Niantic Launches Large Geospatial Model for 80% of Offline Economy

Niantic Spatial Inc. officially launched its Large Geospatial Model (LGM), a foundational AI designed to bring “spatial intelligence” to the offline economy. Following the company’s strategic $3.5 billion sale of its gaming division (including Pokémon GO) to Scopely in 2025, this launch marks Niantic’s full-scale pivot from a gaming giant into a premier enterprise AI firm.

CEO John Hanke highlighted that while current AI is “trapped behind the screen,” Niantic’s LGM is built to understand and navigate the physical world—an arena that accounts for 80% of the global economy.

The Large Geospatial Model (LGM) Explained

Unlike Large Language Models (LLMs) that reason through text, or World Foundation Models (WFMs) that simulate “plausible” environments, the LGM provides ground truth.

• Metric Accuracy: The model uses billions of georeferenced images to create 3D maps with centimeter-level precision, essential for safe machine operation.

• Semantic Understanding: The LGM doesn’t just see shapes; it understands objects. It can identify a “factory floor,” “loading dock,” or “emergency exit,” allowing AI agents to navigate and solve problems autonomously.

• Complementary Intelligence: Niantic expects the LGM to act as the “eyes” for other AI systems, working alongside physical AI from companies like Nvidia and Physical Intelligence.

Industrial Application: Mapping the Offline Economy

The new “Spatial Intelligence” suite is specifically engineered for Industry 4.0, focusing on sectors that cannot be automated through digital reasoning alone.

1. Intelligent Logistics: Optimizing picking routes and autonomous delivery within complex warehouse environments where GPS signal is often denied.

2. Industrial Maintenance: Allowing remote experts to conduct site surveys and risk assessments using high-fidelity digital twins (Gaussian Splats).

3. Robotics Navigation: Providing the mapping layer for humanoid robots and autonomous systems to move safely through “urban mazes” and factory complexes.