Proof over Promise Insights on Real World AI Adoption from 2025 MINDS Organizations 2026

Organizations are pursuing diverse strategies to design optimal AI computing infrastructure Some MINDS organizations are harnessing private infrastructure to fine-tune models and take advantage of capabilities like retrieval-augmented generation (RAG) and agentic systems. Their approach prioritizes sovereignty and compliance while ensuring dedicated performance and control over sensitive workloads. Other MINDS organizations embrace hybrid architectures that combine edge and cloud computing. In these systems, data from machines and sensors is analysed first at the edge for speed and then in the cloud for deeper insights. This process, supported by custom middleware software, bridges the two layers. Such hybrid approaches prove effective in managing costs while keeping solutions scalable. Some MINDS organizations modern AI infrastructure stacks are combining edge computing for latency-sensitive and cost-critical workloads, with cloud and high-performance computing (HPC) for large-scale training and orchestration (see Spotlight 8). Others are also introducing custom silicon to reshape unit economics and gain a competitive advantage through specialized hardware acceleration (see Spotlight 9). Railway intelligence platform BOX 15 Hitachi Rail’s AI-powered digital asset management platform integrates real-time data from trains, signalling systems and infrastructure to optimize maintenance, reduce costs and improve operational efficiency across railway networks. Microgrid optimization BOX 16 AI for commercial building efficiency BOX 17Schneider Electric introduced an intelligence platform to automate complex energy decisions across diverse facilities ranging from campuses to industrial sites, with a cloud-based model predictive control (MPC) optimizer connected to an edge controller, turning each facility’s solar panel, batteries, electric vehicle (EV) chargers and flexible loads into a self-learning microgrid. Ultimately, it cuts emissions and costs without sacrificing sites’ workloads. Siemens has applied an edge-cloud model to commercial building operations. Its AI system optimizes heating, ventilation and air conditioning (HVAC) systems in real time, combining live data, occupancy forecasts and weather insights to optimize comfort and efficiency without new infrastructure. In pilot sites, the system improved comfort compliance by over 25% and cut energy use by more than 6%. SPOTLIGHT 8 Ant Group’s hybrid AI infrastructure approach to transforming patient care The Ant Group is tackling fragmentation within healthcare and improving efficiency across the entire patient journey by deploying a hybrid edge-to-cloud architecture. Real-time voice and digital-human interfaces run at the edge for low- latency, natural patient interactions, while distributed cloud training accelerates model updates at scale. This multimodal system processes text, imaging and sensor data securely. By combining privacy-first design, on-premise compliance and edge-cloud synergy, Ant Group delivers scalable and cost- efficient AI that transforms care delivery from disconnected tools into an intelligent, end-to-end service layer. Impact: Ant Group’s edge-cloud architecture doubles model update speed and reduces medical literature search times by 80%. This strategic blend of edge and HPC drives scalable, secure AI for healthcare, enabling inclusive and integrated patient care with specialist-level reasoning for over 5,000 diseases across 70 departments, and diagnostic accuracy above 90%. Proof over Promise: Insights on Real-World AI Adoption from 2025 MINDS Organizations 21