A roadside camera can capture a vehicle in a fraction of a second. The lasting power comes later, when that observation enters cloud infrastructure designed to store, search, copy, analyze, and share records at scale.
The cloud has a physical address
“Stored in the cloud” can make data sound weightless. It is not. Cloud services run on physical servers inside data centers connected by fiber, power systems, cooling equipment, backup facilities, and administrative networks.
For surveillance systems, this infrastructure is the difference between seeing something and remembering it. A camera is a sensor. The data center gives the sensor durable memory, search tools, user accounts, audit logs, backup copies, and connections to other systems.
Flock Safety's license plate reader policy says LPR data is transferred from its cameras to AWS Government Cloud. Axon says its cloud infrastructure uses Microsoft Azure and Amazon AWS to support evidence storage, databases, artificial-intelligence services, and integrations. These architectures can offer security, reliability, and centralized administration. They also move consequential public records into privately operated infrastructure governed by a mix of law, contract, agency settings, and vendor design.
Collection, storage, and analysis are different decisions
Public discussions often treat a new camera as one decision. In practice, several decisions are bundled together:
- Collection: What does the sensor capture and where does it operate?
- Transmission: How is the record encrypted and sent away from the device?
- Storage: Which cloud environment receives it, and in which legal jurisdiction?
- Retention: When is the routine record deleted, archived, exported, or preserved?
- Analysis: Can software search vehicle attributes, correlate records, or apply automated classification?
- Sharing: Which agencies, prosecutors, contractors, or partner systems can obtain access?
A community may support one part of that chain and object to another. It may accept a short-lived alert for a stolen vehicle but reject broad historical searches. It may support cloud storage with strong auditing but oppose automatic sharing across a large network. Those distinctions disappear when the entire system is debated simply as “a camera.”
Data centers make scale inexpensive
Cloud platforms are built to scale. Adding another camera or another category of digital evidence does not require a city to construct a new evidence room. Storage, search, and computing capacity can expand inside a vendor platform.
That scalability is operationally useful. It can reduce lost evidence, improve chain-of-custody records, and make authorized files available to investigators and prosecutors. Axon emphasizes those benefits in its public material. The same scalability can make it easier to retain more information, combine more sources, and search larger populations. Capacity is not policy. The fact that a system can preserve or correlate a record does not answer whether it should.
Deletion must follow the copies
Retention rules sound simple until a record moves. The original observation may have one deletion schedule, while an alert, export, case file, shared copy, backup, or legal hold follows another. A serious deletion policy should explain the status of each copy rather than describe only the source database.
Public agencies should also know how deletion is enforced, how exceptions are approved, how backups age out, and whether a departing customer can export or fully remove its data. Vendor assurances matter, but contracts, audit rights, and public reporting turn assurances into something that can be tested.
The infrastructure questions citizens should ask
- Which cloud providers and regions store the records?
- Which party owns the data and controls encryption keys?
- Are vendor employees able to access customer records, and under what controls?
- How are searches, downloads, exports, and sharing events logged?
- What happens to backups when the primary record is deleted?
- Can the system connect to other databases without a new public approval?
- What happens to records when the contract ends?
- Which breach, outage, and incident notifications are required?
None of these questions assumes a breach or improper search has occurred. They recognize that public accountability must extend beyond the pole at the roadside to the infrastructure that stores what the pole sees.
The permanent record is an architecture
Modern surveillance is not one camera, one company, or one database. It is a chain of sensors, networks, cloud platforms, policies, people, and permissions. The camera is the visible part. The data center is where visibility becomes institutional memory.
Communities deciding whether to build that memory deserve a complete diagram: what enters, where it goes, who can retrieve it, when every copy disappears, and who is responsible when the rules fail.
This report is independent commentary based on public vendor and government documentation. References to cloud architecture describe published product information and do not identify the physical location of any specific customer record.
Sources on file
- Flock Safety License Plate Reader Usage and Privacy Policy
- Axon: What's Changing in Axon Cloud in 2026
- Axon: Cloud Storage for Digital Evidence
- U.S. Government Accountability Office: Smart Cities, Technologies, and Transparency
CORRECTIONS: Send documented corrections to support@ministryofdissent.com.
