A camera recording around the clock at 2 Mbps writes 21.6 GB a day. A hundred of them write 2.16 TB. Whether it is CCTV storage for a camera estate or a continuously recorded service, nobody argues with those numbers, and almost nobody budgets from them correctly.
The mistake is treating video recording storage cost as something that grows forever. For continuous recording it does not. Once the oldest footage starts being deleted as new footage arrives, the amount stored stops rising and the bill flattens into a plateau. Where that plateau sits is set by three numbers, and the one you control most directly is not the camera or the provider. It is the retention policy.
The formula behind video recording storage cost
For continuous recording, the amount you store at steady state is:
stored data = cameras × bitrate × seconds per day × retention days
and the monthly bill is that figure times the price per GB-month. There is no growth term. A 30-day policy stores 30 days of footage on day 31 and on day 900 alike.
That is the key difference from per-session recording. A service that records verification calls and keeps them for years grows every month — the shape priced in the compliance trap in recording at scale. A camera estate with a fixed retention period reaches its ceiling in one retention cycle and stays there.
So the question is what goes into each of the three terms of video recording storage cost.
Bitrate: the number you can't look up
Bitrate is the term people most want a standard figure for, and there isn't one. It depends on resolution, codec, frame rate, scene content and encoder settings. How much it varies is visible in Axis's own Zipstream white paper: the same storage-reduction technology cuts bitrate by 25% on a well-lit scene with occasional motion, 50% on a daytime overview, and 90% on a dark scene with occasional motion. No single number describes all three.
This post therefore uses 2 Mbps per camera as a round planning assumption, with 1 and 4 Mbps either side to show how sensitive the result is. Replace it with the average your own recorders report — the average, not the configured maximum, because the average is what gets written to disk.
Average bitrate per camera | Written per day | Written per 30 days |
|---|---|---|
1 Mbps | 10.8 GB | 324 GB |
2 Mbps | 21.6 GB | 648 GB |
4 Mbps | 43.2 GB | 1,296 GB |
Codec choice lives in this term too. Storage codecs such as H.265 reduce bitrate for the same picture, with consequences for the live path covered in why the storage codec breaks the live path.
Bitrate sets how fast footage arrives. Retention sets how long it stays — and that is where video recording storage cost is decided.
Retention: the lever on video recording storage cost
Holding cameras and bitrate constant, the stored volume is exactly proportional to retention. Double the days, double the bill. Here is one estate — 100 cameras at 2 Mbps — at list prices from Amazon S3 (Standard, $0.023 per GB-month) and Backblaze B2 (from $6.95 per TB-month):
Retention | Stored at steady state | S3 Standard / month | Backblaze B2 / month |
|---|---|---|---|
30 days | 64.8 TB | $1,490 | $450 |
90 days | 194.4 TB | $4,471 | $1,351 |
180 days | 388.8 TB | $8,942 | $2,702 |
365 days | 788.4 TB | $18,133 | $5,479 |
Two things stand out. The spread between providers is a constant factor of about three, while the spread between retention policies is a factor of twelve. And the jump from 90 to 365 days quadruples the bill — worth weighing against how often footage older than 90 days is actually reviewed in your own estate.
Retention is not only a cost decision. The GDPR's storage-limitation principle in Article 5(1)(e) says personal data should be kept "for no longer than necessary for the purposes for which the personal data are processed," and footage of identifiable people is personal data. The period that is cheapest and the period that is defensible usually point the same way: as short as the purpose allows.
Shorter retention does not automatically lower video recording storage cost, though. On some providers it saves nothing at all.
The minimum-duration trap
Several storage services charge for a minimum period no matter when you delete. Wasabi's minimum storage duration policy states that if objects are deleted early, "a Timed Deleted Storage charge equal to the storage charge for the remaining days will be applied," and uses 90 days as its worked example. Wasabi's pay-as-you-go price is $7.99 per TB-month — but on a 30-day camera retention policy, each byte is billed for 90 days.
Retention on a 90-day-minimum service | Effective price per TB-month stored | 100 cameras at 2 Mbps / month |
|---|---|---|
30 days | $23.97 (3× list) | $1,553 |
60 days | $11.99 (1.5× list) | $1,553 |
90 days | $7.99 (list) | $1,553 |
Every row costs the same. Below the minimum, shortening retention cuts the data you keep and none of the money you pay — and at 30 days the "cheap" provider costs more than S3 Standard. Amazon's colder tiers have the same property: S3 Glacier Instant Retrieval has a 90-day minimum and Glacier Deep Archive 180 days, per the S3 pricing page. Backblaze B2 states it has "no minimum storage duration fees."
