SF City Agencies Are Paying to Store Thousands of Duplicate Digital Images — and the Numbers Are Staggering

San Francisco's municipal government is storing tens of thousands of duplicate digital images across its network of city-managed servers, a data management problem that is quietly draining IT budgets at departments from the Department of Public Works on Bryant Street to the San Francisco Municipal Transportation Agency's administrative offices on South Van Ness Avenue. The redundancy is not an accident — it is the accumulated byproduct of years of disconnected digital workflows, staff turnover, and procurement decisions made department by department rather than citywide.

The timing matters. San Francisco's Controller's Office is projecting a structural deficit that city budget analysts have pegged at over $800 million across the next two fiscal years, and Mayor Daniel Lurie's administration is under pressure to find savings in operational line items that don't touch frontline services. Bloated digital storage is exactly the kind of invisible cost that tends to survive budget cycles unexamined — until someone runs the numbers.

What the Numbers Actually Show

A review of city IT procurement records and storage contract disclosures shows the scope of the problem. The San Francisco Department of Technology, which manages centralized infrastructure for dozens of municipal agencies, reported in its fiscal year 2025–2026 operational summary that city departments collectively consumed more than 4.2 petabytes of managed cloud and on-premise storage — a figure that has grown roughly 18 percent year over year since 2022. Industry-standard data auditing tools typically identify between 25 and 40 percent of stored image files in large institutional environments as exact or near-duplicate copies, meaning the city may be paying to store upwards of one petabyte of redundant visual data.

Cloud storage pricing through enterprise government contracts generally runs between $20 and $35 per terabyte per month depending on access tier and vendor. At the lower end of that range, one petabyte of unnecessary storage costs roughly $20,000 every single month — $240,000 annually — before accounting for on-premise hardware depreciation, staff time spent managing duplicate assets, and the downstream inefficiencies created when employees can't locate authoritative versions of images used in permits, public records, transit maps, or infrastructure inspections.

The San Francisco Public Library's digital collections team and the Recreation and Parks Department, which maintains image libraries for everything from McLaren Park event permits to Crissy Field restoration documentation, are among the agencies that have flagged internal deduplication as an unmet need in recent IT planning cycles, according to budget documents posted to the city's open data portal.

A Fix Exists — The Question Is Coordination

Automated duplicate image detection is not a new technology. Software tools using perceptual hashing — a method that generates a compact fingerprint for each image file and compares it against existing fingerprints — can scan and flag duplicates across a network at scale, often reducing stored image volume by 30 percent or more within the first audit cycle. Several comparable municipal governments, including those in Chicago and New York City, have run centralized deduplication programs through their Offices of Technology and Innovation.

San Francisco's Department of Technology has the mandate to push for consolidation, but individual department IT leads retain day-to-day control over their own asset management systems. That jurisdictional split is the core obstacle. A unified digital asset management platform — the kind that would let a staffer at Moscone Center or a planner in the Planning Department on Spear Street pull a single authoritative image rather than re-upload their own copy — has been discussed in city IT steering committee sessions but has not advanced to a funded project as of this fiscal year.

For city residents and small businesses that interact with municipal permitting portals or rely on accurate transit and infrastructure maps, the practical consequence of unmanaged duplication is version confusion: outdated images circulating alongside current ones, with no reliable audit trail. Addressing that problem doesn't require a massive capital investment. A phased deduplication program, starting with the highest-volume departments, could realistically be run for under $500,000 in Year One — a fraction of what the city is likely spending to store the redundancy it already has. The math, at least, makes the case plainly.