Singapore announced the establishment of a physical AI systems testbed at the Punggol Digital District (PDD) during the ATxSummit 2026, marking a transition toward deploying real-world AI applications. The facility, launching later this year, results from a collaboration between the Infocomm Media Development Authority, JTC Corporation, and the Singapore Institute of Technology. It aims to support multi-operator deployments at scale, enabling companies including Certis, DHL, Grab, and QuikBot to validate robotic services such as delivery, cleaning, and security patrolling in public spaces.

To enable these operations, the Land Transport Authority has implemented a precinct-level exemption framework under the Active Mobility Act. Furthermore, the National Robotics Programme and partners like FieldAI, Thoughtworks, and Unitree are advancing embodied AI (EAI—models controlling physical hardware) capabilities at the new Centre for Intelligent Robotics in PDD. These efforts seek to refine robotic safety, infrastructure integration, and regulatory compliance.

The summit featured several additional AI initiatives. NVIDIA will launch an AI research hub in Singapore, its second such presence in the Asia Pacific, focusing on embodied AI and efficient AI computing to reduce costs and improve energy efficiency. Concurrently, Google will establish a local office for Google DeepMind, while OpenAI signed a Memorandum of Understanding committing over S$300 million to grow local technical teams and research capabilities. Temus also introduced an AI Foundry, aiming to hire 50 Singapore-based professionals to develop enterprise AI accelerators and governance frameworks.

ASTAR's Institute for Infocomm Research is updating its Multimodal Empathetic Reasoning and Learning in One Network (MERaLiON) model, with the new AudioLLM v3 offering paralinguistic intelligence (understanding tone and intent in speech) across Southeast Asian languages. Available via cloud and edge computing, this model targets applications in healthcare and emergency response. Additionally, the Model AI Governance Framework for Agentic AI (systems capable of autonomous decision-making) has been updated with over 10 new case studies from public and private sector contributors to guide real-world deployments.

Singapore announced the establishment of a physical AI systems testbed at the Punggol Digital District (PDD) during the ATxSummit 2026, marking a transition toward deploying real-world AI applications. The facility, launching later this year, results from a collaboration between the Infocomm Media Development Authority, JTC Corporation, and the Singapore Institute of Technology. It aims to support multi-operator deployments at scale, enabling companies including Certis, DHL, Grab, and QuikBot to validate robotic services such as delivery, cleaning, and security patrolling in public spaces.

To enable these operations, the Land Transport Authority has implemented a precinct-level exemption framework under the Active Mobility Act. Furthermore, the National Robotics Programme and partners like FieldAI, Thoughtworks, and Unitree are advancing embodied AI (EAI—models controlling physical hardware) capabilities at the new Centre for Intelligent Robotics in PDD. These efforts seek to refine robotic safety, infrastructure integration, and regulatory compliance.

The summit featured several additional AI initiatives. NVIDIA will launch an AI research hub in Singapore, its second such presence in the Asia Pacific, focusing on embodied AI and efficient AI computing to reduce costs and improve energy efficiency. Concurrently, Google will establish a local office for Google DeepMind, while OpenAI signed a Memorandum of Understanding committing over S$300 million to grow local technical teams and research capabilities. Temus also introduced an AI Foundry, aiming to hire 50 Singapore-based professionals to develop enterprise AI accelerators and governance frameworks.

ASTAR's Institute for Infocomm Research is updating its Multimodal Empathetic Reasoning and Learning in One Network (MERaLiON) model, with the new AudioLLM v3 offering paralinguistic intelligence (understanding tone and intent in speech) across Southeast Asian languages. Available via cloud and edge computing, this model targets applications in healthcare and emergency response. Additionally, the Model AI Governance Framework for Agentic AI (systems capable of autonomous decision-making) has been updated with over 10 new case studies from public and private sector contributors to guide real-world deployments.