The rule is simple: compare providers at your retention period, not at their list price. Effective price = list price × (the greater of your retention or their minimum) ÷ your retention.
Storage is the cost of keeping footage. Video recording storage cost has a second half: using it.
Retrieval and egress: the other half of the bill
Footage that is written and never read costs only storage. Footage that is reviewed, exported or streamed back to a control room costs egress too, and the providers price that very differently.
Provider | Outbound data transfer, as published |
|---|---|
Amazon S3 | $0.09 per GB to the internet for the first 10 TB a month — about $90 per TB pulled out |
Backblaze B2 | Free "up to 3x their average monthly storage," then $0.01 per GB |
Wasabi | Free while monthly egress is "less than or equal to your active storage volume" |
For a typical camera estate, where a small fraction of footage is ever reviewed, egress is minor. It stops being minor when recordings are routinely streamed back for review, or when an investigation pulls weeks of footage at once. Where footage is watched live, the bottleneck is usually elsewhere — watching 500 cameras is a decode problem, not a storage one.
Which leaves the practical question: how do you set retention so it is both cheap and safe?
Setting retention by class, not by estate
The expensive pattern is one retention period for everything, set to satisfy the most demanding camera. The cheaper and more defensible pattern is retention by class, with incidents lifted out.
- Default class, short period. This is footage kept in case something happens. Keep it for the shortest period your purpose and your rules allow.
- Sensitive classes, longer period. Cash desks, entrances or regulated areas may justify more. Give them their own class rather than raising the whole estate.
- Incidents, held indefinitely until released. When something happens, the relevant footage goes under a legal hold and leaves the retention schedule entirely. That is how evidentiary video recording survives a short default period.
- Tier by age where retrieval allows. Older footage can move to a colder, cheaper tier — as long as its retention outlasts that tier's minimum and retrieval latency is acceptable. How lifecycle tiering fits the pipeline is in what runs behind the recording pipeline at scale.
The same arithmetic shows why coverage decisions compound. Every camera you add raises the plateau permanently, which is worth weighing alongside the operator-to-camera ratio that decides whether anyone is watching them live.
The Bottom Line
For continuous recording, video recording storage cost is cameras × bitrate × retention days × price — and it plateaus once retention is reached. Of those terms, retention is the one a policy decision moves directly, and it moves the bill far more than the choice of provider.
Price providers at your actual retention period, because a 90-day minimum makes anything shorter cost the same. Then keep the default short and hold incidents separately.
What's Next
For how per-session recordings grow instead of plateauing, and why egress often outweighs storage, read the compliance trap in recording at scale. For keeping specific footage safely beyond a short default, see evidentiary video recording.
Frequently Asked Questions
How do I calculate video recording storage cost for 24/7 cameras?
Multiply cameras by average bitrate by seconds per day by retention days to get stored data, then multiply by the price per GB-month. One camera at 2 Mbps writes 21.6 GB a day, so 100 cameras on a 30-day policy store about 64.8 TB — roughly $1,490 a month on S3 Standard list pricing or $450 on Backblaze B2.
Does video storage cost keep growing every month?
Not for continuous recording with a fixed retention period. Once the oldest footage is deleted as new footage arrives, stored volume stops growing and the bill flattens. Per-session recordings kept for years are different: they grow until the retention period is reached.
What bitrate should I assume for a surveillance camera?
There is no standard figure; it varies with resolution, codec, frame rate and scene. Axis reports bitrate reductions from 25% to 90% or more for the same technology depending on scene. Use the average your recorders actually report, not the configured maximum.
Why doesn't shortening retention always reduce video recording storage cost?
Some storage services charge a minimum storage duration. On a 90-day-minimum service, footage deleted after 30 days is still billed for 90, so a 30-day and a 90-day policy cost the same. Compare providers at your actual retention period.
Is cold storage cheaper for surveillance footage?
Often, but check the tier's minimum duration and retrieval time. S3 Glacier Deep Archive has a 180-day minimum and is designed for rarely accessed data, so it suits long-held footage that is seldom reviewed, not a 30-day rolling buffer.
How long should CCTV footage be kept?
As long as the purpose requires and no longer — the GDPR storage-limitation principle points the same way as the cost. Sector rules can set specific periods. Use short default retention, longer classes where justified, and legal holds for incidents.
Can recordings be stored on our own infrastructure instead of a cloud bucket?
Yes, if the platform lets recording storage run on infrastructure you control. Samvyo works this way: it is based on SFU architecture and keeps recording processing and storage on your side, so the storage bill follows your own hardware or cloud contract and retention policy.
