Niantic Spatial Wants to Map the 80% of the Economy AI Can’t See

Why It Matters

Mapping the offline majority of economic activity unlocks new data streams for AI, enabling automation and analytics in sectors that have been invisible to text‑based models. This shift could drive efficiency gains and new revenue models across heavy‑industry supply chains.

Key Takeaways

• Niantic Spatial launches Scaniverse for businesses, a front‑door to spatial AI
• Scaniverse captures 3D spaces with smartphones or 360° cameras, no expensive gear
• VPS 2.0 provides near‑centimeter positioning where GPS fails
• Niantic targets the 80% of the economy that remains offline
• NSDK 4.0 will unify Unity, Swift, Android, ROS 2 for spatial apps

Pulse Analysis

The rise of spatial intelligence marks the next evolution beyond large language and vision models. By training large geospatial models on 3D scans, satellite imagery, LiDAR and GPS data, developers can create machine‑readable representations of real‑world environments. This foundation enables AI to reason about physical spaces, opening use cases from autonomous robotics to augmented reality overlays that were previously limited to text‑centric insights.

Niantic Spatial’s Scaniverse platform operationalizes this vision for enterprises. Users can generate high‑fidelity meshes and Gaussian‑splat photorealistic scenes using only a smartphone or a 360° camera, eliminating the need for costly lidar rigs. The captured data feeds into VPS 2.0, which delivers sub‑centimeter localization and six‑degree‑of‑freedom tracking, crucial for navigating warehouses, construction sites, and even underwater facilities where GPS signals drop out. The forthcoming NSDK 4.0 unifies development across Unity, Swift, Android and ROS 2, accelerating integration of spatial maps into existing workflows.

Industry analysts see the 80% of the economy that operates offline as a massive untapped data frontier. As companies like Lampata and OpenUK map the geospatial AI landscape, the challenge will be harmonizing diverse data formats—vectors, rasters, point clouds—into reliable models. Success will empower AI to move from conversational interfaces to operational decision‑making, reshaping logistics, energy management, and manufacturing with real‑time, location‑aware intelligence.