A digital twin, in the original and technically precise sense, is a model that predicts the future state of its physical counterpart. Not describes. Not records. Predicts.

A BIM model describes geometry and specification. A facilities database records maintenance history. A sensor-connected model shows you current conditions. All useful. None of them twins.

A real twin takes current sensor data, feeds it into a calibrated structural or thermodynamic model, and tells you what will happen next — where fatigue is accumulating, when a component is likely to fail, how the building will respond to a load event that hasn’t happened yet. It is a predictive instrument, not a descriptive one.

The gap between what AEC calls a digital twin and what the aerospace and manufacturing industries have been operating for twenty years is enormous. A jet engine twin updates its predictive model in real time from hundreds of sensors and can forecast maintenance requirements months in advance. The average “smart building” can tell you the current temperature in room 214.

Closing that gap requires three things AEC currently lacks at scale: dense sensor instrumentation of structures post-occupancy, calibrated physics-based models that degrade gracefully as conditions change, and the data infrastructure to connect them continuously over a building’s lifetime.

The technology exists. The business model for who pays for it across a fifty-year asset life does not.