Project Smart City: Smart Streetlights. The Savings Depend on the Tariff [2026]

The Pipe · Project Smart City 02

Project Smart City: The Streetlight Pays Back. The Smart Part Might Not.

Energy edition. The LED math is honest. The controls math depends on one question almost nobody asks first.

Written by Scot Free

Streetlights are the most common smart city project in America, and it isn't close. Every city has them, they're already on a pole with power running to it, and the pitch writes itself: cut the energy bill, add a radio, and the pole becomes a platform for everything else.

I've been on the evaluating side of one of these. The last piece was about the enclosure being right. This one is about the arithmetic, which is a separate question and the one that actually decides.

The claim

The standard smart streetlight business case has two layers stacked on one capital request:

• Layer one: swap high-pressure sodium for LED. Cut the wattage roughly in half, cut the maintenance, done.
• Layer two: add networked controls. Dim overnight, get fault reporting so a dead lamp calls in instead of waiting for a complaint, and build the mesh that future sensors ride on.

Layer one is boring and it works. Layer two is the one with "smart" in the name, and it's the one that gets the press release.

What the data shows: layer one holds up

Credit where it's due. The LED conversion case is one of the better-documented pieces of municipal arithmetic in the country, and it survives scrutiny.

• Cost: large-scale municipal LED conversions run roughly $400–$900 per streetlight installed, varying with fixture tier, labor agreements, and pole access.
• Savings: about $100 per light per year in energy, driven by replacing a 250W high-pressure sodium fixture with a 100W LED. DOE puts LED savings against legacy lighting at 25–80% depending on the baseline.
• Payback: 4–9 years on the raw numbers; most projects land at 3–5 years once utility rebates, grants, and favorable financing are applied, and DOE reports that successful municipal retrofits frequently achieve 3–5 years.
• And real projects land where you'd expect. Fifteen Iowa municipalities buying together installed 1,154 LED streetlights and estimated a 9.5-year payback including maintenance savings — at an electricity rate of $0.08/kWh. Meaford, Ontario financed 720 fixtures at a net $408,762, about $568 per fixture including financing, paying the debt over ten years.

Note what makes those last two honest: both state their assumptions. A 9.5-year payback at eight cents a kilowatt-hour is not a worse result than a 3-year payback — it's the same physics at a different electricity rate. The projects that quote three years without naming the rate, the rebate, and the grant are quoting someone else's inputs.

The pipe: what layer two actually is

Networked controls add a controller at each fixture, a radio, a head-end system, and a data path. The hardware adder is small — roughly $50–$150 per pole for networked lighting controls, or $10–$30 incremental for a Zhaga socket on a controls-ready fixture.

The claimed return is a second tranche of energy savings on top of the LED conversion. DOE model specification work put controls at roughly 33% additional energy savings over an LED system without them, added to the 25–40% from the HPS-to-LED swap. Vendor material quotes 15–30% additional, and combined figures of 50–70% total when controls accompany a conversion.

There's a maintenance return too, and it's real: truck rolls drop 30–40% when a failed lamp reports itself instead of waiting for a patrol or a resident complaint.

So far, so good. Now the question that decides whether any of it reaches the budget.

Where it breaks: the tariff

A great many American municipalities do not pay for street lighting by the kilowatt-hour. They pay an unmetered tariff: a fixed charge per fixture per month, calculated from the fixture's rated wattage and an assumed burn schedule.

Read that again with the dimming claim in hand.

If the city is billed on an unmetered tariff, dimming the lights at 1 a.m. saves the city nothing. The invoice is computed from rated wattage and assumed hours. The meter that would have noticed does not exist.

The industry says this out loud when asked directly. How much of the energy saving actually reaches the invoice depends on whether the utility bills the city on a metered or an unmetered tariff. That single sentence sits in the FAQ section of a vendor page, well below the 50–70% savings headline — which is roughly where it sits in most business cases too.

It gets more specific. In a side-by-side of municipal-owned versus utility-owned streetlight fleets, the energy savings line for the municipal-owned case is footnoted as applying at the metered rate only, and the utility-owned column has no energy savings entry at all.

