Niantic Spatial

Niantic Spatial builds foundational models designed to let AI systems operate in physical space. The company's work spans spatial intelligence, computer vision, 3D reconstruction, neural networks, visual positioning, localization, Gaussian splats, mesh generation, LiDAR processing and geometrically accurate mapping. Underpinning this is a proprietary database of more than 30 billion posed images.

Two products sit at the centre of the offering. The first is digital twin creation: geometrically accurate digital twins built from smartphones, 360 cameras, drones, satellites and LiDAR, with output in the form of meshes and Gaussian splats. The second is a Visual Positioning System (VPS), which converts aerial and ground capture devices into precise positioning and orientation sensors, delivering accurate localization in places where GPS fails. The VPS is reported to be powered by 50 million neural networks and 150 trillion parameters.

The company positions its work around what it describes as the 80% of economic activity that takes place beyond screens. Stated industry verticals include robotics, the public sector, energy and industrial applications. Further details on founding, team size, funding and geographic footprint are not disclosed here; more information is available at nianticspatial.com.