Abstract
From 2011 to 2026, organizations outside the traditional geospatial industry became major developers of geographic data and technology. Ride-hailing firms built spatial indexes to price and dispatch trips. Social networks and cloud providers generated building and road data with machine learning and released it openly. Automakers bought a map company to keep it out of a competitor’s hands. Database vendors made geometry a native column type. This paper examines development across four overlapping phases: map data, purpose-built tooling, general-purpose data infrastructure, and consortium-scale open data. It identifies four mechanisms: user innovation, vertical integration, platform convergence, and collective infrastructure investment. It also distinguishes between innovations originating outside the geospatial sector and established geospatial technologies that outside organizations brought to larger markets. The case evidence supports both patterns. Among incumbents, the clearest documented response has been a shift toward integration and analytical software. The evidence is less conclusive on whether open data has reduced proprietary base-data prices. Testing that claim would require historical pricing, switching, or margin data that are not available in this review.
Introduction
In 2011, geographic information was still produced and distributed through a fairly well-defined set of institutions. National mapping agencies and survey firms captured the physical world. A small number of commercial map-data suppliers, chiefly Navteq (later Nokia HERE) and Tele Atlas (later part of TomTom), licensed road networks to navigation and device makers. Desktop and enterprise GIS vendors supplied the analytic software, and a handful of satellite operators sold imagery, largely to governments. Users outside that ecosystem consumed its outputs; they rarely produced them.
Over the next fifteen years, that division of labor changed. Firms in mobility, social media, consumer electronics, cloud computing, agriculture, and gaming found that their products depended on spatial data and spatial computation that the traditional sector either could not supply at the required scale, resolution, or price, or would supply only on terms that constrained them. Many developed their own capabilities, and some subsequently made those capabilities available to others.
Two earlier strands of scholarship anticipated parts of this shift. Goodchild (2007) described volunteered geographic information as a new class of contribution in which ordinary people act as sensors and authors, and argued that it would reshape the relationship between GIS and the public. Haklay (2013) cautioned that the reduced technical barriers associated with this “neogeography” delivered a narrower form of democratization than its proponents claimed, with participation shaped by technical skill and institutional power. Developments since 2011 reflect elements of both arguments. Participation in mapmaking expanded, although some of the most influential new contributors were large corporations rather than individual volunteers.
Four mechanisms of diffusion
Von Hippel’s (2005) account of user innovation is the paper’s starting point: firms and individuals develop products for their own use, often ahead of manufacturers, and frequently share the results because the private benefit of use exceeds the benefit of withholding. Uber’s H3 fits that definition closely. Acquisitions, new platform features, and shared data initiatives reflect different incentives and have different implications for established geospatial firms. The review distinguishes four mechanisms.
User innovation in the strict sense: a firm builds a geospatial capability to solve its own operational problem and later releases it. Uber’s H3 is the clearest example.
Vertical integration and collective control of a supplier: a firm or group of firms takes ownership of map production that it previously bought. Apple’s rebuilt base map and the automakers’ purchase of HERE belong here. These transactions changed control of an existing capability.
Platform convergence: a general-purpose computing platform adds geospatial functions so that its existing customers can do spatial work without leaving it. Spatial types in cloud data warehouses and embedded databases are the main examples. These additions primarily serve platform customers rather than the vendor’s own spatial operations.
Collective infrastructure investment: competitors pool the cost of producing a shared input and compete above it. Corporate editing of OpenStreetMap (OSM) and the Overture Maps Foundation are the main examples.
The economics of base maps help explain the incentive for collective investment. Base map data carries high fixed costs and low marginal costs, and every firm that maps the same streets duplicates work its competitors are also doing. Sharing a common base layer can therefore reduce duplicated expense while leaving companies free to compete through other products and services. By the early 2020s, several large firms had organized around that approach.
Scope and method
This paper uses “traditional geospatial sector” to mean organizations whose primary product is geographic information, geographic software, or geospatial capture: GIS software vendors, commercial map-data suppliers, Earth observation (EO) operators, and survey and photogrammetry firms. “Non-traditional” organizations are those whose primary product lies elsewhere but who developed, acquired, or packaged geospatial data, software, or methods. The distinction is not absolute. Google, for instance, has operated mapping products for two decades despite its much broader business.
Cases were selected on three criteria. Each involved an organization outside the traditional sector, or a unit of one whose parent’s primary business lay outside it. Each either produced a geospatial dataset, tool, or standard released publicly or adopted beyond the originating firm, or acquired or internalized control of a strategic geospatial capability previously supplied by the traditional sector. The second route admits cases such as Apple’s rebuilt base map, the HERE acquisition, and the purchase of The Climate Corporation, which changed who owned geospatial capability without releasing anything. Each case falls at least partly within 2011–2026 and has documentation in a regulatory filing, standards document, peer-reviewed paper, or the organization’s own announcement. The selection favors cases with public footprints, which biases it toward large Western technology firms and toward releases rather than internal capabilities. It excludes Chinese platforms such as Baidu and Amap, defense and intelligence programs, and autonomous-vehicle HD mapping, each of which would warrant separate treatment.
The four phases were derived from the layer of the geospatial stack, each case primarily affected (base data, application tooling, computational infrastructure, and shared data governance) and ordered by when activity in that layer became concentrated. The phases organize the discussion rather than establish strict historical boundaries. Several cases span more than one phase. Table 1 summarizes the cases.
The review draws on peer-reviewed literature where it exists, Open Geospatial Consortium (OGC) standards documents, regulatory filings, and primary corporate announcements, supplemented by trade and technology press. Corporate announcements are treated as evidence of what a firm did and said, not of the effects it claimed. The analysis also distinguishes four stages that the evidence supports to different degrees: origination (who built a capability), availability (whether it was released), adoption (whether others used it), and economic displacement (whether it changed incumbents’ revenues or prices). Most sources establish origination and availability. Evidence for adoption is available in some cases, while economic displacement remains difficult to demonstrate. Appendix A records the qualitative source assessment, and Appendix B records the verification status of each case.
