PassiveLogic Launches Generative AI for Autonomous Buildings

- PassiveLogic's core technology creates a "physics-based digital twin" of a building using a descriptive language called Quantum. This detailed virtual model allows the AI to simulate millions of future control decisions to find the most optimal path for managing systems in real-time. - The on-site "Hive" controller, which includes a 12-core processor and an NVIDIA GPU, runs the autonomous system at the edge, meaning it can fully operate without a cloud connection. Multiple Hive units can be linked to create a distributed supercomputer for larger, more complex buildings. - The company has raised over $125 million from investors including NVentures (NVIDIA's venture arm), Brookfield, and Johnson Controls, indicating significant backing for its approach to generative autonomy in the physical world. - The system is designed for both new construction and retrofits, with a "Quantum Create" visual application suite that allows engineers and integrators to build custom digital twins for various building topologies. - PassiveLogic's AI compiler, Differentiable Swift, is reportedly over 300 times faster than Google's TensorFlow and enables the AI to work backward from an optimized goal state rather than simulating trillions of scenarios forward. - A key feature is "self-commissioning," where the Hive controller guides installers through wiring with on-screen error detection and can run the commissioning process on its own once connected. - Case studies of similar AI-driven building automation in facilities like The Edge in Amsterdam have demonstrated significant gains in energy efficiency and occupant comfort by optimizing lighting, HVAC, and occupancy detection. - The global market for Autonomous Buildings is projected to reach $1.3 trillion by 2030, vastly outpacing the traditional building automation market, which is estimated at $191 billion.