The Price Nobody Publishes

The largest sum of money in this atlas is nineteen billion dollars, and it was never spent. It belongs to PlanIT Valley, a city that Portugal announced outside Paredes in 2008 with a full technical prospectus and a planned total investment of roughly $19 billion that was never raised. The city was never built. The number survives anyway, because a number that is announced circulates, and a number that is spent often does not.

That is the shape of the public record on smart-city money, and it is worth stating plainly before anyone builds a ranking on it: of the 1,077 published entries in this atlas, 215 state a cost — under a fifth. The rest say nothing. We can therefore rank what cities have disclosed, and we have, on the costs page. What we cannot do is tell you what the most expensive smart-city projects are, and neither can anyone else.

Where the silence falls

If disclosure were random, that would be an annoyance rather than a finding. It is not random. Transport entries disclose a cost roughly two and a half times as often as governance entries, and the gap follows a mundane logic: where a physical thing has to be procured, a contract value usually has to be tendered in public and a figure exists whether or not anybody wanted it to. Where the spending is software licences, consultancy days and staff time, no such moment arrives, and the cost simply never becomes a document.

This matters more than it sounds. It means the visible part of smart-city spending is systematically the part that looks like traditional infrastructure — the metros, the tunnels, the barriers — while the part that is distinctively "smart" is the part that disappears. A reader who studies the disclosed figures learns about civil engineering procurement. The platforms, the dashboards, the analytics contracts, the multi-year vendor relationships: those are mostly absent, and their absence is not evenly felt, because they are exactly what a city considering a smart-city programme is trying to budget for.

The number that is published is not the number that was spent

Even where a figure exists, it is rarely the figure a reader assumes. Cairo's East Nile monorail is reported at three different sums by its own sources: about €2.7 billion for the construction contract, about €4.1 billion once thirty years of operation and maintenance are included, and about $4.5 billion as the 2019 contract value on that same basis. None is wrong. They are answers to different questions, and nothing on the face of any of them tells you which question was asked. This atlas carried the line under two separate records for a while, each quoting a different one of those figures, which is how easily the confusion propagates — the duplicate was withdrawn in August 2026 and the surviving entry now carries all three.

The same problem runs through the whole set. Sydney's driverless metro is the largest disclosed cost in the atlas, and the entry itself carries two figures — A$8.3 billion for the Northwest line, and roughly A$20.5 billion for City & Southwest after 2023 budget revisions. Chicago's deep tunnel system reports over $3 billion across thirty years of construction beginning in 1975, a nominal figure that has never been restated in today's money by any source we could find. Washington D.C.'s court-ordered sewer programme was estimated at $2.6 billion when the consent decree was signed in 2005 and is now put at about $3.5 billion.

None of these is a scandal. All of them mean that a league table of costs is a league table of accounting conventions unless somebody states, for every row, what the figure covers and when it was struck. That is the reason the ranking on the costs page carries a "what the figure covers" column and will look less impressive than a cleaner-looking list somewhere else. The cleaner list is the less honest one.

The part we cannot prove

Here is the claim this essay is actually about, and it needs to be labelled precisely, because it is a suspicion and not a finding.

The developments most likely to sit above everything in our ranking are in this atlas already. When this piece was first published, all seven of them stated nothing at all. We then went looking, source by source, and the result is worth reporting because it did not come out the way we expected.

Three of the seven turned out to have a figure that meets this atlas's standard, and now carry one. Shanghai's Yangshan Phase IV terminal was approved by China's National Development and Reform Commission in November 2014 at about ¥12.848 billion, 30% equity and the rest bank credit. President Joko Widodo put the cost of Indonesia's new capital at a provisional Rp 466 trillion when he spoke at the site in March 2022, of which he expected 19–20% to come from the state budget. Masdar City was announced in 2008 at $22 billion and cut twice during 2010 — first by 10–15%, then by a further $2.5 billion to $16 billion.

Read those three again and the pattern is the point. One is a regulatory approval. One is a provisional requirement announced at a groundbreaking. One is a budget that was revised downwards twice within two years. Not one of them is an account of money spent. The Yangshan approval is the closest thing to a hard number in the set, and it is a permission to spend issued three years before the terminal opened; no statement of what it finally cost has been published. Masdar's is the most instructive, because the cuts name what was deleted to reach them — the raised podium, rooftop solar on every building and the driverless pod network, which is to say the features the city had been announced for.

The other four — NEOM's The Line, Songdo, Sidewalk Toronto's Quayside and Singapore's Tengah district — still state nothing. Large figures circulate for several of them, from announcements, promotional material and press repetition rather than from any published project account, which is why this atlas does not carry them and why the costs page does not rank them.

So the silence is narrower than we first wrote and worse than it looks. What a determined reader can obtain for the biggest projects in this field is what somebody intended to spend, sanctioned, or hoped to raise. What nobody publishes is the other document — the one written afterwards.

And the suspicion itself survives the correction, so it should be stated plainly one more time: we think the real top ten is mostly composed of projects whose spending has never been accounted for, and we cannot demonstrate it. The evidence that would settle the question is precisely the evidence nobody publishes. Anyone who tells you confidently which smart-city project is the most expensive in the world is either working from an announcement — the PlanIT Valley error, dressed up — or has access to accounts we would very much like to see.

What does follow

One thing survives without any speculation at all. A cost that is never published cannot be compared against an alternative, cannot be audited after the fact, and cannot be used by the next city to plan with. Every procurement officer who has tried to benchmark a platform contract against a comparable city has run into this: the comparable city's number does not exist.

That is a smaller claim than "cities are hiding what they spend", and it is the one the record actually supports. It is also, unfortunately, the more expensive problem. Concealment can be legislated against. A number that was never produced in the first place has to be created before it can be disclosed — and nothing in the current arrangement asks anyone to produce it.

The cases in this atlas where a cost is known and instructive are, almost without exception, cases where something went wrong and an outside body went looking: an auditor, a court, a parliamentary committee. Busan and Gimhae's light rail payments are public because a revenue guarantee turned them into a recurring line in two municipal budgets, not because anybody chose to publish them. That is a poor mechanism for producing a public record, and at the moment it is the main one we have.