch ASIL D Autonomous Driving Platform
Momenta, XHEART, and QNX have jointly developed a production-ready Physical AI autonomous driving platform that combines full-stack automated driving software, AI-native automotive silicon, and a safety-certified operating system.
The platform integrates Momenta’s autonomous driving stack, XHEART’s X7 automotive SoC, and QNX OS for Safety, built on QNX SDP 8.0.
The resulting system is certified to ISO 26262 ASIL D, the highest Automotive Safety Integrity Level defined by the functional safety standard.
For automakers, the collaboration addresses a central challenge in advanced driver-assistance and autonomous driving systems: increasing AI compute requirements while maintaining the rigorous functional safety requirements needed for production vehicles and highly regulated international markets.
🚗 Three-Layer Platform Combines AI, Silicon and Safety #
The platform brings together three complementary technologies.
| Layer | Technology | Primary Role |
|---|---|---|
| Autonomous driving | Momenta full-stack solution | Perception, planning and driving intelligence |
| Compute | XHEART X7 SoC | Automotive AI and high-performance processing |
| Operating system | QNX OS for Safety | Functional safety foundation and system isolation |
This architecture reflects a broader shift in automotive computing.
Modern intelligent vehicles require significantly more compute than traditional electronic control architectures, but additional AI capability cannot come at the expense of safety certification and deterministic system behavior.
The three companies are therefore targeting the intersection of AI performance, functional safety, and production scalability.
🛡️ ISO 26262 ASIL D Certification Sets the Safety Baseline #
The platform has been certified to ISO 26262 ASIL D by TÜV Rheinland.
ASIL D represents the highest functional safety integrity level within ISO 26262 and is associated with systems where failures can have potentially severe consequences.
For autonomous driving platforms, achieving this level of certification can simplify one of the most difficult stages of vehicle development: integrating increasingly complex AI workloads into a safety-critical production architecture.
Why ASIL D matters for intelligent vehicles #
AI-based perception and decision systems introduce substantial software and computational complexity.
At the same time, automotive OEMs must establish that safety-critical functions remain appropriately controlled even when:
- AI workloads fail
- Software components behave unexpectedly
- Hardware experiences faults
- External inputs become unreliable
- Individual services malfunction
- System resources become constrained
A safety-certified operating system can provide foundational mechanisms for managing these risks as part of the broader vehicle safety architecture.
Certification does not automatically make every autonomous driving function safe. OEMs still need to perform system-level safety analysis, verification, validation, and certification appropriate to their vehicle programs.
However, beginning with an ASIL D-certified foundation can significantly reduce the amount of foundational safety engineering required from individual vehicle developers.
🧠 Momenta Provides the Physical AI Layer #
Momenta contributes its full-stack autonomous driving technology to the platform.
The company’s approach is centered around Physical AI, where machine learning systems perceive and reason about the physical world and translate those capabilities into real-world vehicle behavior.
Its technology stack is intended to provide the intelligence layer required for automated driving applications.
From perception to vehicle control #
A production autonomous driving platform must coordinate multiple computationally intensive functions, including:
- Environmental perception
- Object detection and tracking
- Sensor fusion
- Scene understanding
- Prediction
- Path planning
- Decision-making
- Vehicle control
These workloads increasingly rely on large AI models and high-performance inference hardware.
This makes the underlying compute and operating-system architecture increasingly important as autonomous driving systems move toward more sophisticated AI-based approaches.
⚙️ XHEART X7 Provides AI-Native Automotive Compute #
XHEART contributes its X7 automotive-grade SoC, designed specifically for high-performance intelligent computing.
The company positions the chip as an AI-native compute foundation for applications built around large models.
Autonomous driving is one of the most demanding use cases for such hardware.
The SoC must support substantial AI workloads while operating within automotive constraints involving:
- Thermal budgets
- Power consumption
- Reliability
- Real-time responsiveness
- Hardware redundancy
- Functional safety
- Long product lifecycles
The X7 is therefore intended to bridge the gap between large-scale AI compute requirements and automotive-grade deployment constraints.
Silicon becomes increasingly important for Physical AI #
As autonomous driving models become larger and more sophisticated, the underlying semiconductor architecture increasingly determines what can be deployed inside a vehicle.
Traditional automotive processors were primarily designed around deterministic control workloads.
Physical AI introduces a very different computational profile, involving neural-network inference, sensor processing, multimodal perception, and increasingly complex decision models.
Purpose-built AI silicon can provide the required computational throughput while allowing the platform to be designed around automotive power and safety requirements from the beginning.
🖥️ QNX OS for Safety Provides the Software Foundation #
QNX contributes QNX OS for Safety, built on QNX SDP 8.0.
QNX has extensive experience providing operating-system technology for safety-critical embedded systems, making its software particularly relevant to automotive applications where reliability and functional safety are core requirements.
The operating system serves as the foundational software layer connecting the automotive hardware with higher-level autonomous driving applications.
