Whatever happened to smart cities?

Before there were traffic lights, Swedish city junctions were controlled by a person standing in the middle of them. A photograph in the collection of the National Museum of Science and Technology shows a traffic policeman at a signal post at the junction of Regeringsgatan and Hamngatan in Stockholm.¹¹ The signal is a box with two windows and the word STOPP (stop), and it is turned by hand.

So the question has never been whether traffic should be controlled, but by what. First a policeman, then a timed signal, then a detector in the tarmac, then a connected sensor. The smart city is the next link in a long chain, not a break from it.

This article tests what was promised in Sweden between 2014 and 2018, and what was reported later. The review follows four documented promises from 2014 to 2018.^{1, 2, 4, 6} Two of the four had no measurable targets.

What a smart city is, according to Swedish policy documents

The term has no definition in Swedish law. The closest description is in the government's communications to parliament, where a smart city is a city that uses digitalisation and connectivity to improve its own services, and where everything from water, sewage, energy and transport to lighting and traffic lights is connected.⁷

That the definition exists only in policy documents and not in law has a practical consequence. No agency is tasked with measuring how smart a city is, and there are no official statistics on how many connected street lights or sensors Sweden has. The figures that exist come from individual municipalities' own accounts and from evaluations of individual programmes.

Four promises, two of which could be measured

The period 2014 to 2018 produced four dated Swedish lines of promises. One of them contained both a figure and a year. One contained figures but no target year, one had a year but no figures, and one had neither.

The first was an EU-funded demonstration project with the City of Stockholm as coordinator, decided on 19 Dec 2014. The project's targets were up to 1,500 new jobs at demonstration level and 60% lower energy demand.^{6, 12} The project also included Cologne and Barcelona.^{6, 12} The second was the government's interim target of 12 Apr 2018.^{2, 13} The third was Stockholm's own strategy from spring 2017, which set the goal of becoming the world's smartest city by 2040 but left the measures to be worked out later.^{4, 14} The fourth was the government's collaboration programme for smart cities, which ran from 2016 to 2018 without concrete targets.^{1, 16}

Of the two measurable promises, one was only half reported. The demonstration project reported 123,000 square metres of renovated floor space, which is a concrete outcome, but the public final reporting contains energy results for individual measures, and we have not been able to verify an overall follow-up of the promise of 1,500 jobs and 60% lower energy demand.⁶ The second, the interim target, was reported in full and failed.

The target that could be tested

The interim target of 12 Apr 2018 is the only one with both a figure and a year: the share of passenger transport by public transport, cycling and walking was to be at least 25% in 2025, measured in passenger kilometres.^{2, 13} The starting point was given in the same document as about 20% for the period 2000 to 2016.^{2, 13}

The Swedish Environmental Protection Agency's follow-up of the environmental objectives, updated on 31 Mar 2026, gives the verdict in one line: the interim target has been assessed as not achieved by the target year.^{3, 15}

There is a detail in the 2018 communication that makes the story more interesting. Already then the government wrote that a recurring view among the consultation bodies was that following up the target could be difficult and that more policy instruments were needed to reach it, and that there were requests to specify what the target included and how it would be measured.^{2, 13} So the objection came eight years before the outcome, and it was about the same thing.

The programme that could never be tested

The government's collaboration programme for smart cities was one of five such programmes in the 2016 to 2018 term of office.^{1, 7, 16} The Swedish National Audit Office reviewed them and reported the result on 17 Mar 2020.^{1, 16}

Two sentences from that review carry the whole subject. The first: the Swedish National Audit Office has not identified any systematic follow-up work for the collaboration programmes.^{1, 16} The second: the government nevertheless assesses the programmes as successful in its reporting to parliament.^{1, 16}

It is the state's own auditor saying that the state does not know, and that it reported success anyway. In its latest follow-up of targets, Transport Analysis describes the same kind of problem in the cycling field. There, according to Transport Analysis, the Swedish National Audit Office points in particular to inadequate follow-up of the effects of the measures.^{8, 17}

Stockholm's own review

On 13 May 2026 the City Executive Office in Sweden's largest municipality published a look back at eight years of work on the smart city.^{4, 18} It is unusually direct, and it is the city's own words about its own work.

The platform few came to use

The ambition in 2017 was a shared platform for connected sensors that would meet the needs of the whole city.⁴ Three suppliers were chosen from 33 companies in a competitive dialogue, but the procurement was abandoned.⁴ The city writes that the ambition to procure both technology and expertise that way did not succeed.⁴ The platform was bought as a licence instead, and the expertise part never happened.⁴

Today, according to the look back, only a few departments and companies use the platform.⁴ The city also notes that what the world imagined as the next big technical breakthrough for a data-driven city has not become reality, and that data collected has in practice often stayed with the part of the organisation that produced it.⁴

The city gives several explanations, and one of them is not about technology. When departments and companies have bought solutions, their own needs and their own budgets have been in charge, which has meant that sharing data and services has often been given low priority.⁴

The initiatives that grew in number anyway

The same document contains the counterweight. Twelve initiatives in 2022 had become 35 by 2025, and eleven of them are established solutions in operation.⁴ So the central platform idea from 2017 became thirty-five separate initiatives, of which eleven have made it all the way to operation.

