Dean Takahashi co-wrote this story with Sam Smith.
Niantic Spatial and Spexi Geospatial today announced a strategic partnership that brings together Spexi’s next-generation aerial data network with Niantic Spatial’s industry-leading reconstruction technology, turning drone imagery into large-scale 3D intelligence for physical AI on demand.
Under the partnership, both companies’ customers will access Niantic Spatial’s industry-leading 3D modeling pipeline through its Reconstruction API. Customers can commission drone captures and receive high-fidelity, geometrically accurate 3D reconstructions, in the form of 3D Gaussian splats, accessible through a Niantic Spatial Viewer and measurement tool embedded within the Spexi World platform. Geo-referenced with coordinates, they can be easily dropped into a map of a large area.
Spexi has also been chosen as a preferred drone imagery provider for training Niantic Spatial’s real-world foundation models for physical AI, expanding coverage to city-scale data and beyond. These metric scale 3D reconstructions of physical environments stitch together multiple drone scans of real-world spaces to make intelligent models. Grounded in geometry, they are applicable for simulation, location, and training.
“We started working early on with Niantic to try and create flight missions with drones automatically to take pictures at just the right locations and then the right perspectives to output very high quality digital twins,” Lakeland said. “They were working on Gaussian Splats and still are and working on that pipeline, using the images to help train AI models, like real-world foundational models for spatial AI. Essentially is is what they’re building in the background.”
The collection of data is turning out to be powerful input for that purpose. So Spexi works on the ideal flight missions, the exact heights and pattners the drones need to fly, and more so they can create digital twins with the most impact.
“We’re looking at flying over all cities across North America and other countries globally,” Lakeland said.
That’s putting Niantic Spatial in a position to go to market with construction companies and other verticals where precise 3D data that matters.
“For physical AI to work in the real world, it needs a foundation grounded in reality. Combining Spexi’s capture network with our reconstruction technology and real-world models gets us significantly closer to that and delivers real operational value for our customers,” said Inhi Cho Suh, CEO, Niantic Spatial, in a press release. “Until now, high-quality 3D reconstruction has largely operated at the scale of an object or building. This partnership takes it to city scale and more.”
Spexi operates an advanced aerial data network that has attracted more than 10,000 drone pilots, and mapped over six million acres at 2.8 centimeter resolution – ten times sharper than satellite imagery – using autonomous and standardized flight protocols optimized for machine learning. At the same time, Niantic Spatial’s reconstruction pipeline has been calibrated to Spexi’s capture workflows, ensuring every flight yields maximum quality output.
“Together, Spexi and Niantic Spatial deliver a drone-to-3D pipeline that will redefine the next generation of physical AI, unlocking more accurate, up-to-date, and immersive representations of the built environment,” said Bill Lakeland, CEO, Spexi. “Partnering with Niantic Spatial means customers can now go from raw imagery to actionable 3D intelligence in one seamless workflow. That’s a step change in what drone data can achieve for real-world applications.”
Use cases include infrastructure inspection, insurance risk assessment, energy site analysis, asset management, and 3D measurement. Built for enterprises where spatial insight drives real decisions, the partnership delivers industry-leading levels of detail and a scalable architecture designed for city-scale reconstruction.
Lakeland said the company had been doing testing of digital twins of urban areas in the past couple of years. It has worked with its crowdsourced drone operators to create missions for capturing imagery for Gaussian Splats. The drones capture 25-acre parcels of imagery all over the world in an automated way.
Niantic Spatial is now a separate company from Pokemon Go maker Niantic, which is ownedby Scopely. It focused on its geospatial technology to build games like Pokemon Go, but now it is expanding into other markets with detailed maps of the planet’s cities. The data can be used in everything from VR applications to AR apps.
Regarding possible gaming applications, Lakeland said it could be the foundation for a powerful engine to run real-world gaming applications. That could include real-world operations with digital foundation models.
The company is focused on mapping out 300 cities in the U.S., but i has probably doubled its imagery data in te past coupel of years. Spexi is also focusing on capturing more data on areas where it has already collected information, and it is also coming up with real-time data that is regularly updated.
“The update requirement for models is a really important,” he said.
That’s because maps for first responders has to be accurate and it can’t use generative AI to fill in details.
“Synthetic doesn’t work when you think of a first responder trying to get access to a window or something that needs to be in an exact location,” Lakeland said. “That’s an important growth factor for us.”
One example of a site that needs constant daily updates is a construction site, where the drone flights can capture just how much work is being accomplished on a day to day pace with a construction project. In that respect, the job is almost never done. The network of drone operators is well over 10,000 now.
“They’re saying, ‘Hey, can we fly more?’,” Lakeland said. “I’m quite excited about the physical AI space. That is going to be an unlock for a lot of future growth and value in this, in the space that we’re in.”
The company has 37 people and it has raised $21 million to date.