AI Has Conquered the Internet. Now It Needs to Learn the Real World.

Generative AI is very good at the 20% of the economy that lives on a screen. The other 80% — factory floors, construction sites, farms, oil refineries, city infrastructure — has barely been touched. Niantic, the company behind Pokémon GO, is quietly building the map that changes that.

John Hanke, CEO of Niantic Spatial, framed the problem with unusual clarity: "AI is in many ways trapped inside the screen, deeply knowledgeable about concepts derived from the mountains of text on the internet, and yet woefully ignorant about the world outside the door of the data centre, much less the factory floor, the farm, the construction site, the oil refinery and the cities in which we live."

That sentence describes the actual frontier of AI development. Not the next chatbot. The physical world.

What Spatial Intelligence Actually Is

Large language models are trained on text. They know everything that has ever been written down and very little about how a robot should navigate a wet factory floor at 3 am or route a supply drone across rugged terrain without GPS precision.

Niantic Spatial is building what it calls a large geospatial model — not designed for human operators, but natively for machines. A living map of the world that robots and AI agents can query for navigation, task planning, path optimization, and environmental awareness. Their stated ambition is ambitious, bordering on audacious: a continuously updated spatial intelligence layer that tells machines not just where things are, but what they mean and how to act on them.

Near-term use cases include robot navigation through urban and industrial environments, computing suburban fire risk, and optimizing city layouts. Future iterations will add semantic understanding — the difference between knowing a door exists and knowing it leads to a hazardous materials storage area.

The Industrial Automation Stack Taking Shape

Niantic isn't building this alone. The company expects its geospatial model to integrate with physical AI systems from Physical Intelligence, Skild AI, and Flexion Robotics, as well as world models from World Labs, General Intuition, and Nvidia. Hardware partners include Boston Dynamics, Agility Robotics, and Apptronik.

NVIDIA's Jensen Huang has called humanoid robots "the next multi-trillion-dollar industry." That claim sounds hyperbolic until you map it against the economic reality: if AI tools have captured measurable productivity gains in the 20% of the economy that runs online, and the physical world represents 80%, the math on industrial automation becomes hard to argue with.

The bottleneck has never been the robots. It's been the intelligence layer that tells them where they are, what's around them, and what to do next. That's the gap Niantic is positioning itself to fill.

Why Marketers and Growth Leaders Should Care About Factory Floors

This might feel distant from marketing strategy. It isn't. The physical automation wave Niantic is enabling will reshape supply chains, logistics, manufacturing costs, and labor markets at a scale that dwarfs what generative AI has done to white-collar work so far.

When the cost of physical labor in manufacturing and logistics drops significantly — and the geospatial and robotics infrastructure being built right now suggests it will — the downstream effects reach pricing, inventory, fulfillment speed, and customer experience in ways that every growth team will need to account for.

Hanke made a point worth sitting with. The goal of physical AI, done right, should be "increasing the standard of living" — not, as he put it, engineering a "confection of pixels and waveforms optimised to capture your attention until the next ad arrives."

That's a pointed critique of where most AI investment has gone so far. It's also a useful reminder that the most consequential AI work of the next decade may happen in places without screens at all.

Winsome Marketing helps growth teams understand where AI is heading — across every sector — and build strategies that account for what's actually coming. Let's talk.