Video delivery cost per viewer: pricing a stream properly

Quick answer

One hour of 1080p video at a 5 Mbps average bitrate transfers about 2.25 GB. Delivered through CloudFront in North America at 0.085 dollars per GB, that is roughly 0.19 dollars per viewer-hour. At 720p and 3 Mbps it is about 1.35 GB and 0.11 dollars. At 4K and 15 Mbps it is 6.75 GB and about 0.57 dollars. For 100,000 viewer-hours per month of 1080p, delivery costs about 19,000 dollars at list price, which is why bitrate ladder tuning and modern codecs are the main levers: AV1 at equivalent quality cuts bitrate roughly 30 to 50 percent against H.264.

Video is different from every other workload in one respect: the cost is almost entirely bandwidth. The transcoding is a one-time charge amortized over every view. The storage is small next to the delivery. What you pay is the per-GB meter, multiplied by bitrate, multiplied by watch time, multiplied by audience. Getting this number right before launch is the difference between a viable product and a rounding error that eats the margin.

Bytes per viewer-hour

Quality	Avg bitrate	GB per hour	Cost per hour at 0.085/GB
360p	0.8 Mbps	0.36 GB	0.031 USD
720p	3 Mbps	1.35 GB	0.115 USD
1080p	5 Mbps	2.25 GB	0.191 USD
1440p	9 Mbps	4.05 GB	0.344 USD
4K	15 Mbps	6.75 GB	0.574 USD

The conversion is simple: bitrate in Mbps times 3,600 seconds divided by 8 divided by 1,000 gives GB per hour. A 5 Mbps stream is 2.25 GB per hour. Once you have that, everything else is multiplication.

Scaling to a real audience

Suppose a service delivers 100,000 viewer-hours per month, mostly at 1080p. That is 225 TB of egress. Through CloudFront at the North America and Europe list rate of 0.085 dollars per GB, and after the 1 TB free tier, the bill is roughly 19,000 dollars per month. Serving the same traffic directly from S3 at 0.09 dollars per GB would cost about 20,400 dollars, and without the CDN you also lose caching, so origin load and latency both worsen.

Geography matters enormously. CloudFront charges about 0.085 dollars per GB in North America and Europe, around 0.12 dollars in most of Asia Pacific, and around 0.11 dollars in South America. An audience that is 40 percent Asia Pacific raises the blended rate to roughly 0.099 dollars per GB, adding about 3,100 dollars per month at that volume. Price classes that restrict delivery to cheaper edge groups reduce the rate at the cost of latency for distant viewers.

Codec choice is the biggest single lever

Bitrate is not fixed; it is a function of codec efficiency and encoder tuning. Against H.264 at the same perceptual quality, HEVC typically saves 25 to 35 percent of bitrate and AV1 saves 30 to 50 percent. Moving a 1080p ladder from H.264 at 5 Mbps to AV1 at 3 Mbps takes the per-viewer-hour cost from 0.191 dollars to 0.115 dollars, a 40 percent cut. On 225 TB per month that is about 7,600 dollars saved.

The cost of doing so is encoding. AV1 encoding is substantially more expensive in CPU than H.264, though hardware encoders have closed much of the gap. Because encoding is paid once per asset and delivery is paid per view, the break-even favors the better codec heavily for popular content and barely at all for long-tail assets watched a handful of times. Encode the top 10 percent of your catalog in AV1 and leave the tail in H.264.

Per-title and per-scene encoding

A fixed bitrate ladder wastes bytes on simple content. An animated explainer with flat colors and little motion can look identical at 2 Mbps where a sports clip needs 8 Mbps. Per-title encoding analyzes each asset and assigns a ladder suited to its complexity, commonly saving 20 to 30 percent of average bitrate across a mixed catalog with no visible quality change. On the 225 TB example that is around 4,300 dollars per month.

