Disaster · Water Security

Why So Many Big Cities Are Running Out of Water — and Why Fixing Leaks Beats Building Dams

📅 Sep 23, 2026 🏷️ Explainer · Evergreen 🕐 9 min read

By Luminesca · Updated Sep 23, 2026
Analysis compiled from public reporting with AI-assisted drafting. See our editorial policy.

Water stress used to be a story about deserts. In 2026 it is increasingly a story about cities — including cities that get plenty of rain. Mapping by Watershed Investigations and The Guardian found that half of the world's 100 largest cities draw from catchments under high or extreme water stress, with 38 of them in the most severe band: Beijing, Delhi, Los Angeles, New York, Rio de Janeiro. Satellite gravimetry analysed at University College London shows about 1.1 billion people living in metropolitan regions with a long-term drying trend, against roughly 96 million in cities that are getting wetter. The United Nations now describes the condition as water bankruptcy. The uncomfortable part is where the diagnosis points: at pipes, meters and the aquifers beneath the city, not at the sky.

Half the biggest cities, one shared problem

The mapping is blunt. Fifty of the 100 largest cities sit where withdrawals for public supply and industry are approach or exceed what the catchment can sustainably yield, and 38 are classed as extremely stressed.

Direction of travel matters as much as level. Using two decades of NASA's GRACE satellite gravity data, researchers at UCL measured total water storage over major urban regions. Chennai, Tehran and Zhengzhou show pronounced drying; Tokyo, Lagos and Kampala show the opposite. The resolution is coarse — satellites see regions, not neighbourhoods — but the direction is clear.

The institute behind that framing, the UN University Institute for Water, Environment and Health, argues "crisis" is the wrong word, because a crisis implies recovery. Its 2026 assessment estimates that nearly 75 per cent of the world's population lives in a water-insecure country, that more than 70 per cent of major aquifers are in long-term decline. The World Bank separately puts global freshwater loss at about 324 billion cubic metres a year, roughly the annual needs of 280 million people.

The cheapest new water is the water you already have

Before a city hunts for new supply, it should count what it is already losing. On World Bank benchmarking, utilities leave roughly a third of the water they treat unbilled — pumped, treated and pressurised, then lost to leaks, illegal connections, faulty meters or broken billing. Programme utilities in southern Africa all reported losses above 45 per cent.

This is not a poor-country problem. France's networks run at about 80 per cent efficiency, losing close to a billion cubic metres a year. The United States loses more than two trillion gallons of treated water annually, at a cost industry bodies put near $6.4 billion. The Netherlands holds leakage at 5 to 6 per cent after decades of disciplined maintenance — which is the point: the difference is managerial, not geographic.

Burkina Faso's utility ONEA kept non-revenue water below 20 per cent for years in one of the world's poorest countries, then saw losses creep back up: efficiency is maintained, not achieved. Ghana reports 45.5 per cent; Addis Ababa's own water balance put losses near 42 per cent. At a World Bank workshop in Pretoria this March, five utilities mapped a route from above 45 per cent to 35 per cent — 31 million cubic metres a year, enough for a million people, worth $57 million annually.

The aquifers are the part you cannot refill

Reservoirs recover after a wet winter. Aquifers do not, at anything like the same speed. A 2026 analysis built on 21.5 years of GRACE satellite gravity data finds groundwater depletion dominating continental freshwater decline, most heavily in Asia, where the preprint measures roughly 55 cubic kilometres a year. Africa is the one large region where groundwater is rising; parts of Europe are edging towards long-term decline.

Over-pumping is hard to reverse for a physical reason. When an aquifer is drawn down faster than it recharges, pore spaces collapse and the ground subsides — parts of the Tehran plain are sinking tens of centimetres a year. That damages pipes and foundations and permanently reduces how much the formation can hold: the loss is capacity, not just volume.

Tehran is this year's clearest case: a sixth consecutive year of drought, reservoirs down to single-digit or low-double-digit percentages of capacity, pressure halved for most households, and a president who has raised rationing and evacuation as options. Urban demand grew from about 346 million cubic metres in 1976 to 1.2 billion.

Where the water actually goes

Two consumers dominate. Agriculture takes about 70 per cent of global freshwater withdrawals, much of it by flood irrigation that evaporates or runs off before crops use it; drip systems typically cut use by 30 to 50 per cent while holding yields.

For cities the question is less how much water exists than how much is reused. Singapore's NEWater reclamation meets about 30 per cent of demand and targets 55 per cent by 2060. Israel treats close to 90 per cent of its domestic wastewater and sends about 85 per cent to farms. Globally the picture is far behind: the domestic sector generates some 267.5 billion cubic metres of wastewater a year and barely 55 per cent of it is treated at all, leaving well over 120 billion cubic metres discharged with little or no treatment.