Three further items that belong in the same column and usually aren't:

• Stranded asset payments. Where the utility owns the existing fixtures, converting can trigger a payment of as much as $200 per fixture for the asset being retired early. That's a real line item against the capital cost, and it lands before any savings start.
• Tariff approval is a process, not a form. Rhode Island required a Municipal Streetlight Tariff through its Public Utilities Commission as a condition of the Municipal Streetlight Investment Act. If the savings depend on a tariff change, the project's real critical path runs through a regulatory body, not a procurement office.
• The controls have an operating cost. O&M on streetlight fleets runs around $30–$48 per fixture per year in published figures, and the networked layer adds its own: the head-end subscription, the cellular or mesh backhaul, and somebody who knows what to do when a node stops reporting. That cost recurs for the life of the asset; the capital savings are counted once.

None of this makes the controls worthless. Fault reporting is genuinely valuable, and the maintenance return doesn't depend on the tariff at all — a truck roll you don't make is money you don't spend regardless of how the energy is billed. The problem is that the energy half of the claim, which is the half with the big percentage next to it, can be worth exactly zero dollars in a given city, and whether it is turns on a billing arrangement that nobody in the demo is talking about.

The demonstration problem

There's a second arithmetic trap, and it's structural rather than financial.

Downtown STL Inc. ran an open design competition and deployed 135 smart streetlights on Market Street as a $400,000 demonstration project. Work the division: roughly $2,963 per fixture, against the $400–$900 range for a conversion.

That is not a criticism of the project. A demonstration includes the design competition, a custom dual-zone luminaire, a controller platform built alongside it, and — in that case — five months of getting new small-cell, streetlight and utility-pole specifications approved by the Board of Public Service, which the city's own CTO noted normally takes far longer.

The point is that demonstration economics and deployment economics are different numbers, and the press coverage reports the first while the business case assumes the second. A pilot that cost three thousand dollars a pole proves the technology works. It proves nothing about what the citywide rollout costs, and the gap between those two figures is where a lot of these programs quietly stop.

Why the pilot was the point

Which raises the question underneath this whole series: are cities approving these to save money, or to be seen doing smart city work?

The research is less coy about this than you'd expect.

• There is no gatekeeper. Scholars note there is no official gatekeeping process for designating cities as "smart," so cities are free to adopt the moniker — capital and world cities to safeguard their status as innovation centers, and third- and fourth-tier cities to signal expansionist ambitions. The label is self-assigned, and there is a world rankings table for it.
• Grants initiate projects. In a grounded-theory study of smart technology adoption, a participant describes a project whose initiation was triggered by federal funding through a smart cities grant program. The money arrived first; the use case followed.
• And the practitioners know what happens next. From the same study: these smart projects have not been designed for longevity, and that is what has stopped a lot of councils from pursuing the next phase after the pilot, because they view it as a failure to achieve the long-term objectives. Compounding it: frequent personnel turnover on these projects, and conflicting agendas across three levels of government.
• The failure literature agrees. A 2026 review of smart city megaproject failures identifies a tendency for these projects to promise more than existing systems can deliver, and notes that municipal officials' commitment to smart city technologies correlates with a broader appreciation for economic development policy — the technology is attached to a political program, not just an operations budget.

So: sometimes yes. Not cynically, usually — a grant appears, a council wants a visible win, a pilot gets funded, and the thing genuinely works at 135 poles. Then the expansion requires a tariff fight, a capital request at deployment prices rather than demonstration prices, and a maintenance commitment that outlasts the administration that started it. Nothing failed. Nothing scaled either.

That is the same shape as Quayside: every break was governance and scope, none of it physics.

The four questions

If you're evaluating one of these — on either side of the table — these decide it, and all four can be answered before anyone builds anything:

• Metered or unmetered? If unmetered, the dimming savings are zero until the tariff changes. Ask this first. It takes one phone call and it reorders the entire business case.
• Who owns the poles? Municipal-owned and utility-owned are different projects with different savings, and utility-owned may carry a stranded asset payment of up to $200 a fixture.
• Is the quoted payback at your electricity rate? Iowa got 9.5 years at $0.08/kWh. The same hardware at a higher rate pays back faster. Neither number transfers.
• What's the ten-year operating cost of the network layer? Not the fixtures — the head-end, the backhaul, and the person. Capital savings are counted once; this recurs.