Adaptive bitrate and the waste of over-delivery

ABR players pick the highest rendition the connection allows, which is not always the rendition the screen needs. A 1080p stream delivered to a phone at 5 inches is bytes spent on pixels nobody can resolve. Capping the maximum rendition by device class, for example 720p on phones, typically cuts mobile delivery volume by 40 percent. Similarly, aggressive pre-buffering wastes bytes for viewers who abandon in the first 30 seconds, and abandonment rates on short-form catalogs can exceed 30 percent. Tune the initial buffer to a few seconds and let ABR fill in.

Cache hit ratio decides how much origin traffic you pay

Origin fetches from S3 into CloudFront are free, so the cache hit ratio does not change egress cost directly, but it changes origin compute, origin request fees, and latency. At 0.0004 dollars per 1,000 GET requests, a 225 TB catalog served in 2 MB segments is about 112 million segment requests; at a 95 percent hit ratio the origin sees 5.6 million of them, costing a trivial 2.24 dollars. The real value of a high hit ratio is covered in CDN cache hit ratio economics.

Commit and negotiate

List pricing is the wrong benchmark at video scale. CDN providers offer committed-volume pricing that commonly lands between 0.01 and 0.04 dollars per GB at hundreds of terabytes per month, a fraction of list. Any serious video business should be modeling against a negotiated rate, and should know its cost per viewer-hour precisely enough to price the subscription around it. Price your delivery infrastructure against the resource catalog before you launch.

FAQ

How much data does an hour of video use?

Multiply bitrate in Mbps by 3,600 seconds, divide by 8, then by 1,000 to get GB per hour. A 720p stream at 3 Mbps uses 1.35 GB per hour, 1080p at 5 Mbps uses 2.25 GB, and 4K at 15 Mbps uses 6.75 GB. At a CDN rate of 0.085 dollars per GB those cost about 0.115, 0.191, and 0.574 dollars per viewer-hour respectively.

What does video delivery cost at scale?

100,000 viewer-hours per month at 1080p and 5 Mbps is 225 TB of egress, costing roughly 19,000 dollars per month at the CloudFront North America and Europe list rate of 0.085 dollars per GB. Geography shifts this: an audience 40 percent in Asia Pacific, where rates run around 0.12 dollars per GB, raises the blended rate to about 0.099 and adds roughly 3,100 dollars.

Does codec choice meaningfully reduce cost?

It is the largest single lever. At equivalent perceptual quality, HEVC saves 25 to 35 percent of bitrate against H.264 and AV1 saves 30 to 50 percent. Moving a 1080p ladder from 5 Mbps H.264 to 3 Mbps AV1 cuts per-viewer-hour cost from 0.191 to 0.115 dollars. Since encoding is paid once and delivery per view, the better codec pays off most on popular content.

What is per-title encoding worth?

A fixed bitrate ladder overspends on simple content. Animated material with flat colors can look identical at 2 Mbps where sports needs 8 Mbps. Per-title encoding assigns a ladder suited to each asset's complexity and commonly saves 20 to 30 percent of average bitrate across a mixed catalog with no visible quality loss, which on 225 TB per month is around 4,300 dollars.

Should I cap video quality on mobile devices?

Usually yes. Adaptive bitrate players select the highest rendition the connection allows, not the one the screen needs, so a 1080p stream on a five-inch phone spends bytes on pixels nobody can resolve. Capping the maximum rendition at 720p for phones typically cuts mobile delivery volume by about 40 percent with no perceptible quality difference.

How does C3X help with video delivery cost?

C3X prices delivery infrastructure from Terraform before deployment, including CDN distributions, origins, storage, and transcoding resources. Because video economics are dominated by per-GB rates multiplied by volume, having the priced plan at design time is what lets you compare a codec, ladder, or price-class decision against a concrete monthly number.

What to do next

Model your delivery cost before your first stream. C3X prices Terraform against a live resource catalog. Start with the quickstart.