The editorial frames this as a consent betrayal: tens of millions of players walked to landmarks and scanned environments believing they were earning PokéStop refreshes, not contributing to autonomous weapons navigation. The core critique isn't that games or militaries are bad, but that the most valuable training dataset for drone targeting in a decade was assembled under a ToS that promised an AR Pikachu.
The article argues Niantic's true product was never the games — it was always the planet-scale 3D map being crowd-sourced under the guise of entertainment. The 2025 split (selling games to Scopely for $3.5B while Niantic Spatial kept the VPS data) and the immediate Vantor defense partnership reveal the mapping business was the asset all along.
The partnership pitch positions VPS as a solution to a real and growing battlefield problem: GPS jamming and spoofing now dominate contested airspace in Ukraine and the Pacific. A downward-facing camera matching terrain features against a model trained on 10M+ real-world scans can recover position without satellites, which Vantor frames as essential geospatial intelligence infrastructure for the DoD and allied militaries.
Niantic — the studio behind Pokémon Go, Ingress, and Pikmin Bloom — spent a decade convincing tens of millions of players to walk to real-world landmarks and point their phones at them. The in-game reward was a PokéStop refresh or an AR portal scan. The actual product was a planet-scale, crowd-sourced 3D map: Niantic's Visual Positioning System (VPS), a model that can localize a camera in physical space without GPS, indoors or out, by matching what the lens sees to a learned representation of the world.
In early 2025, Niantic spun the mapping business out as Niantic Spatial and sold the games division to Saudi-owned Scopely for $3.5B. The spatial company kept the data. This week, Niantic Spatial announced a partnership with Vantor — the renamed Maxar Intelligence defense arm — to use that VPS model as the foundation for drone navigation in GPS-denied environments. Vantor sells geospatial intelligence to the U.S. Department of Defense and allied militaries; "GPS-denied" is industry shorthand for contested airspace where jamming or spoofing has knocked satellites offline, which now describes most of Ukraine and a growing share of the Pacific.
The technical pitch is straightforward: a drone with a downward-facing camera can match terrain features against the VPS world model and recover its position with no satellite fix. Niantic claims its model has been trained on over 10 million player-submitted location scans. None of those scans were submitted by people who thought they were contributing to a military targeting stack.
The HN thread on this story ran predictably hot, but the substantive critique isn't "games bad, military bad." It's that the most valuable training dataset for autonomous weapons navigation in a decade was assembled under a consumer ToS that promised an AR Pikachu. Niantic's privacy policy has always reserved the right to use scan data for "research and development" and to share with "successors and affiliates." Niantic Spatial is, technically, an affiliate. Vantor is, technically, a customer of an affiliate. The chain holds legally. It does not hold morally, and the company knows it — which is why the press release leads with "safety" and "navigation" and never says "weapons."
Compare this to how the defense industry has historically acquired training data. Palantir builds bespoke pipelines. Anduril instruments its own sensors. Maxar — Vantor's former name — flies its own satellites. None of those approaches give you a model that knows what the inside of a Tokyo subway station looks like from a pedestrian's eye height. Pokémon Go got the military a dataset that satellites physically cannot collect: street-level, human-scale, indoor-inclusive visual geometry, captured under lighting conditions and at angles a soldier or drone operator would actually encounter. This is not a small upgrade. This is the kind of data advantage that wins procurement contracts.
The community reaction split along three lines. The civil liberties wing pointed out that players in countries now in active conflict — Ukraine, Israel, Iran — contributed to a map that may now be used to target their cities. The contracts wing pointed out that every consumer-app ToS contains the same affiliate-transfer language, which means every behavioral dataset at every B2C company is one acquisition away from this exact outcome. The pragmatist wing pointed out that VPS-style models will exist regardless, and a commercial dataset with public oversight is arguably better than a black-box one assembled by a defense prime.
All three are correct simultaneously. The lesson is not that Niantic did something uniquely sinister — it's that the standard SaaS-era data playbook now has a defense-industry exit, and the founders signing those ToS pages a decade ago did not price that option.
Three concrete implications if you ship product.
First, audit what spatial, sensor, or behavioral telemetry you collect that has dual-use value. Camera frames, IMU traces, audio fingerprints, gait data, indoor floor plans, BLE scans, even high-resolution timestamp patterns of user movement — all of these have a defense buyer somewhere. If you are not deliberately deciding whether to be in that market, you are passively in it the moment an acquirer wants to be. Treat the dataset as a regulated asset, not a byproduct.
Second, your Terms of Service almost certainly already authorizes this. "Successors and affiliates," "research and development," "to improve our services and develop new products" — these clauses are boilerplate, and they cover everything Niantic just did. If you want to actually constrain downstream use, you need explicit purpose limitation in the ToS, not just GDPR Article 6 lawful-basis hand-waving. EU users who scanned PokéStops have a stronger case than U.S. users, but neither has a great one. Write the limitations you want into the contract on day one — retrofitting them after a $3.5B exit is theoretical.
Third, if you're building anything that touches user-contributed spatial or visual data — AR apps, robotics SDKs, mapping tools, even something as innocuous as a fitness tracker with GPS — assume your competitors will pitch the same dataset to defense buyers. The right response is not necessarily refusal. It's pricing. A drone-navigation training corpus that took ten years and tens of millions of voluntary contributors to build is not a commodity. If Niantic Spatial's valuation triples on the Vantor deal, every founder sitting on adjacent data should be re-running their cap table math.
The Niantic-Vantor deal is the first clean public example of consumer-game telemetry powering autonomous weapons infrastructure. It will not be the last. Expect the next twelve months to surface similar arrangements at rideshare companies (street-level imagery), delivery apps (urban routing under adversarial conditions), and any robotics startup with a real-world fleet. The honest question for founders is no longer "could our data be used for this" — the answer is yes — but "what do we want the ToS, the cap table, and the press release to look like when it is."
If you are looking for something to channel that energy into, you could help improve open street map using streetcomplete: https://streetcomplete.app/
I stopped scanning pokestops because the effort has outweighed the rewards. A lot of the time, the requests show up as "research tasks" for a point of interest that I quickly passed by and have no interest in returning to, besides the tasks related to taking pictures of your buddy pokemon
This article is based on reporting from Trouw: https://www.trouw.nl/redactie/PokemonGo/I was interviewed for the Trouw piece and briefly quoted. This isn't to detract from the DroneXL piece, which adds its own angle.
This shouldn't be a surprise. But at this point it feels like if you don't completely avoid participating in digital society, your data will be used against you or groups/countries you support.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
As someone who works in this space, the headline is a bit of a stretch. The overlap in the locations of Pokemon Go Player data and any active Drone heavy theaters of war is a tiny sliver (or zero?).The military contractor (Vantar/Maxar) in question basically admits so but just "reserves th