Strategic collaboration showcases real-time Vision-Language-Action (VLA) model workflows powered by AMD Ryzen™ AI Embedded platforms and AMD ROCm™ open software at AMD Advancing AI 2026.
MulticoreWare, Inc., an industry pioneer in artificial intelligence software solutions, accelerated computing, and physical AI engineering, announced its continued strategic technical collaboration with AMD to develop edge intelligence and future generation of autonomous robotics.
With the evolution of artificial intelligence workloads directly into physical robots, autonomous vehicles, and edge devices from centralized cloud platforms, industrial innovators demand compute architectures for executing advanced AI models with ultra-low latency. In light of this new trend, MulticoreWare teamed up with AMD for their Advancing AI 2026 conference in their joint presentation “Enabling Physical AI on AMD.” The interactive demonstration showcased how advanced Vision-Language-Action (VLA) models enable real-time perception and decision making using AMD Ryzen™ AI Embedded platform.
During the demonstration, MulticoreWare and AMD showcased multimodal VLA models executing directly on integrated GPUs within the AMD Ryzen™ AI Embedded platform utilizing the open-source AMD ROCm™ software stack. The session offered practical engineering blueprints for AI developers, system architects, and original equipment manufacturers (OEMs) seeking to build highly responsive, production-ready intelligent machines.
Also Read: SAP and Cyberwave Bridge the Gap Between Digital and Physical AI with Autonomous Robotics Deployment
“Physical AI is reshaping how machines perceive, decide and act in the real world,” said Sumit Shah, Head of Product Management and Marketing, Adaptive and Embedded Computing Group, AMD. “AMD Ryzen™ AI Embedded X100 Series processors deliver a scalable, open x86 Embedded platform that unifies AI, real-time control and industrial reliability to enable the generation of autonomous systems without locking developers into a single compute architecture or software stack.”
“A Physical AI system depends on a tightly integrated loop between perception and actuation. It must operate in real time, on real hardware, and in environments that are inherently unpredictable,” said Vish Rajalingam, VP & GM, Mobility and Transportation BU at MulticoreWare. “That makes it a hardware-software co-design challenge, not simply an AI inference problem. Building on the open-source AMD Robotics Software Suite, we work closely with OEMs to optimize the entire stack so that latency, reliability and accuracy targets are consistently achieved in production environments. That’s the integration MulticoreWare and AMD deliver together to move intelligent robotic systems from prototype to deployment.”
This milestone builds upon more than 15 years of continuous engineering collaboration between the two companies. Over the past decade and a half, MulticoreWare has provided algorithm optimization, AI runtime tuning, and software engineering services across the broader AMD hardware portfolio including AMD Ryzen™ AI, AMD EPYC™, AMD Instinct™, AMD Radeon™, and AMD adaptive computing solutions.
