Sweden

What the record shows

The Swedish record is about legitimacy rather than function. Its best-known entry worked technically within weeks and still needed a vote to survive; its two failures were not broken machines but a scheme stopped by an appeals board and a service nobody rode. Stockholm is unusually good at building things that work and unusually exposed to the question of whether anyone agreed to them.

The congestion charge is the case, and it is one of the most transferable lessons in this atlas. The 2006 trial cut cordon traffic by about 22 per cent against a 10 to 15 per cent target and cut queue times by 30 to 50 per cent. Most residents had opposed the scheme before it started. A referendum that September backed it by a narrow majority, it became permanent the following year, and support has held since. The order of events is the finding: people did not approve a congestion charge and then experience freer streets — they experienced the streets first and approved afterwards. Every city in this atlas that has tried to win the argument before running the trial should read that sequence carefully, including Stockholm's own zero-emission zone, which was legislated in 2024 and wired with plate-reading cameras before a county appeals board blocked it days ahead of its planned launch at the end of that year. The dispute was still unresolved in mid-2026. The cameras were never the obstacle.

Stokab is the other pillar, and its design decision is the interesting part. The city has owned it outright since 1994 and it leases dark, unlit fibre to any operator, business or institution on equal terms — and it has never sold retail internet itself. The network now reaches over 90 per cent of regional households and 99 per cent of commercial buildings, with more than a hundred operators leasing capacity on it. A city does not have to become an internet provider to fix a broadband market; it can build the shared physical layer once and let a competitive market run on top. Set that against the same instinct meeting different law elsewhere — Chattanooga's municipal fibre, which runs into a state territorial limit a federal appeals court declined to preempt, and Cape Town's open-access network, wholesaled to private providers rather than sold to residents.

The second failure is the more poignant one, because it lasted. Autopiloten ran driverless electric minibuses on a free pilot from January 2018 and then, from that October, as a scheduled route in the regional transport network — five years of real public transport service, with a steward aboard able to take control, which is rarer in this atlas than any technical milestone. About 10,000 people rode the free pilot in its first months. By 2023, weekly ridership on the scheduled line had fallen to roughly 200 rides, and the transport authority ended the service that September. A driverless shuttle graduating from demonstration to timetable is the outcome every pilot in this atlas hopes for, and it turns out not to be proof of durability: this one was cancelled for the oldest reason in transit.

Where Stockholm has built lasting systems, the mechanism is usually a market or a metering layer rather than the hardware. Open District Heating lets data centres, supermarkets and other businesses near the pipes sell their excess heat instead of venting it; suppliers install their own heat exchangers, the utility updates temperature-indexed prices daily in an online portal and settles monthly, and more than twenty businesses have plugged in. The pipe was already there. What turned a one-way network into a two-way market is the pricing and metering layer — heat billed like a commodity, at a price published every day. The Hammarby Sjöstad vacuum waste network makes the ownership version of the same point, and it is worth reading directly against Bergen's: both are Envac pneumatic systems in a Nordic city, and in Stockholm the property-owner associations, not the municipality, own the pipes and carry the operational risk — 100 to 120 blockages a year, caused mostly by residents putting the wrong thing down the inlet.

The two district demonstrators are honest about being demonstrators. GrowSmarter retrofitted three 1960s buildings in Årsta deeply enough to cut their final energy demand by 3,333 MWh a year, and its own lesson is that replication stalls without financing models for ordinary landlords rather than for EU lighthouse consortia. Smart Energy City fitted about 150 apartments with real-time electricity, heat and hot-water monitoring plus demand response, ran from 2015 to 2018 and then closed with a final report instead of continuing as a live service — which is a legitimate ending for a research project and one the atlas records too rarely. The weak spot is the one the other Nordic capitals share: Stockholm's open data has been published by department since 2011 with no single sitewide licence, and the traffic office's own portal describes its road-network and parking data as licence-free without any confirmable dedication behind the phrase. Set beside Zurich's 937 datasets, every one licensed and most of them public-domain, it is the clearest illustration on this ladder that "freely reusable" is a description and a licence is an instrument. If the entries below agree on one lesson, it is that in Stockholm the hard part has rarely been making the thing work — it has been securing permission for it to keep working.