Array of Things
Chicago called it a fitness tracker for the city and announced 500 sensor nodes by the end of 2018. About 130 ran, and they were retired in September 2021.
Data & IoT is the plumbing category: it rarely has a ribbon-cutting of its own, but it's what every other category quietly runs on. A traffic signal that "negotiates" with its neighbours, a grid that balances solar supply — all of it depends on sensors that report reliably and platforms that can make sense of what they report. It's also the category where cities have made their most public mistakes, which makes it one of the most honest to read.
The first wave of urban sensor networks, from the early 2010s, was mostly research-driven: Santander's SmartSantander testbed, EU-funded from 2010, federated over 20,000 sensors across four European cities — 12,000+ of them under Santander's own streets and lampposts alone — to answer a fairly academic question: what can city-scale sensing actually tell you? Chicago's Array of Things, launched in 2016, took a similar research pedigree public, putting 100+ modular nodes on lampposts to measure air quality, noise and pedestrian flow as fully open data — a "fitness tracker for the city," in its own framing.
The second wave shifted from sensing to modelling: not just collecting data, but building a live, queryable copy of the city itself. Rotterdam's digital twin now tracks over 80 ships a day and 460 million tonnes of cargo a year through Europe's busiest port, alongside the city's flood and energy-transition planning. Virtual Singapore went further still — a roughly $53 million, National Research Foundation-led 3D model completed in 2022, the first digital twin built for an entire country rather than a single city or district.
Not every sensor deployment survived contact with the public unchanged. Hong Kong's 400-lamppost smart-pole pilot triggered a privacy backlash in 2019 severe enough that protesters tore several down, forcing the government to disable Bluetooth vehicle-tracking and licence-plate recognition; an advisory panel later recommended camera-free alternatives like LIDAR instead. San Diego's US$30 million "CityIQ" network — 3,000+ nodes with cameras, microphones and environmental sensors — was deactivated in September 2020 after the ACLU raised concerns about police access to the footage, only returning in 2023 under a narrower, 500-camera agreement with new operators. Both are included here deliberately: a "smart city" data project is also a governance question about who can see what, and these two show what it takes to answer that question after the fact rather than before.
Expect digital twins to keep expanding in scope (from single buildings to whole ports to whole countries), open-data platforms to remain the least controversial and most durable pattern in this category, and privacy-by-design — sensors that report aggregate counts rather than raw video or Bluetooth MACs, settled before deployment rather than after a backlash — to become the default lesson cities take from cases like Hong Kong and San Diego.