Table 1. Cases by business problem, capability, mechanism, and diffusion path
| Case | Originating business problem | Capability | Mechanism | Diffusion path | Implication for incumbents |
|---|---|---|---|---|---|
| Apple Maps rebuild (2018) | Dependence on third-party map data for a core device app | Proprietary base map from vans and iPhone probe data | Vertical integration | Internal; not released | Lost or reduced supplier role |
| HERE acquisition (2015) | Automakers’ dependence on maps for automated driving | Control of a map supplier | Collective control of a supplier | Ownership change | Supplier absorbed by its customers |
| Corporate OSM editing (2016–) | Need for editable base data without licensing limits | Paid editing at scale | Collective infrastructure | Open database | Substitute base data |
| Microsoft / Google building footprints (2018–2023) | Building coverage for maps and cloud workloads | ML extraction from imagery | User innovation, then release | Open data | Substitute foundational data |
| Meta Map With AI / Rapid (2019) | Road coverage for its own map products | ML road prediction with human review | User innovation, then release | Open tool and data | Lower cost of OSM production |
| Uber H3 and kepler.gl (2018) | Real-time pricing, dispatch, and analysis | Hexagonal index; web visualization | User innovation | Open source | Alternative aggregation and visualization tools |
| Climate Corporation (2013) | Agricultural risk and agronomy | Field-level weather and agronomic modeling | Capability acquisition by an outside vertical | Absorbed into a seed company | Agriculture as a buyer of geospatial analytics |
| Flight Simulator / Blackshark.ai (2020) | Global scenery for a game | Inferred global 3D reconstruction | User innovation (commissioned) | Commercial product | New entrants in 3D reconstruction |
| Niantic Spatial (2025) | Location-based games | Large geospatial model; visual positioning | User innovation, spun out | New geospatial company | New competitor from outside the sector |
| Earth Engine (2010–) | Environmental monitoring at planetary scale | Cloud EO analysis platform | Platform convergence | Hosted platform | EO analysis without traditional remote sensing software |
| BigQuery, DuckDB, Databricks spatial (2018–2025) | Spatial work inside general data platforms | Native spatial types and SQL | Platform convergence | Commercial and open source | Spatial analysis without GIS licenses |
| GeoParquet, STAC (2018–2025) | Interoperable cloud-native data | Formats and catalog specifications | Mixed community standards | OGC adoption | Standards originating outside the OGC |
| Overture Maps Foundation (2022–) | Duplicated, non-interoperable base data | Shared open base map and stable IDs | Collective infrastructure | Open data | Base layer shared among competitors |
Phase one: The map-data layer changes hands (2012–2019)
Some of the earliest changes occurred in base-map production and ownership. When Apple launched its own Maps application in 2012, it relied on third-party data, and the launch went badly enough that Apple’s chief executive publicly apologized (Slivka, 2012; TomTom, 2012). Six years later Apple disclosed that it was rebuilding the map with data from its own fleet of sensor-equipped vans and anonymized trip segments contributed by iPhones, with the stated goal of owning every input to the map for quality and privacy reasons (Panzarino, 2018). Apple told Reuters that TomTom would remain a data provider; TomTom’s shares nonetheless fell as much as 5% from their session high on the news before closing down 1.7% (“Apple Rebuilds Mapping App,” 2018). Apple demonstrated that a device maker with a sufficiently large installed base could become a producer of map data, using its devices as part of the capture network.
Automakers took a different route to controlling their map supply. When Nokia put HERE up for sale in 2015, a consortium of Audi, BMW, and Daimler bought it, completing the purchase in December 2015 for about €2.55 billion, roughly US$2.8 billion (Lunden, 2015). The buyers said their aim was to keep HERE available as an open and independent platform for the automotive industry and other sectors (Ribeiro, 2015); a contemporaneous deal analysis characterized the motive as concern that a technology company might acquire HERE and license it back to automakers at prohibitive cost (“Audi, BMW and Daimler,” 2015). The transaction illustrates collective vertical integration rather than user innovation. Three direct competitors chose joint ownership of an existing supplier over dependence on an outside platform; the map itself was unchanged by the transaction.
For developers, the issue was often the cost and terms of access. In May 2018 Google announced that it was consolidating its mapping APIs as Google Maps Platform under a single pay-as-you-go plan with US$200 of free monthly usage, and that from June 11 all projects would need a billing account (Google Maps Platform Team, 2018). When the new prices took effect in July, the free tier for dynamic maps fell from 25,000 map loads per day to about 28,000 per month, and the overage price rose from US$0.50 to US$7.00 per thousand requests (Ewbank, 2018). Contemporary coverage anticipated that the change would push developers toward OSM and other alternatives (Ewbank, 2018). At least one large consumer application had already moved: according to contemporaneous reports from OSM contributors, Pokémon GO switched its in-game map display to OSM data in December 2017, a change those contributors noticed through an influx of new editors (“Pokemon GO Now Uses OSM Globally,” 2017). Niantic’s own account of the decision has not been located. Meta launched its own basemap for Facebook, Instagram, and WhatsApp in late 2020, built on the Daylight distribution of OSM (Adkins, 2023).
OSM became an important outlet for corporate contributions. Anderson et al. (2019a) documented that Apple, Microsoft, and Facebook, among others, had begun employing paid editors to contribute to the OSM database at scale. Their analysis of quarterly snapshots found that corporate editing had a global footprint but varied considerably by firm in geographic reach and edit type; corporate teams had a major influence on road networks, while volunteer mappers continued to contribute most building and point-of-interest edits and the majority of edits overall. The authors placed this within a longer history of corporations both contributing to and benefiting from OSM, and Anderson’s accompanying conference presentation reported more than 17 million edits by paid contributors over the preceding five years (Anderson et al., 2019b).
Corporate participation also extended from manual editing to machine-generated data. Microsoft and Google developed extraction systems for their mapping needs and made the resulting datasets public, closer to von Hippel’s account of user innovation. In June 2018 Microsoft released 124,885,597 building footprints covering the United States, extracted from Bing aerial imagery with a deep-learning segmentation model followed by a polygonization step, and published them under an open license with the stated aim of increasing building coverage in OSM (Dempsey, 2018). The dataset was later re-extracted from newer imagery for much of the country, and Microsoft’s announcement now reports 129,591,852 footprints (Bing Maps Team, 2018). Microsoft counted about 30.6 million US building footprints in OSM at the time (Bing Maps Team, 2018). The release therefore made available, at no cost, a dataset roughly four times the size of OSM’s US building inventory. How much of it was subsequently imported into OSM is a separate question that this review does not measure.
