I’m excited to welcome blackshark.ai—the company building the first semantic 3D digital twin of the structures on the surface of the Earth—to the M12 portfolio as lead of their Series A funding round, alongside Point72 Ventures.

Tremendous effort goes into mapping our world and keeping that map information fresh and up-to-date. But map data is typically only available in 2D formats, flattening the depth and physical qualities of our environment. Satellite and aerial imagery—combined with photogrammetry techniques—enables the creation of 3D models for specific locations, but a substantial gap remains to scale this to the entire world. Even the inclusion of DSM (digital surface model) data isn’t enough to automatically recreate detailed 3D structures for any arbitrary location on the surface of the planet.

Blackshark.ai answers this challenge by bringing together its innovations in a massively scaled-out deep learning-powered geospatial compute engine. The team is leveraging procedural content generation knowhow from their experience in gaming to create realistic 3D representations of all of the structures on the globe.

Like many others, I first came to know about blackshark.ai’s capabilities based on their contribution to the new Microsoft Flight Simulator. But it quickly became apparent that the applications of their tech extend beyond gaming; many B2B applications will benefit from the advancement in 3D modeling of geospatial information.

The platform can process petabytes of input geospatial data very efficiently while accurately extracting semantic information, such as: building footprints, building heights, land use, bodies of water, or even infrastructure assets like streets or rail tracks. To give a sense of the sophistication of their ML pipeline, blackshark.ai can accurately extract every single building footprint on the entire earth in just a couple of days using the full power of distributed AI compute in the cloud.

The output from this geospatial compute engine enables the definition of a layer of 3D semantics on the source data to create a new kind of processed base map. When given the visual effects treatment we are so familiar with from the gaming and simulation world, stunning synthetic 3D environments that are able to be queried, searched, and updated regularly can be instantiated for all of Earth.

Blackshark.ai’s founders, including CEO Michael Putz, are seasoned entrepreneurs, and previously led a game development studio from which blackshark.ai spun out in 2020. Headquartered in Graz, Austria the company is tapping into the incredible local talent pool specializing in computer vision and AI.

This latest fundraise will be used to ramp up global go-to-market efforts, sales, and customer success, and accelerate additional hiring in machine learning and computer vision. I’m excited to have Point72 Ventures Managing Partner Sri Chandrasekar as a co-investor and fellow board member in the company, and for Applied Intuition CEO Qasar Younis to join as an investor as well.