Data centres are the newest claimant. One 2026 industry assessment puts AI-related water consumption at up to 9.3 trillion litres by 2030 — comparable to the annual domestic needs of about 1.3 billion people in sub-Saharan Africa, on top of their electricity demand.

Why desalination only half-works

For a coastal city the obvious answer is the sea, and the technology has improved. Modern seawater reverse osmosis with isobaric energy recovery uses 2.5 to 4 kilowatt-hours per cubic metre, against 7 to 12 without it — a reduction of 60 to 70 per cent that turned desalination from a last resort into mains supply for Dubai, Tel Aviv and Singapore.

It is still the most expensive new water in the portfolio. Levelised costs for a well-run seawater plant in 2026 sit around $0.45 to $0.80 per cubic metre, dominated by electricity, with the best plants — the UAE's Taweelah, at over 900,000 cubic metres a day — near $0.49. The harder constraint is the byproduct: at roughly 45 per cent recovery, every cubic metre of fresh water leaves about 1.2 cubic metres of brine at about twice seawater salinity, and a UN-backed estimate puts global brine output near 142 million cubic metres for every 95 million of drinking water. Brine disposal is now a first-order permitting problem.

So the policy shift is towards efficiency rather than concrete. The World Bank's April 2026 action plan targets water security for a billion people by 2030 through leakage control, irrigation modernisation, reuse and data-driven planning. Its water practice is explicit that returns come fastest where water is already being lost.

What it means

Water is becoming a constraint on growth rather than a background utility. UN-Habitat projects urban populations rising from 55 per cent of the world to 68 per cent by 2050, urban water demand up 50 to 70 per cent within three decades, and about 1.9 billion urban residents facing seasonal shortages by mid-century. The World Bank warns unmanaged water stress could cut regional GDP by up to 6 per cent by 2050.

There is also a financial loop that makes the problem self-reinforcing. Utilities with high non-revenue water score badly on creditworthiness, which limits commercial borrowing, which keeps networks leaking. Lenders now treat operational efficiency as a precondition, not an aspiration: cutting losses is the cheapest new supply, and for many operators the only route back to bankability.

What changes next is likely to be more administrative than dramatic: tariff reform and ring-fenced utility accounts, performance-based leakage contracts, real groundwater monitoring, reuse mandates, and permit conditions that make large water users account for what they take. Cities rarely run dry in one morning. They run a deficit for years — in pipes and in ledgers — until a dry year makes it visible.

Frequently asked questions

If so many big cities are short of water, why do they keep growing?

Because the constraint rarely stops growth — it changes its price. Cities can import water, food and manufactured goods, so a shortfall usually shows up as higher bills, rationing, trucked supply, subsidence or lost industrial output rather than a hard stop. That is exactly what makes the problem easy to defer: UN-Habitat expects urban populations to rise from 55 per cent of the world to 68 per cent by 2050, and the fastest-growing cities are often in the regions with the strongest drying signals.

What does 'day zero' actually mean?

Day zero is the point at which a city's piped municipal supply can no longer meet basic demand — taps are shut off, and water arrives by tanker or standpipe. Cape Town approached it in 2018 and avoided it through drastic consumption cuts; Chennai's main reservoirs ran nearly empty in 2019; Tehran is the current case, in its sixth consecutive drought year. The term is useful as a warning and misleading as a threshold, because most cities degrade toward shortage over years rather than crossing a line in one day.

Isn't desalination the answer?

It is part of the answer for coastal cities, not a general fix. Seawater reverse osmosis has become far cheaper — levelised costs of roughly $0.45 to $0.80 per cubic metre for a well-run plant in 2026, with energy recovery cutting consumption by 60 to 70 per cent — but it remains the most expensive new water available, it is energy-intensive, and it produces about 1.2 cubic metres of brine for every cubic metre of fresh water at typical recovery rates. For an inland city it is usually not an option at all.

Why does groundwater matter more than reservoirs?

Because reservoirs refill and aquifers largely do not. A wet year can refill a reservoir within a season; an over-pumped aquifer recharges over decades, and once pore spaces collapse the ground subsides and the formation permanently holds less water. The 2026 GRACE-based assessment found groundwater depletion to be the dominant contributor to continental freshwater decline, with more than 70 per cent of major aquifers in long-term decline. That is why monitoring matters more than the headline storage figure.

What can a single city realistically do?

Four things, in rough order of returns. Find and bill the water it already treats: non-revenue water of 35 to 45 per cent is common, and performance-based contracts have cut it by ten points. Meter and price honestly, with utility accounts ring-fenced from municipal budgets so the money goes back into the network. Reuse wastewater rather than discharging it, following Singapore's and Israel's examples. And manage demand on the agricultural and industrial side, where drip irrigation cuts use by 30 to 50 per cent and where new large users such as data centres can be asked to account for what they take.