The progress also shows in how far the initiatives have come. At the 2022 survey, most of the twelve initiatives were on the first two steps of the development ladder, tests and pilots, and only three had reached higher levels.⁴ At the 2025 follow-up the weight had shifted from tests towards pilots, but most were still on the first two steps.⁴

What the money bought

The open project database of Vinnova, Sweden's innovation agency, shows what the agency funded. Between 2016 and 2024, 87,105,126 kronor went to 16 innovation platform projects in two phases.⁵ Eleven of them went to seven municipalities: Borås, Gothenburg, Kiruna, Lund, Malmö, Stockholm and Umeå.⁵ Five went to a research institute and two other organisations outside the municipalities.⁵

The database has a field where the projects themselves write what they achieved. All sixteen have filled it in, but one of them only writes that the project ended early and that results therefore cannot be described.⁵ The others list methods, networks, coordinator roles, knowledge alliances and models for working across boundaries. In one case, an internal innovation index that rose from 70 to 72 in the staff survey.⁵ The projects' own descriptions of results are mainly about organisation and ways of working.

The wording in the results field speaks for itself. One municipality reports that an internal innovation index in the staff survey rose from 70 in 2017 to 72 in 2018.⁵ Another that 134 parties were involved, 51 property owners and developers took part and 13 innovations were tested together.⁵ A third that new roles were set up at office level, such as innovation coordinators and development coordinators.⁵ A fourth that it built up interest and willingness to develop the innovation capacity of the municipal group.⁵

It is entirely possible that such work is worth the money. But what is reported is not sensors, kilometres or units in operation, and it is the funder's own database that shows it.

What is actually in operation

The current state can be counted, and it is smaller and more everyday than the promises. In 2025, 48 of 290 municipalities, or 16.6%, took part in the national climate contract, against nine cities in 2020.^{10, 19} Among the municipalities that answered an ongoing survey by the Agency for Digital Government, 30% said they had automated business processes widely or in some departments, and 45% within individual functions.⁹

On the ground, 41 kilometres of new state footpaths and cycle paths were built in 2025, and 111 kilometres of new municipal cycle paths in 2024 by the 49 municipalities that measure it.⁸ These are kilometres, municipalities and solutions in operation, measured afterwards.

The climate contract grew step by step: nine cities signed in 2020, 23 had done so by 2022, and in 2025 48 municipalities took part, of which 33 pioneer cities and 15 twin cities, together with six government agencies.¹⁰

The survey is not yet published, and the response rate among municipalities was 35% at the turn of 2025 and 2026.⁹ The figures should be read with that caveat. Besides business processes, 32% say they have automated repetitive tasks widely or in some departments, and 28% that they have automated decisions.⁹

On the traffic side, Transport Analysis also reports that 74 crossings for pedestrians, cyclists and mopeds were newly built or improved in 2025, and that the 49 municipalities that measure together invested just under a billion kronor in new cycling infrastructure in 2024.⁸

The state has changed its tone

Two government communications, seven years apart, on the same subject. The 2018 communication says that the development of smart cities and regions with digitally driven innovations such as smart grids and transport solutions should be promoted.⁷

In the 2025 communication a heading instead reads that digitalisation needs to happen with a greater focus on security, and the text notes that it is unavoidable that the smart city therefore brings new vulnerabilities.⁷

So the state itself has moved the idea from the opportunities column to the risks column, and the move is dated.

What the figures do not say

None of the figures here measures how smart a city is, because nobody measures that. They measure how much was promised, how much was followed up and how much is in operation in the few places where someone counts.

There are also no national statistics for two of the three things the episode description mentions. Traffic can be followed through the accounts of the Swedish Transport Administration and the municipalities, but for connected street lighting and for sensors in waste handling there is no combined Swedish count at all.

And the most important lesson is about method rather than technology. Clear criteria and a date for follow-up make promises easier to test. Two of the four promises in this article could never be tested, and that was not because of the technology.

What to take away

The review follows four documented promises from 2014 to 2018, two of which had no measurable targets.^{1, 2, 4, 6} The interim target of at least 25% walking, cycling and public transport in 2025 was assessed as not achieved, eleven of 35 initiatives in Sweden's largest municipality were established solutions in operation in 2025, and 87,105,126 kronor went to 16 innovation platform projects.^{3, 4, 5} What decides the outcome over decades is not the technology but whether a promise has criteria and a date, since no agency measures how smart a city is.^{1, 4} Follow the Swedish Environmental Protection Agency's follow-up of interim targets, the number of municipalities in the climate contract, and Transport Analysis's measure of newly built footpaths and cycle paths.^{3, 8, 10}

The next episode is S3:E4 Automation, about what happened to work when the machines came.

Education, not advice.