Facebook followed in 2019 with Map With AI and the RapiD editor, a modified version of OSM’s iD editor that presents machine-predicted roads for human mappers to review before they enter the map (Denham, 2019). Facebook reported using the system to add more than 300,000 miles of previously unmapped roads in Thailand to OSM in about 18 months (Gao et al., 2019), and by March 2020 Microsoft’s open building datasets were available inside the same editor (Bing Maps Team, 2020). Google extended the pattern to regions with the thinnest coverage. Its Open Buildings dataset, built from 50 cm satellite imagery by a team based in Ghana, began with 516 million building detections across 64% of the African continent (Sirko et al., 2021) and by its third version contained 1.8 billion buildings across Africa, Latin America, the Caribbean, and South and Southeast Asia, released under a CC BY 4.0 license (Google Research, 2023).
By the end of the 2010s, firms serving consumer applications, advertising, and cloud workloads were substantial producers of openly available road and building data.
Phase two: Building the tools they needed (2013–2020)
Other companies developed spatial software because existing tools did not fit their operational workloads.
Uber provides a particularly clear example of user innovation. A ride-hailing business has to aggregate millions of events into spatial units fast enough to set prices and dispatch drivers in real time, and administrative polygons and square grids handle that poorly. Uber developed H3, a hierarchical hexagonal grid system that maps any latitude–longitude pair to a 64-bit cell index at a chosen resolution, explicitly for pricing and dispatch, and open-sourced it in 2018 (Brodsky, 2018). The same year Uber released kepler.gl, a browser-based geospatial visualization application built on its own deck.gl WebGL framework and intended for analysts and data scientists of varying skill levels (He, 2018). These sources establish origin and release, not the extent of downstream adoption, which this review does not measure. One documented diffusion path is personal as well as technical: H3’s original author later published an H3 extension for DuckDB while working at Foursquare (Litwintschik, 2023).
The Climate Corporation illustrates acquisition of an existing capability. It was founded in 2006 by former Google engineers (“Monsanto Acquires Weather Data Company,” 2013) and was already a geospatial analytics business offering hyper-local weather monitoring, weather simulation, and agronomic modeling as risk-management and decision-support tools for growers when Monsanto agreed to acquire it in October 2013 for approximately US$930 million, completing the purchase on November 1 (Monsanto Company, 2013a, 2013b). Monsanto did not originate the technology. The acquisition is significant because a seed and chemical company valued field-level spatial analysis sufficiently to spend nearly a billion dollars acquiring the business.
Gaming created demand for global 3D reconstruction at a scale uncommon in conventional mapping projects. Microsoft Flight Simulator, released in 2020, set out to let players fly anywhere on Earth. Bing Maps’ photogrammetric 3D data covered only about 400 cities at full fidelity, so Microsoft engaged Blackshark.ai, a team of about fifty people in Graz, which used machine learning and large-scale cloud computing to reconstruct some 1.5 billion buildings from two-dimensional satellite imagery across the rest of the planet (Lardinois, 2020). The result was a global 3D model generated largely by inference, commissioned for an entertainment product.
Niantic offers a second, distinct example from gaming. Its location-based games, Pokémon GO above all, accumulated a large corpus of player-captured imagery tied to real places. In March 2025 Niantic agreed to sell its games business to Scopely for US$3.5 billion and to spin its geospatial technology into a separate company, Niantic Spatial, which launched in May 2025 with US$250 million in funding (“Niantic Spatial Spin-Off,” 2025; Niantic Spatial, 2025a). The new company’s stated long-term goal is a large geospatial model built on more than 30 billion posed images, and it describes its visual positioning system as offering centimeter-level precision (Niantic Spatial, 2025a, 2025b). Both are company claims; no independent accuracy benchmark is cited here. In this case, a consumer gaming business generated a proprietary street-level imagery collection and later established a separate business around the resulting geospatial technology.
Phase three: Geography becomes a data type (2010–2025)
General-purpose data infrastructure also began incorporating geospatial capabilities. Over roughly fifteen years, spatial computation moved from specialist software into the tools that data engineers and analysts already used. The mechanism here is platform convergence: the platform vendors were mostly not solving their own operational problems but completing their products for customers who increasingly needed spatial functions.
Google Earth Engine predates the other cases. Google.org introduced it at the UN climate conference in Cancún in December 2010 as a philanthropic platform for environmental monitoring, initially aimed at forest monitoring by scientists and developing-country governments (Kolodny, 2010). Gorelick et al. (2017) later described the mature system, which placed petabyte-scale satellite archives next to Google’s computing infrastructure and exposed them through a programming interface. Its designers stated the intent plainly: to serve not only remote sensing scientists but a much wider audience lacking the technical capacity to run analyses on supercomputers or large commodity clusters (Gorelick et al., 2017). Earth Engine removed the download-and-preprocess step that had dominated EO workflows and made planetary-scale analysis a matter of writing a script.
The cloud data warehouse followed. In August 2018 Google introduced BigQuery GIS in public alpha, with functions following the SQL/MM Spatial standard, and described BigQuery as the first cloud massively parallel processing warehouse to support geospatial data types and functions, according to Google (Jennings, 2018). In 2023 DuckDB, an embedded analytical database, released an official spatial extension that added a geometry type and spatial functions built on the long-established open-source libraries GEOS, GDAL, and PROJ (Gabrielsson, 2023). Databricks introduced Spatial SQL in early 2025, with native GEOMETRY and GEOGRAPHY types, and by December 2025 the feature was generally available with about 90 spatial functions (Marten & Johns, 2025). These releases brought spatial operations into cloud warehouses, a local analytical database, and a lakehouse platform. In each case, a spatial join became an ordinary SQL operation that a data engineer could write without acquiring a GIS license or learning a GIS desktop application.
