Lynx, a leader in safe and secure software platforms for mission-critical edge systems, today announced the general availability of LYNX MOSA.ic.AI, a unified CPU and GPU AI deployment platform purpose-built to enable deterministic, certifiable AI in mission-critical environments.

Combining the proven LYNX MOSA.ic deterministic CPU software platform and Lynx CoreSuite certified GPU software stack into a single, cohesive architecture, MOSA.ic.AI gives defense, avionics, and industrial programs a clear, actionable path to certifying AI-enabled systems under DO-178C and other related security and safety standards. The launch marks a significant strategic milestone and directly addresses what the industry has been calling the “AI deployment gap”: the structural divide between powerful AI development frameworks optimized for experimentation and the deterministic, certification-compliant execution discipline that regulated environments demand.

The Expanding Market for AI-Enabled Mission Systems

AI is rapidly becoming a foundational capability in next-generation mission systems across aerospace, defense, and autonomous platforms. Industry analysts estimate that the global AI in aerospace and defense market will exceed $25-30 billion by the early 2030s, driven by applications such as sensor fusion, autonomous navigation, predictive maintenance, and intelligent mission support. Edge AI deployments in robotics and industrial systems are expected to grow at more than 20% annually as organizations move AI capabilities from cloud experimentation into operational systems.

The AI Deployment Gap: A Structural Problem

The mainstream AI development ecosystem (built on frameworks such as CUDA, PyTorch, and TensorFlow) was designed for throughput, experimentation, and speed, not for the deterministic execution, analyzable system behavior, and certification discipline required in regulated environments.

As a result, many AI initiatives advance rapidly in the prototype stage but stall when teams attempt to deploy them in operational systems that must meet stringent regulatory standards. Deploying AI in these environments requires bounded execution timing, predictable memory behavior, and clear certification evidence across the system architecture.

Introducing LYNX MOSA.ic.AI: Deterministic AI Deployment for Mission-Critical Systems

LYNX MOSA.ic.AI is not a training platform and does not replace AI development frameworks. Instead, it operates downstream of model development as a deterministic runtime environment that enables AI workloads to meet the execution discipline required for security and safety certification.

Lynx has unified two previously independent product lines into a single, integrated platform for heterogeneous compute systems.

This unique unified architecture, enabled by Lynx’s unikernel, called LynxElement, allows CPU and GPU workloads to operate within a shared deterministic execution model, reducing integration complexity and enabling a consistent safety framework across the entire compute stack.

How MOSA.ic.AI Delivers Determinism

MOSA.ic.AI is distinguished by several unique architectural properties that alternative solutions cannot match, such as static execution models, fixed memory allocation with bounded execution timing, portability across heterogeneous hardware like ARM, AMD and Intel (not just NVIDIA/CUDA-dependent platforms), and support for running frameworks such as PyTorch and computational libraries such as BLAS. These capabilities eliminate runtime surprises that undermine certification, ensure program leads can move applications between hardware architectures without rebuilding certifications from scratch, and run mixed-level AI execution tuned to program requirements.

“Mission-critical edge programs have been watching AI transform adjacent industries while facing a wall when it comes to their own deployments. MOSA.ic.AI tears that wall down,” said Tim Reed, Chief Executive Officer and Board Member, Lynx. “The industry has been waiting for a platform that doesn’t ask programs to choose between the AI tools they know, and the certification discipline their systems require. With MOSA.ic.AI, Lynx delivers the deterministic execution foundation that makes both possible, turning what has been a multi-year certification challenge into an immediate, actionable path forward.”

Industry Support for LYNX MOSA.ic.AI

The shift to heterogeneous compute – combining CPUs, GPUs, and specialized accelerators – is gaining momentum across embedded and edge programs. But for mission-critical application developers, value depends on deterministic compute and a clear path to certification. MOSA.ic.AI helps close that gap by providing a unified execution environment across CPU and GPU workloads, built on open standards rather than proprietary dependencies,” said Chris Rommel, Executive Vice President, VDC Strategy.

"As AI/ML technology is incorporated into safety-critical systems, the ability to analyze and validate system behavior becomes increasingly important. Platforms that provide efficient workload processing and support sound AI/ML execution in alignment with performance and safety requirements make it easier for engineering teams to model, simulate, and verify system behavior as part of the development process. Lynx's CoreSuite libraries and MOSA.ic together with SCADE support AI/ML integration and facilitate fulfillment of both performance and safety requirements,” said Gunther Siegel, Sr. Director, R&D, Ansys, part of Synopsys.

“Edge AI systems increasingly combine real-time control workloads with high-performance AI inference. Enabling these workloads to run efficiently across CPUs and GPUs while maintaining predictable system behavior is a key requirement for next-generation mission platforms,” said Peter Muller VP Global Customer Application Center at Congatec.

MOSA.ic.AI is validated on 11th Gen Intel Core processors, which provide the performance, integrated graphics capabilities, and industrial-grade reliability required for safety-critical applications and the heterogeneous compute foundation needed for AI workloads at the edge.

“Through Intel’s collaboration with Lynx, customers gain coordinated support for AI solutions with MOSA.ic.AI-based deployments on proven Intel platforms. Lynx’s ongoing support for the VkCoreSC graphics stack in Lynx CoreSuite also addresses a critical requirement for long-lifecycle programs—ensuring platform stability and ecosystem continuity from initial development and certification through decades of deployment, while helping accelerate pathways for certifying deterministic AI-enabled systems,” said Prakash Kartha, General Manager, Federal and Aerospace Division, Edge Computing Group, Intel Corporation.

Availability

Lynx offers LYNX MOSA.ic.AI now through direct sales and partner channels.