Niantic Inc, a former Google spin-off responsible for the hugely popular Pokémon GO and other virtual reality games, has announced that it will split into two arms — one focused on gaming and the other on AI-powered geospatial data.
Niantic’s gaming arm will be acquired by Scopely, a company involved in mobile games, for US$3.5 billion. A further sweetener of US$350 million of cash from Niantic takes the total value to US$3.85 billion.
Of more interest for the geospatial sector, Niantic is spinning off its geospatial AI business into a new company, Niantic Spatial Inc.
To be led by John Hanke, Niantic Spatial will be funded with US$250 million of capital, including US$200 million from Niantic’s balance sheet and a US$50 million investment from Scopely.
Niantic says that all of its original investors will also continue to be shareholders of Niantic Spatial.
Niantic Spatial is powered by what it calls a third-generation digital map that captures the content of the world at what it claims is “a level of fidelity never before achieved,” and which “enables both humans and machines to understand it in ways never before possible”.
Last year, the company released its Spatial Platform, which integrates spatial computing, extended reality, GIS and AI to create produce experiences that blend the digital and physical worlds with centimetre-level precision.
“Existing maps were built for people to read and navigate but now there is a need for a new kind of map that makes the world intelligible for machines, for everything from smart glasses to humanoid robots, so they can understand and navigate the physical world,” said Hanke.
Niantic says the platform is suited for sectors such as manufacturing, warehousing, logistics, construction, tourism, entertainment and education, by virtue of its precise tracking, AR-guided navigation and real-time 3D spatial visualisation.
The company’s Visual Positioning System (VPS) and computer vision stack powers experiences for millions of users of Pokémon GO and other games.
As well, its Scaniverse capture technology uses Gaussian splatting technology to render 3D models.
The company is also developing a Large Geospatial Model which, it says, will enhance “spatial reasoning in Large Language Models (LLMs) by providing a spatially grounded and semantically rich understanding of real-world locations,” and which is built on a proprietary database of more than 30 billion posed images.
“Today’s LLMs represent the first step towards a future where a variety of expert models collaborate to reason and understand complex problems, and many of those problems will require deep and accurate knowledge of the physical world,” said Hanke.