Interoperable formats and catalogs developed alongside the databases. GeoParquet, a specification for storing geometries and coordinate reference metadata in the Apache Parquet columnar format, reached version 1.0.0 in 2023 after development by a community that included cloud, analytics, and geospatial firms (Cloud-Native Geospatial Forum, 2023). The OGC formed a GeoParquet Standards Working Group the same year to carry the specification through formal standardization (Cloud-Native Geospatial Forum, 2023). The SpatioTemporal Asset Catalog (STAC), a specification for describing satellite imagery and other spatiotemporal files, followed a similar course: developed outside the OGC by an open community, it was approved as an OGC Community Standard on September 9, 2025, and published on October 14, 2025 (Open Geospatial Consortium, 2025). The OGC’s own justification for adopting STAC described its benefit as “describing geospatial data to newer, non-geospatial companies” (Open Geospatial Consortium, 2023).
Geospatial foundation models provide a more recent example of integration with general-purpose computing. In August 2023 NASA and IBM released Prithvi, a geospatial foundation model trained on NASA’s Harmonized Landsat and Sentinel-2 imagery using IBM’s AI infrastructure, and distributed it openly through Hugging Face, the general-purpose machine learning model hub (IBM, 2023). In July 2025 Google DeepMind introduced AlphaEarth Foundations, a model that compresses multi-source Earth observations into annual embeddings, and released those embeddings in Earth Engine as a Satellite Embedding dataset with more than 1.4 trillion embedding footprints per year, after testing with more than 50 organizations (Google DeepMind, 2025). Those distribution channels, a machine learning model hub and a cloud analysis platform, were themselves products of the technology sector at large, not of the geospatial industry.
Phase four: Consortium-scale open data (2022–2026)
By the early 2020s, large corporate map producers had overlapping investments in base data that were difficult to combine. Each was building and maintaining base map data, each was also contributing to OSM, and the resulting datasets did not interoperate cleanly. In December 2022 Amazon Web Services, Meta, Microsoft, and TomTom founded the Overture Maps Foundation under the Linux Foundation to develop interoperable open map data as a shared asset (Linux Foundation, 2022). The founding announcement described a model in which members combine resources to build a base map under an open license while remaining free to contribute and commercialize their own data on top of it (Linux Foundation, 2022).
Overture’s milestones arrived in stages. Monthly beta releases began in July 2023. In July 2024 four themes (buildings, places, divisions, and base) reached general availability, and the transportation theme followed in December 2024 with 86 million kilometers of roads (Prioleau, 2025). The Global Entity Reference System (GERS), which assigns stable identifiers to entities such as buildings, places, addresses, and road segments so that other datasets can be attached to them, was declared generally available separately in June 2025 (Overture Maps Foundation, 2025). Releases were distributed directly into AWS, Azure, Snowflake, BigQuery, and Databricks (Prioleau, 2025), so the data arrives where the non-geospatial analyst already works.
Open releases also came from individual firms outside the consortium. In November 2024 Foursquare, whose point-of-interest data had long been a commercial product, released more than 100 million global places with 22 core attributes under the Apache 2.0 license, updated monthly (Little, 2024). Foursquare’s explanation echoed the economic rationale for sharing foundational data: proprietary place data had slowed adoption, and a company without a dominant distribution platform could not build a comprehensive place database alone (Little, 2024).
The economics outlined earlier help explain the formation of Overture and the broader move toward open base data. Base map data had become an input that every large platform needed and none could easily differentiate on. The participating firms could reduce duplication in base-data production while competing in other parts of their businesses. Most of the founding organizations were not primarily geospatial companies.
Origin or scale? The strongest counterargument
There is an important qualification to this account. Many of the cases rest on work done inside the traditional geospatial community. DuckDB’s spatial extension, like much of the open geospatial stack, is built on GEOS, GDAL, and PROJ (Gabrielsson, 2023), long-established open-source libraries developed within the geospatial community. kepler.gl renders through react-map-gl, a wrapper for Mapbox’s map library (He, 2018). Earth Engine’s catalog is built around public archives, chief among them the Landsat record produced by NASA and the U.S. Geological Survey (Gorelick et al., 2017). OSM itself predates the period by seven years. GeoParquet’s development drew in geospatial firms alongside cloud and analytics companies, and Planet’s catalog was among the production implementations cited in the OGC’s case for STAC (Cloud-Native Geospatial Forum, 2023; Open Geospatial Consortium, 2023). Much of the apparent external innovation may therefore reflect the ability of larger technology companies to package, scale, and distribute capabilities developed by geospatial specialists.
The cases support that interpretation in many instances. They also suggest separating technical origination from subsequent scaling and distribution.
Origination from outside the sector is clearest where a firm’s operational problem had no existing solution at the required scale. H3 was a new spatial index designed around dispatch. Continent-scale building extraction from imagery, as Microsoft and Google practiced it, depended on deep-learning pipelines and imagery archives that traditional data vendors did not operate at that scale. Niantic’s large geospatial model is built on a corpus that only a consumer game could have collected.
Scaling and packaging describes the platform and consortium cases. Spatial SQL in warehouses largely reimplemented or embedded established geometry libraries and the OGC’s Simple Features model. Overture assembles OSM, corporate contributions, and open government data rather than inventing a new kind of map. In these cases the decisive contribution was distribution: placing existing geospatial capability inside platforms with very large user bases and removing licensing and installation barriers.
This distinction changes how the broader shift should be understood. Geospatial specialists continued to develop important libraries and standards that other organizations incorporated into widely used products. Deployment, distribution, and governance increasingly involved firms outside the traditional industry, with consequences for where commercial value accrued. The external contribution included new techniques, but much of its importance lay in bringing established capabilities to new markets and users.
Effects on the traditional geospatial sector
Pressure on base-data licensing: what the evidence shows
Commercial base-map suppliers faced new alternatives, although the financial effects are difficult to isolate. Diffusion is well documented: large platforms substituted open or self-built data for licensed data (Panzarino, 2018; “Pokemon GO Now Uses OSM Globally,” 2017; Adkins, 2023), and Overture now supplies a shared base layer at no cost. Whether that substitution compressed the prices and margins of proprietary base data is a separate claim, and the available evidence is mixed.