Safety and AI workloads must coexist #
One of the key architectural challenges in autonomous vehicles is allowing computationally intensive AI workloads to coexist with safety-critical software.
A production vehicle cannot simply treat every process as equally trusted.
Safety-critical functions need appropriate isolation, resource management, and failure containment so that a fault in one software component does not automatically compromise unrelated functions.
This becomes increasingly important as vehicles consolidate more workloads onto centralized or high-performance compute platforms.
🌍 Designed for Global Automotive Deployment #
The partners are positioning the platform for global OEM deployment rather than limiting it to a single domestic market.
This is significant because autonomous driving regulations vary considerably between jurisdictions.
A platform designed around recognized international functional safety requirements can provide automakers with a stronger foundation for adapting vehicle systems to different regulatory environments.
The companies specifically reference applications relevant to European regulatory requirements, including models designed to comply with UN R171 Driver Control Assistance Systems (DCAS) regulations.
Functional safety is becoming a global engineering requirement #
Automakers expanding intelligent driving products internationally need to satisfy increasingly sophisticated regulatory and safety requirements.
That means automotive AI platforms must address more than raw inference performance.
They must also provide evidence of:
- Functional safety
- Software reliability
- Hardware reliability
- Fault handling
- System-level validation
- Traceability
- Regulatory compliance
A certified foundation can therefore become a competitive differentiator for autonomous driving suppliers.
📈 The Platform Targets Production Scale, Not Just Demonstration Systems #
A notable aspect of the collaboration is its emphasis on production readiness.
The goal is not simply to demonstrate that AI models can operate on an automotive SoC.
The platform is designed to provide automakers with a reusable foundation for deploying intelligent driving systems across vehicle programs.
This distinction is important.
A research prototype can tolerate custom hardware, experimental software, and manually managed infrastructure.
A production automotive platform must instead support:
- Long-term software maintenance
- Hardware qualification
- Functional safety processes
- Vehicle integration
- Manufacturing requirements
- Regulatory approval
- Software updates
- Multiple vehicle configurations
The combination of Momenta’s software, XHEART’s silicon, and QNX’s safety-certified operating environment is intended to address these requirements as an integrated platform.
🔗 Vertical Integration Could Accelerate Automotive AI Deployment #
The collaboration also illustrates a broader trend toward vertically integrated Physical AI platforms.
Instead of treating autonomous driving software, AI processors, and operating systems as independent components, the partners are coordinating all three layers.
This creates potential advantages in system optimization.
Hardware-software co-design #
Momenta can optimize autonomous driving workloads around the capabilities of XHEART’s X7.
XHEART can tune its silicon around the computational requirements of advanced driving models.
QNX provides the safety-oriented operating environment needed to integrate these workloads into production automotive systems.
This creates a feedback loop between:
AI models → algorithms → compute architecture → operating system → vehicle platform
Such co-design can become increasingly valuable as autonomous driving workloads become more computationally demanding.
🧩 Safety Certification Does Not Eliminate System-Level Validation #
Although ASIL D certification is a major milestone, it is important to distinguish component or platform certification from complete vehicle-level safety certification.
An automotive OEM remains responsible for validating how the platform is integrated into a specific vehicle.
That includes analyzing the complete system across:
- Sensors
- Actuators
- Vehicle networks
- Power systems
- Compute hardware
- Autonomous driving software
- Human-machine interfaces
- Redundancy mechanisms
- Operational scenarios
The safety case must ultimately account for the behavior of the complete vehicle system.
Therefore, the value of an ASIL D-certified foundation lies in reducing foundational safety risk and engineering effort rather than eliminating the OEM’s certification responsibilities.
🎯 Physical AI Is Moving Toward Production Infrastructure #
The collaboration between Momenta, XHEART, and QNX reflects a broader transformation in autonomous driving.
The industry is moving from relatively specialized driver-assistance algorithms toward increasingly general AI systems capable of interpreting complex physical environments.
That evolution creates three simultaneous requirements:
- More AI compute to process increasingly sophisticated models.
- More capable software architectures to deploy those models efficiently.
- Stronger safety foundations to ensure the resulting systems can operate reliably in production vehicles.
The new platform attempts to address all three.
🧭 A Full-Stack Approach to Safe Autonomous Driving #
Momenta, XHEART, and QNX are combining autonomous driving intelligence, AI-native silicon, and safety-certified operating-system technology into a single production-oriented platform.
The XHEART X7 provides the computational foundation, Momenta contributes the autonomous driving intelligence, and QNX OS for Safety provides the functional-safety-oriented software foundation.
Its ISO 26262 ASIL D certification is particularly significant for automakers targeting international markets where functional safety is a fundamental requirement.
The larger implication extends beyond this individual platform.
As Physical AI becomes more sophisticated, successful autonomous driving systems will increasingly depend on close coordination between models, silicon, operating systems, and safety engineering.
The ability to combine high-performance AI compute with production-grade functional safety could become one of the defining requirements for the next generation of globally deployed intelligent vehicles.