How countries rank their smart cities

Every autumn, German newsrooms write the same headline: which city topped the Bitkom index this year. Every spring, Zurich celebrates another win in a Swiss index built from opinions rather than sensors. And the EU quietly refuses to rank anything at all. Behind the scoreboards sit four very different ideas of what "best" even means — and none of them measures the thing this atlas keeps finding: whether projects actually work.

Germany: measured yearly, by fixed rules

Germany's scoreboard is the Bitkom Smart City Index, published annually since 2019 by the digital-industry association Bitkom. It is the strictest of the four models: every German city above 100,000 inhabitants is scored on the same 37 indicators (13,529 data points in the 2025 edition) across five fixed categories: administration, IT and communication, energy and environment, mobility, and society and education. The data comes from the cities themselves plus official statistics, normalised so small cities can compete with big ones.

Munich has led the last three editions (2023–2025), scoring 90.2 points in 2025 ahead of Hamburg (89.6) and Stuttgart (88.0). Munich's smart-city groundwork is older than its winning streak — its EU-funded Smarter Together district retrofit ran from 2016.

Two things about that construction are worth pausing on. The scoring body is the industry association for the sector being scored — the same organisation whose members sell the technology a good rank rewards. And most of the input is supplied by the cities themselves, which means the index measures, in part, how thoroughly a municipality documents its own digitalisation. Neither makes the ranking worthless; both mean a place can move several positions without a single resident noticing a change. Wiesbaden's DIGI-V, which rebuilt the control and sensing layer of all 227 signals on the city's main road network, is one line in such a table — the same one line a city earns for a pilot on a single street.

The global yardstick asks a different question entirely

The world's most-quoted list, the IMD Smart City Index from Lausanne (first edition 2019, 148 cities by 2026), doesn't count sensors or e-government services at all. It asks residents how their city feels (about 400 per city in the 2026 edition, up from roughly 120 in earlier years): health and safety, mobility, activities, opportunities, governance — smoothed over a three-year average and grouped by human-development tier so Lagos isn't graded against Geneva.

That design produces a very different leaderboard. Singapore won the first edition in 2019; Zurich has topped every edition since 2023, most recently in March 2026 ahead of Oslo and Geneva. Nobody would claim Zurich deploys more urban tech than Singapore or Seoul — but its residents are exceptionally satisfied, and that is literally what this index measures.

The EU deliberately doesn't rank — it selects

The European Union runs the third model: no scoreboard, but a mission. In April 2022 it selected 112 cities (100 from member states, 12 from associated countries, out of 377 applicants) to become climate-neutral by 2030. Selected cities get tailored support through the NetZeroCities platform and can earn the EU Mission Label once the Commission endorses their Climate City Contract — 107 cities held the label by mid-2026. Nine German cities are in: Aachen, Dortmund, Dresden, Frankfurt, Heidelberg, Leipzig, Mannheim, Munich and Münster.

The logic is the inverse of a ranking: instead of grading everyone once a year, pick the committed few and hold them to a contract.

Where there's no index, there's an award

The fourth model is the jury prize. The biggest is handed out at Smart City Expo World Congress in Barcelona, the industry's largest gathering (running since 2011), whose World Smart City Awards drew 462 entries from 65 countries in 2025. The 2025 city title went to Rome for its data, 5G and digital-twin strategy ahead of the Jubilee year; 2024's went to Shenzhen — a city this atlas knows well for the world's first fully electric bus fleet and Tencent's car-free Net City. And a win travels: Tel Aviv still cites the "World's Smartest City" title it took here in 2014 for its DigiTel resident card. Awards reward showcase projects and momentum; they are marketing as much as measurement, and cities treat a win accordingly.

Four models, four definitions of "best"

The lists your search results are built from

Beneath those four models sits a fifth layer with no methodology at all: the "top smart cities" listicle. It matters because search snippets and answer machines feed on it. We compared six recent lists name by name — Wikipedia's, BCG's AI-cities report, Interesting Engineering's, earth.org's, a software vendor's evergreen top ten and an aggregator's top twenty. Together they name 63 different cities, and 44 of them appear on exactly one list — seven in ten. Only five cities make four or more of the six lists: Amsterdam, Dubai, London, New York and Singapore. The starkest pair: the aggregator cites Interesting Engineering as its own measuring stick, yet shares only two cities with it — London and Dubai. Lists that were measuring the same thing could not disagree this much.

The two most rigorous global sources are careful in a different way: neither the 2026 IMD report nor BCG's AI-cities index names a single concrete project or system — 180 pages on 148 cities without one named deployment between them. A ranking that names nothing it could be checked against cannot tell you whether anything it rewards actually runs.

And the IMD's own analysis quietly sides with the sceptics. Its 2026 report notes that almost every city among the bottom twenty of the 2026 ranking has a higher average Technology score than a Structures score — in the index's own data, governance out-predicts hardware. That is the pattern this atlas keeps finding from the project side: sensors are procurable, responsibility is not.

What the scoreboards miss

Research keeps landing on the same two critiques. Indicator rankings mostly count inputs — technology deployed, services offered — rather than outcomes like cleaner air or faster commutes; recent academic work argues rankings should be rebuilt around livability instead. And perception surveys inherit the biases of who answers them: studies of citizen-reported data show low-income neighbourhoods systematically report problems less, so "how it feels" data can flatter cities that are pleasant for the people most likely to be surveyed. That reporting bias runs far deeper than rankings; our deep-dive on what 311 and civic-reporting maps actually measure traces it through Boston, New York and Bangkok.

No current ranking measures the thing that decides success in this atlas again and again: not whether the hardware exists, but whether anyone actually operates it.

— the pattern across our documented entries, from Dhaka's unused signals to Sydney's fifty-year-old running ones

The source-checked projects here make the point concretely: Dhaka and Algiers installed automated traffic signals that never ran, while Sydney's SCATS — the same idea, older hardware — has adapted signal timings continuously since 1975. A ranking that counts "adaptive signals: yes" scores all three the same. That gap between deployment and operation is exactly where rankings end and this atlas begins.