TomTom provides a useful test case. It co-founded Overture and rebuilt its product around it: its Orbis Maps platform, launched in 2023 and rolled out globally in 2024, combines Overture’s base layer and OSM with TomTom’s proprietary data and conforms to the Overture specification (TomTom, 2024; Collins, 2024). Its 2024 results, however, do not show a collapse in map licensing. Group revenue fell 2% to €574 million, but Location Technology revenue was nearly flat at €489 million against €491 million the year before; the decline was concentrated in automotive, down 5%, which TomTom attributed to challenging market conditions and lower near-term volumes, while its enterprise business grew (TomTom, 2025). TomTom later reported a 2024 net loss of €17 million, guided to lower 2025 revenue, and announced about 300 job cuts in a restructuring around AI-driven product development (Roos, 2025). These figures are consistent with an incumbent under pressure, but the pressure TomTom itself identifies is cyclical weakness in automotive, not substitution by open data.
Google’s pricing provides another, more limited observation. In March 2025 Google replaced the US$200 monthly credit introduced in 2018 with a free usage threshold for each core service and extended automatic volume discounts from 100,000 to more than 5 million monthly billable events; in Google’s own worked example, a customer without a negotiated contract making 2 million geocoding requests a month would see its bill fall from US$7,900 to US$5,050 (Google for Developers, 2025). The change was not an across-the-board cut: Google’s documentation also shows a lower-volume customer whose bill rises because a per-service allowance replaced the general credit (Google for Developers, 2025). Lower prices for high-volume customers from the dominant proprietary provider are what competitive pressure from open data would predict. Google gave no reason for the change, and other explanations, including competition from Apple, HERE, and Mapbox or cloud-platform bundling, are equally available.
The evidence supports a narrower conclusion. Open and self-built base data removed some of the largest potential customers from the licensing market and created a no-cost alternative for many others. Incumbents have visibly moved their value proposition toward integration, quality control, and application services. A demonstrated compression of base-data licensing prices, attributable to open data rather than to automotive cycles or proprietary competition, would require evidence this review does not have: licensing price series, contract-level switching data, or incumbent segment margins over the period.
Incumbents repositioning toward integration and analytics
A similar repositioning is apparent in Earth observation. Advent International acquired Maxar Technologies for about US$6.4 billion in a deal announced in December 2022 and completed in May 2023 (Advent International, 2023), divided the company into intelligence and space-systems units in 2023, and in October 2025 renamed them Vantor and Lanteris (Pettibone, 2025). Vantor presented itself as a spatial intelligence company building Tensorglobe, a platform under development for fusing satellite, aerial, and ground sensor data into a continuously updated 3D model, rather than as an imagery vendor (Pettibone, 2025). Like TomTom, the company is positioning integration and analytical services alongside its underlying data products.
Launch economics also changed the conditions under which Earth observation firms operated. Jones (2018) estimated that the Space Shuttle’s launch cost per kilogram to low Earth orbit was roughly twenty times that of SpaceX’s Falcon 9, and the CSIS Aerospace Security Project found that, in inflation-adjusted terms, per-kilogram launch costs fell gradually from 1957 to 2005 and much more steeply between 2005 and 2020 (Roberts, 2022). Cheaper launch lowered the cost of putting imaging constellations into orbit for incumbents and new entrants alike, and the price was set by a launch company, not by any EO operator.
Accommodation and participation
Established GIS software firms also participated in the new arrangements. Esri, which Overture describes as the market leader in location intelligence, became Overture’s first General Member (Prioleau, 2025). In the specific case of cloud-native formats and catalogs, the OGC adopted GeoParquet and STAC, specifications that had been built and deployed in production before formal standardization began (Cloud-Native Geospatial Forum, 2023; Open Geospatial Consortium, 2025). Whether that pattern extends across the OGC’s broader portfolio is not tested here; the OGC API family, for instance, was developed within the consortium, and STAC was explicitly aligned with it.
A hypothetical illustrates the operational consequence. An insurer’s analytics team that, in 2012, needed building footprints joined to a hazard layer would typically have licensed footprint data, licensed GIS software, and staffed GIS analysts to run the overlay. The same team in 2026 could query open building footprints from Overture or Microsoft inside its existing data warehouse, index them to H3 cells, and perform the join in SQL. The GIS vendor and the data vendor are no longer necessary participants in that workflow, though they may still compete to be in it on the strength of data quality, support, and domain-specific analytics.
Limits of the diffusion
Wider access to geospatial data and software does not necessarily distribute control over their production. Haklay’s (2013) observation that lower technical barriers produce only a partial democratization applies with more force when the new producers are among the largest firms in the world. The datasets now available at no cost were generated by organizations with access to proprietary imagery, large labeling teams, and global compute. They were released because releasing them served those organizations’ interests, and their continued maintenance depends on those interests persisting.
Corporate OSM contributions illustrate the concentration of influence. In the areas where corporate teams worked, Anderson et al. (2019a) found that corporate editing had become a major influence on the road network, even as volunteers continued to account for most edits overall. A volunteer project’s road layer in parts of the world now reflects, to a meaningful degree, the editorial priorities of a few firms.
Data quality and accountability present additional limitations. Machine-generated footprints and roads arrive with confidence scores, not survey-grade guarantees; Google’s Open Buildings dataset carries a per-building confidence value precisely so users can filter out uncertain detections (Google Research, 2023). Inference-based global models are designed for coverage first. The traditional sector’s emphasis on lineage, accuracy standards, and liability did not become obsolete in this period. It became something that has to be added back, often at the integration layer where incumbents are now concentrating.
Open base data continues to coexist with proprietary capture networks. Apple’s van fleet and iPhone probe data, Niantic’s 30 billion posed images, and automotive sensor data remain closed, and the commercial maps of Google, Apple, HERE, and TomTom continue to serve navigation and automotive customers at scale. The diffusion described here opened one layer of the stack while new proprietary assets accumulated in others, particularly in street-level imagery and the 3D reconstructions now feeding spatial AI.
Conclusion
Over the period examined, firms that once primarily consumed geographic information also became producers and owners of significant geospatial capabilities. Some built capabilities their operations required and released them. Some took ownership of the suppliers they had depended on. Platform vendors absorbed geometry into general-purpose databases because their customers needed it there. Competitors pooled base map production because duplicating it no longer made economic sense. Many drew on existing geospatial methods and infrastructure, although some developed new techniques to meet their own needs.
Established geospatial firms joined consortia, adopted new formats, and placed greater emphasis on integration, quality assurance, and analytics. A large part of the base layer is now shared infrastructure, alongside proprietary maps that remain in commercial use. Competition increasingly includes fused sensor models, street-level capture, and spatial AI systems. Several firms active in those markets would not have described themselves as geospatial companies in 2011.