The Underground Take

I sat on the evaluating side of one of these and the technology was never the question. The honest version is that a good product can lose to a payback period, and that's not a failure of engineering.

What strikes me about the tariff point is that it's not hidden. It's published, it's in the vendor FAQ, and any utility account rep can answer it in a sentence. It goes unasked because the energy savings number is the most attractive part of the deck and nobody wants to be the person who asks the question that makes it zero. That's the same failure as the downtime benchmark: a large number everyone repeats, resting on a denominator nobody divided by.

The rule is boring and it keeps working. Before you model the savings, find out how the invoice is calculated. The pipe carries the data; the tariff decides whether the data is worth anything. Ask the tariff question first.

Next in Project Smart City: Networks & Connectivity — nine programs, one row each, and what the arithmetic looked like from the inside.

Sources

• U.S. DOE Better Buildings, Outdoor Lighting Challenges and Solution Pathways, March 2016 — stranded asset payments up to $200/fixture; networked controls at roughly 33% energy savings over LED without controls, added to 25–40% from HPS-to-LED; O&M ~$48/fixture/year; Rhode Island Municipal Streetlight Tariff and the Municipal Streetlight Investment Act; DOE MSSCL model specifications for Networked Outdoor Lighting Control Systems.
• U.S. DOE / IAMU, Iowa Municipal Utilities LED Street Lighting brief — 15 municipalities, 1,154 fixtures, ~549,743 kWh annual savings, 9.5-year payback including maintenance at $0.08/kWh.
• SFPUC / ETCC, San Francisco efficient street lighting evaluation — networked control economics, payback and NPV methodology, ~$30/streetlight/year inventory cost.
• Photometrics, "LED Conversion vs. Optimization," 2026 — $400–900/light conversion cost, ~$100/light/year energy savings, 4–9 year raw payback and 3–5 years with rebates and grants, municipal-owned vs utility-owned comparison with the metered-rate-only footnote. Cites Maryland Municipal LED Streetlight Program data and the DOE Better Buildings Outdoor Lighting Toolkit.
• Dimonoff, "Smart Street Lighting Cost and Payback" — 4–8 year combined payback, 50–70% energy range with controls, 30–40% reduction in truck rolls, and the metered/unmetered tariff caveat.
• Meaford, Ontario conversion figures ($408,762 for 720 fixtures, ~$568/fixture including financing, 10-year debt payback) and controls hardware adders ($50–150/pole; Zhaga $10–30 incremental) as published in industry cost guides.
• Labyrinth Technologies releases and St. Louis Public Radio, St. Louis on the Air, Jan. 24, 2020 — 135 streetlights on Market Street, $400,000 demonstration project for Downtown STL Inc.; dual-zone luminaire; five-month specification approval through the Board of Public Service per the city's CTO. Per-fixture figure is this author's arithmetic on published totals.
• "Navigating the institutionalisation of smart technologies for urban sustainability: A Grounded Theory Approach," Cities (2025) — grant-triggered project initiation; practitioner statements on longevity and post-pilot stalling; personnel turnover; conflicting agendas across government levels.
• "Smart cities as spatial manifestations of 21st century capitalism," Technological Forecasting and Social Change (2024) — no official gatekeeping process for the "smart" designation; tier-based signaling, citing Joss et al. 2019, Noori et al. 2020, Tan and Taeihagh 2020.
• "Why smart city megaprojects failed: Learning from the past and planning for the future," Cities (2026) — overpromising relative to system capability (citing Kitchin 2021); correlation between officials' smart city commitment and economic development policy orientation.

A note on these sources: cost and payback figures in this category come from a mix of federal program documentation, utility studies, and vendor material, and the vendor figures tend to sit at the favorable end of every range. Where a number is load-bearing here — the tariff caveat and the controls savings adder — it is sourced to DOE documentation or stated by the vendor literature against its own interest. All per-fixture divisions of published project totals are this author's arithmetic and are labeled as such.

Corrections and additions: Scott@IoTunderground.com.