In 2018, Uber data scientists plotted millions of ride coordinates in Toronto and turned off the underlying street map. The sheer density of human movement perfectly drew the city anyway. Negative space outlined Lake Ontario, major parks, and the exact footprints of buildings. The telemetry data had become the map. Getting to that realisation required solving a massive computational bottleneck. Uber was drowning in location data. Traditional cartographic software was built for static maps. Feeding those systems the high-velocity data of a global rideshare network essentially caused browsers to freeze and crash. They needed a completely new architecture. Uber brought in Shan He, a former physical architect who had pivoted to computer science at MIT. She recognised that data scientists needed to manipulate massive datasets without writing custom rendering code. She led the creation of Instead of using standard web rendering, the framework bypassed the browser's main thread and offloaded the complex geometric calculations directly to the user's graphics processing unit. The result was a web application that could smoothly render over a million data points and thousands of trips simultaneously. Anyone could build complex 3D visualisations in seconds. The internal discoveries saved Uber millions. By mapping Estimated Time of Arrival errors across Manhattan, analysts visualised severe supply shortages near the water. The physical boundaries of the rivers were forcing cars into northbound vectors, silently breaking the global dispatch algorithms. They mapped pick-up success rates using highly granular hexagonal grids over 3D building geometries. The visualisations pinpointed the exact alleyways and complex building exits that consistently caused cancellations. Uber immediately used this data to rewrite their pick-up recommendation engine. Because the rendering engine only processed coordinates and time, it was completely indifferent to the subject matter. An engineer used it to model theoretical urban airspace logistics for flying cars. Academics used the exact same tool to track the spatial distribution of tick-borne viruses and map satellite orbits. Uber made the strategic decision to release under an open-source licence. It became the industry standard almost overnight. Airbnb used it to track rental pricing volatility during the pandemic. City planners used it to untangle commute patterns across New York. Then, the core engineering team left Uber to found a startup called Unfolded. They built enterprise-grade data management tools on top of their open-source rendering engine. They raised $6 million, proved the enterprise value of their architecture, and were acquired by Foursquare in 2021. A tool originally built to stop browsers from crashing when visualising taxi rides became a pretty important tool in geospatial visualisations.