References
Adkins, J. (2023, February 7). Basemap improvements for Facebook, Instagram, and WhatsApp. Engineering at Meta. https://engineering.fb.com/2023/02/07/web/basemap-facebook-instagram-whatsapp-improvements/
Advent International. (2023, May 3). U.S. private equity firm Advent International and BCI complete acquisition of Maxar Technologies [Press release]. https://www.adventinternational.com/news/u-s-private-equity-firm-advent-international-and-bci-complete-acquisition-of-maxar-technologies/
Anderson, J., Sarkar, D., & Palen, L. (2019a). Corporate editors in the evolving landscape of OpenStreetMap. ISPRS International Journal of Geo-Information, 8(5), Article 232. https://doi.org/10.3390/ijgi8050232
Anderson, J., Sarkar, D., & Palen, L. (2019b, September 22). Corporate editors in the evolving landscape of OpenStreetMap: A close investigation of the impact to the map & community [Conference presentation]. State of the Map 2019, Heidelberg, Germany. https://pretalx.com/sotm2019-at/talk/W8BA9K/
Apple rebuilds mapping app. (2018, July 1). The Financial Express. https://old.thefinancialexpress.com.bd/sci-tech/apple-rebuilds-mapping-app-1530430087
Audi, BMW and Daimler €2.8 billion acquisition of Here from Nokia. (2015, December 4). IFLR1000. https://iflr1000.com/Deal/Profile/3304
Bing Maps Team. (2018, June 28). Microsoft releases 130 million building footprints in the USA as open data. Bing Maps Blog. Retrieved October 16, 2026, from https://blogs.bing.com/maps/2018-06/microsoft-releases-125-million-building-footprints-in-the-us-as-open-data
Bing Maps Team. (2020, March 18). Facebook’s street maps RapiD editor now includes Microsoft building footprints. Bing Maps Blog. https://blogs.bing.com/maps/2020-03/facebooks-rapid-editor-now-includes-microsoft-building-footprints-e9626fa48a4ed1dcd7e36c5954964c95
Brodsky, I. (2018, June 27). H3: Uber’s hexagonal hierarchical spatial index. Uber Engineering Blog. https://www.uber.com/blog/h3/
Cloud-Native Geospatial Forum. (2023, September). GeoParquet 1.0.0 released. https://cloudnativegeo.org/blog/2023/09/geoparquet-1.0.0-released/
Collins, M. (2024, September 11). TomTom’s Orbis Maps leverages open data from Overture Maps Foundation. Geo Week News. https://www.geoweeknews.com/articles/tomtom-orbis-maps-overture-maps-foundation-open-data-ecosystem/
Dempsey, C. (2018, June 28). 125 million building footprints in the US now available as open GIS data. Geography Realm. https://www.geographyrealm.com/almost-125-million-building-footprints-us-now-available-open-data/
Denham, A. (2019, July 25). Facebook releases AI mapping tools. I Programmer. https://www.i-programmer.info/news/145-mapping-a-gis/12958-facebook-releases-ai-mapping-tools.html
Ewbank, K. (2018, July 16). Google Maps API changes billing rules. I Programmer. https://i-programmer.info/news/145-mapping-a-gis/11978-google-maps-api-changes-billing-rules.html
Gabrielsson, M. (2023, April 28). PostGEESE? Introducing the DuckDB spatial extension. DuckDB Blog. https://duckdb.org/2023/04/28/spatial.html
Gao, X., Klaiber, C., & Underwood, J. (2019, July 22). AI is supercharging the creation of maps around the world. Tech at Meta. https://tech.facebook.com/artificial-intelligence/2019/7/ai-is-supercharging-the-creation-of-maps-around-the-world/
Goodchild, M. F. (2007). Citizens as sensors: The world of volunteered geography. GeoJournal, 69(4), 211–221. https://doi.org/10.1007/s10708-007-9111-y
Google DeepMind. (2025, July 30). AlphaEarth Foundations helps map our planet in unprecedented detail. https://deepmind.google/blog/alphaearth-foundations-helps-map-our-planet-in-unprecedented-detail/
Google for Developers. (2025). March 2025 changes to Google Maps Platform billing and pricing. Google Maps Platform documentation. Retrieved October 16, 2026, from https://developers.google.com/maps/billing-and-pricing/march-2025
Google Maps Platform Team. (2018, May 3). Introducing Google Maps Platform. Google for Developers Blog. https://developers.googleblog.com/introducing-google-maps-platform/
Google Research. (2023). Open Buildings V3 polygons [Data set]. Google Earth Engine Data Catalog. https://developers.google.com/earth-engine/datasets/catalog/GOOGLE_Research_open-buildings_v3_polygons
Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., & Moore, R. (2017). Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031
Haklay, M. (2013). Neogeography and the delusion of democratisation. Environment and Planning A: Economy and Space, 45(1), 55–69. https://doi.org/10.1068/a45184
He, S. (2018, May 29). From beautiful maps to actionable insights: Introducing kepler.gl, Uber’s open source geospatial toolbox. Uber Engineering Blog. https://www.uber.com/blog/keplergl/
IBM. (2023, August 3). IBM and NASA open source largest geospatial AI foundation model on Hugging Face [Press release]. https://newsroom.ibm.com/2023-08-03-IBM-and-NASA-Open-Source-Largest-Geospatial-AI-Foundation-Model-on-Hugging-Face
Jennings, C. W. (2018, August 16). What’s happening in BigQuery: Integrated machine learning, maps, and more. Google Cloud Blog. https://cloud.google.com/blog/products/data-analytics/whats-happening-bigquery-integrated-machine-learning-maps-and-more
Jones, H. W. (2018, July 8–12). The recent large reduction in space launch cost (Paper ICES-2018-81) [Conference paper]. 48th International Conference on Environmental Systems, Albuquerque, NM, United States. https://ntrs.nasa.gov/api/citations/20200001093/downloads/20200001093.pdf
Kolodny, L. (2010, December 2). Google Earth Engine revealed at COP 16. TechCrunch. https://techcrunch.com/2010/12/02/google-earth-engine-launch/
Lardinois, F. (2020, August 17). Meet the startup that helped Microsoft build the world of Flight Simulator. TechCrunch. https://techcrunch.com/2020/08/17/meet-the-startup-that-helped-microsoft-build-the-world-of-flight-simulator/
Linux Foundation. (2022, December 15). Linux Foundation announces Overture Maps Foundation to build interoperable open map data [Press release]. https://www.linuxfoundation.org/press/linux-foundation-announces-overture-maps-foundation-to-build-interoperable-open-map-data
Litwintschik, M. (2023, March 3). Geospatial DuckDB. Mark Litwintschik’s blog. https://tech.marksblogg.com/duckdb-geospatial-gis.html
Little, G. (2024, November 19). Foursquare Open Source Places: A new foundational dataset for the geospatial community. Foursquare. https://foursquare.com/resources/blog/products/foursquare-open-source-places-a-new-foundational-dataset-for-the-geospatial-community/
Lunden, I. (2015, December 4). Nokia closes its $2.8B sale of Here to the Audi, BMW and Daimler car consortium. TechCrunch. https://techcrunch.com/2015/12/04/nokia-closes-its-2-8b-sale-of-here-to-the-audi-bmw-and-daimler-car-consortium/
Marten, K., & Johns, M. (2025, December 18). Databricks spatial joins now 17x faster out-of-the-box. Databricks Blog. https://www.databricks.com/blog/databricks-spatial-joins-now-17x-faster-out-box
Monsanto acquires weather data company Climate Corporation. (2013, October 2). Agri-Pulse. https://agri-pulse.com/articles/3287-monsanto-acquires-weather-data-company-climate-corporation
Monsanto Company. (2013a). Form 10-K for the fiscal year ended August 31, 2013. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/0001110783/000111078313000205/mon-20130831x10k.htm
Monsanto Company. (2013b, November 4). Prospectus supplement (Form 424B2). U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/0001110783/000119312513427964/d619920d424b2.htm
Niantic Spatial. (2025a, May 29). Niantic Spatial: Day one. https://nianticspatial.com/blog/niantic-spatial-day-one
Niantic Spatial. (2025b, June 10). Niantic Spatial and Snap’s multi-year strategic partnership to build AI-powered map. https://www.nianticspatial.com/blog/vps-snap-investment
Niantic Spatial spin-off nets $250m funding after Scopely acquisition. (2025, March 13). PocketGamer.biz. https://www.pocketgamer.biz/niantic-spatial-spin-off-gets-250m-funding-after-scopely-acquisition
Open Geospatial Consortium. (2023). STAC Community Standard work item justification (OGC Document 23-051). https://portal.ogc.org/files/105785
Open Geospatial Consortium. (2025). SpatioTemporal Asset Catalog (STAC) Community Standard (Version 1.1, OGC Document 25-004). https://docs.ogc.org/cs/25-004/25-004.html
Overture Maps Foundation. (2025, June 25). Overture Maps Foundation launches general availability of its Global Entity Reference System [Press release]. Geo Week News. https://www.geoweeknews.com/articles/overture-maps-foundation-global-entity-reference-system-gers/
Panzarino, M. (2018, June 29). Apple is rebuilding Maps from the ground up. TechCrunch. https://techcrunch.com/2018/06/29/apple-is-rebuilding-maps-from-the-ground-up/
Pettibone, R. (2025, October 1). Maxar retires its name as Vantor and Lanteris emerge from corporate split. Forecast International FlightPlan. https://flightplan.forecastinternational.com/2025/10/01/maxar-retires-its-name-as-vantor-and-lanteris-emerge-from-corporate-split/
Pokemon GO now uses OSM globally – influx of new users [Online forum post]. (2017, December 2). OSM-talk mailing list. OpenStreetMap Foundation. https://lists.openstreetmap.org/pipermail/talk/2017-December/079816.html
Prioleau, M. (2025, February 18). Overture Maps Foundation: Making open data the winning choice. Overture Maps Foundation. https://overturemaps.org/blog/2025/overture-maps-foundation-making-open-data-the-winning-choice/
Ribeiro, J. (2015, August 3). Nokia sells Here Maps business to Audi, BMW Group, and Daimler. PCWorld. https://www.pcworld.com/article/2955852/nokia-reaches-deal-to-sell-here-business-to-audi-bmw-group-and-daimler.html
Roberts, T. G. (2022, September 1). Space launch to low Earth orbit: How much does it cost? Center for Strategic and International Studies, Aerospace Security Project. https://aerospace.csis.org/data/space-launch-to-low-earth-orbit-how-much-does-it-cost/
Roos, N. (2025, July 1). TomTom navigates 300 job cuts in AI-driven restructuring. Bits&Chips. https://bits-chips.com/article/tomtom-navigates-300-job-cuts-in-ai-driven-restructuring/
Slivka, E. (2012, September 28). Tim Cook publishes open letter on Maps for iOS 6. MacRumors. https://www.macrumors.com/2012/09/28/tim-cook-publishes-open-letter-on-maps-for-ios-6/
Sirko, W., Kashubin, S., Ritter, M., Annkah, A., Bouchareb, Y. S. E., Dauphin, Y., Keysers, D., Neumann, M., Cisse, M., & Quinn, J. (2021). Continental-scale building detection from high resolution satellite imagery (arXiv:2107.12283). arXiv. https://doi.org/10.48550/arXiv.2107.12283
TomTom. (2012, June 12). Apple to license TomTom’s content [Press release]. https://www.tomtom.com/newsroom/press-releases/general/18051/apple-to-license-tomtoms-content/
TomTom. (2024, January 9). TomTom Orbis Maps sets new standard for mapmaking with unprecedented coverage and visualization [Press release]. https://www.tomtom.com/newsroom/press-releases/general/425007330/tomtom-orbis-maps-sets-new-standard-for-mapmaking-with-unprecedented-coverage-and-visualization/
TomTom. (2025, February 4). Fourth quarter and full year 2024 results [Press release]. https://www.tomtom.com/newsroom/press-releases/earnings-other/28906/fourth-quarter-and-full-year-2024-results/
von Hippel, E. (2005). Democratizing innovation. MIT Press.
Appendix A: Source assessment (ACRR summary)
Sources were assessed for Authority, Currency, Reliability, and Relevance. The assessment is a qualitative judgment recorded for transparency, not a reproducible scoring method. Verdicts: C = Credible, UC = Use with caution.
| Source | Type | Authority | Currency | Reliability notes | Verdict |
|---|---|---|---|---|---|
| Anderson et al. (2019a, 2019b) | Peer-reviewed article; conference presentation | University researchers; ISPRS journal; State of the Map academic track | Data to 2019; corporate editing has grown since | Quantitative, transparent method; the 17 million figure comes from the presentation, not the article | C |
| Gorelick et al. (2017) | Peer-reviewed article | Platform’s own engineers | Foundational | Authors describe their own system; strong on design, not on outcomes | C |
| Goodchild (2007); Haklay (2013); von Hippel (2005) | Peer-reviewed / scholarly book | Field-defining authors | Pre-period, foundational | Conceptual framing only | C |
| OGC documents (2023, 2025) | Standards body | Primary | Current | Primary record of approval and publication dates | C |
| Monsanto (2013a, 2013b); TomTom FY2024 results (2025) | Regulatory filings; exchange-regulated results release | Primary, legally accountable | Period-appropriate | Audited or regulated disclosure | C |
| Google Maps Platform Team (2018); Google for Developers (2025) | Vendor announcements and pricing documentation | Primary | Period-appropriate; 2025 page is a living document | Authoritative on prices; silent on motive | C |
| Linux Foundation, Overture, Prioleau, TomTom, Advent, IBM, Niantic Spatial, Little (Foursquare) | Corporate self-report | Primary for what was announced | Current | Promotional; used for dates, scope, and stated intent; technical performance claims flagged as claims | C (facts) / UC (claims) |
| Bing Maps, Uber, DuckDB, Databricks, Jennings (Google Cloud), DeepMind, Meta engineering posts | Corporate technical blogs | Primary for their own releases | Current | Release facts verified; the 2018 Bing post was revised after publication and now reports a later dataset count | C (facts) / UC (claims) |
| Jones (2018); Roberts (2022) | NASA conference paper; CSIS data repository | Government researcher; established think tank | 2018–2022; launch costs have continued to change | Methods disclosed; per-kg estimates are approximations | C |
| TechCrunch (Panzarino, Lunden, Lardinois, Kolodny), PCWorld, Bits&Chips, Geo Week News, Forecast International, PocketGamer.biz, IFLR1000, Financial Express, Agri-Pulse, Geography Realm | Trade, technology, and business press | Established outlets, secondary | Period-appropriate | Each checked against the claim it supports; IFLR1000 used only for a characterization of motive | UC |
| I Programmer (Ewbank, 2018; Denham, 2019) | Technology press | Secondary | Period-appropriate | Ewbank’s free-tier and overage figures are not restated in Google’s announcement; Denham now supplemented by Gao et al. (2019) | UC |
| Litwintschik (2023) | Practitioner technical blog | Experienced data engineer; secondary | Period-appropriate | Used only for the existence and authorship of the DuckDB H3 extension | UC |
| OSM-talk mailing list (2017) | Community forum | Contemporaneous practitioner reports | Period-appropriate | Establishes timing of the Pokémon GO switch, not its motive; no Niantic confirmation located | UC |
Appendix B: Source Verification Status (October 16, 2026)
“Verified” means the cited source was retrieved and confirmed to state what the paper attributes to it. It does not certify the underlying fact beyond the source.
| Case or claim | Status | Basis |
|---|---|---|
| Apple Maps rebuild; TomTom share move | Verified | Panzarino (2018); Financial Express report of Apple’s statement to Reuters |
| HERE acquisition (€2.55 billion, December 2015; stated aim) | Verified | Lunden (2015); Ribeiro (2015) |
| Google Maps Platform 2018 change | Verified in part | Google Maps Platform Team (2018) confirms plan, credit, and June 11 billing; Ewbank’s tier and overage figures remain secondary |
| Pokémon GO switch to OSM | Partially supported | Contemporaneous OSM community report; no Niantic statement |
| Meta basemap on Daylight OSM (late 2020) | Verified | Adkins (2023) |
| Corporate OSM editing; 17 million paid edits | Verified | Anderson et al. (2019a, 2019b) |
| Microsoft building footprints | Verified with correction | Dempsey (2018); Bing Maps Team (2018) as revised |
| Facebook Map With AI; Thailand roads | Verified | Gao et al. (2019) |
| Google Open Buildings | Verified | Google Research (2023); Sirko et al. (2021) author list confirmed on arXiv |
| Uber H3 and kepler.gl | Verified | Brodsky (2018); He (2018) |
| Climate Corporation acquisition | Verified | Monsanto (2013a, 2013b) |
| Flight Simulator / Blackshark.ai | Verified with correction | Lardinois (2020) |
| Niantic Spatial | Verified as company claims | Niantic Spatial (2025a, 2025b); PocketGamer.biz (2025) |
| Earth Engine launch (2010) and design aims | Verified | Kolodny (2010); Gorelick et al. (2017) |
| BigQuery GIS | Verified | Jennings (2018) |
| DuckDB spatial extension; DuckDB H3 extension | Verified | Gabrielsson (2023); Litwintschik (2023) |
| Databricks Spatial SQL | Verified | Marten & Johns (2025) |
| GeoParquet 1.0.0; OGC working group | Verified | Cloud-Native Geospatial Forum (2023) |
| STAC approval and publication | Verified | Open Geospatial Consortium (2023, 2025) |
| Prithvi | Verified | IBM (2023) |
| AlphaEarth Foundations | Verified | Google DeepMind (2025) |
| Overture chronology, Esri membership, distribution | Verified | Linux Foundation (2022); Prioleau (2025); Overture Maps Foundation (2025) |
| Foursquare OS Places | Verified | Little (2024) |
| TomTom Orbis and FY2024 results; 2025 job cuts | Verified | TomTom (2024, 2025); Collins (2024); Roos (2025) |
| Google 2025 pricing | Verified | Google for Developers (2025) |
| Maxar acquisition and Vantor/Lanteris split | Verified | Advent International (2023); Pettibone (2025) |
| Launch-cost comparisons | Verified | Jones (2018); Roberts (2022) |
| Goodchild (2007); Haklay (2013); von Hippel (2005) | Verified (bibliographic) | Publisher and repository records |
