Pricing

Priced against what it has to carry

There are no tiers on this page, because there is no meter. A deployment is quoted once, against your catalogue, your audience and the hardware you already own — and the calculator below shows what the rented alternative costs at your volume, using AWS's own published rates.

Three shapes

How the work is bought

Which one you want depends on whether you intend to operate this yourself, and that is a decision you can change later without changing platforms.

Most common

Build and hand over

Discovery, architecture, deployment, cutover, and the repositories handed to your team. Fixed scope, quoted once, with the applications published under your accounts.

Priced against catalogue, audience and existing hardware
One quote after discovery, not a running meter
Your team operates it afterwards

What this includes

Build, then run it for you

Managed operation

The same deployment, and we keep operating it — on call, patched, capacity planned. A monthly fee against a named scope rather than a per-viewer meter.

Monthly, against an agreed scope
No per-subscriber or per-gigabyte line
You can take it in-house at any point

What this includes

Before you commit

Consulting and review

An architecture review, a cost review of what you run today, or a second opinion on a platform decision. Days rather than months, and it ends in a document you own.

Day rate, scoped up front
Frequently ends in “do not rebuild this yet”
Credited against a deployment if one follows

What this includes

Build versus rent

What your traffic costs on somebody else's infrastructure

Move the sliders. The rates are AWS's published prices for this region, the arithmetic is on screen, and the result is allowed to come out against us.

Your numbers

Change anything. Nothing is sent anywhere, and nothing asks for an email address.

Hours watched per month250,000

Average delivered bitrate2.9 Mbps

The live service runs at 2.885 Mbps.

Catalogue held1,200 hours

New content encoded monthly60 hours

Top rendition

What that costs on rented infrastructure

$31,733

per month · $380,798 a year · 326 TB delivered

Delivery — CloudFront egress: $31,173
Encoding — MediaConvert output minutes: $490
Origin — S3 Standard: $70

Owned capacity replaces the first two lines with hardware bought once and a transit bill that does not move with the audience. Which is cheaper depends on how long you intend to keep operating — below roughly a terabyte a month renting wins, and this will show you that rather than hide it.

AWS published rates · ap-southeast-1 · read 2026-08-25 · Singapore egress price group · excludes compute, support and transfer into the origin

The full rate cards, the worked petabyte case and the sources behind every figure are on the economics page.

Three worked cases

Find the one that looks like you, then change the numbers

Each of these is a real set of slider positions rather than an illustration. Set the calculator above to the inputs shown and it returns the same figures — they are generated from its own rate card, not typed next to it. What every case shows is the rented side, because that is the side with a published price; ours is quoted against your deployment.

A regional broadcaster

One linear channel and a back catalogue, moving to streaming for the first time. The audience is national, the catalogue is modest, and most of the watching is on the channel rather than the archive.

Delivery is already the largest line and the catalogue has barely started growing. This is the shape where owning the edge pays for itself first, because the bill is almost all bytes.

Hours watched monthly: 120,000
Average bitrate: 2.5 Mbps
Catalogue held: 800 h
Encoded monthly: 40 h
Delivered: 135 TB

Rented, at these numbers

$13,838

per month · $166,051 a year

Delivery $13,471 · encoding $326 · origin $41

A mid-size telco

The service is bundled into data packs, so the audience arrives all at once and watches on phones. A larger catalogue, a much larger delivered volume, and most of it inside the operator's own network.

At this volume the rented bill is the business case on its own — and it is the case where an owned edge inside the operator's own network changes both the cost and the delivery quality, because the traffic never leaves.

Hours watched monthly: 1,200,000
Average bitrate: 3.2 Mbps
Catalogue held: 3,000 h
Encoded monthly: 120 h
Delivered: 1,728 TB

Rented, at these numbers

$141,748

per month · $1,700,980 a year

Delivery $140,575 · encoding $979 · origin $194

A migration off a rented platform

An operator already streaming, with a deep back catalogue and a subscriber base to move without losing. New content is slow; the archive is the asset.

Storage becomes a real line here rather than a rounding error, and it is the line that never falls, because an archive only grows. A full re-encode of that catalogue is a one-off cost on the rented side and a scheduling decision on the owned one.

Hours watched monthly: 600,000
Average bitrate: 2.9 Mbps
Catalogue held: 12,000 h
Encoded monthly: 30 h
Delivered: 783 TB

Rented, at these numbers

$71,319

per month · $855,825 a year

Delivery $70,369 · encoding $245 · origin $705

What none of these shows is our number, and that is deliberate rather than coy. The owned side resolves against five inputs that are specific to a deployment — delivered volume, catalogue size and growth, peak concurrency, encode hours, and the amortisation period your finance side will accept — and a band quoted before those are known is a band about somebody else’s service. What you can do here without talking to anyone is establish the size of the problem, and below roughly a terabyte a month the calculator will tell you to rent.

After launch

What actually recurs, once it is running

The question a CFO asks second, and the one most vendor pages skip. On an owned deployment the recurring cost is not a licence — it is power, transit, capacity and people.

Line	What it is	How it behaves as you grow
Transit or peering	The committed bandwidth your edge is bought against	Fixed inside the commit; steps when the commit is renegotiated
Power, rack and cooling	What the hardware costs to keep running	Flat until you add nodes
Hardware refresh	Amortisation of what was bought, and eventual replacement	Planned, not metered — and the input that moves a comparison most
Cloud, in the hybrid pattern	The application tier on AWS, if that is the shape you chose	Scales with the application, not with viewers watching
Payment gateway fees	Charged by the gateway to your merchant account	Per transaction, and never through us
Support and engineering	The agreement you sign with us	Fixed term, not per viewer and not per gigabyte
Store fees	Charged by Apple and Google on in-app purchase, where you use it	Per transaction, on your own developer account

No line on this table is per gigabyte delivered or per viewer watching. That is the whole of the argument on this page, stated as an invoice rather than as a claim.

What decides the quote

Six inputs, and what each one is worth

These are the questions discovery asks, in roughly the order they change the number.

What moves the number	Why	What reduces it
Catalogue size	Every hour has to be encoded once per rendition and then held forever	A shorter ladder, and re-encoding only what is actually watched
Concurrent viewers at peak	Peak decides edge capacity, and peak is a match night rather than an average Tuesday	Separating live from on-demand so one surge cannot size the whole estate
Existing hardware	Owned capacity that already exists changes the capital half completely	Using what you have, even where it is not what we would have specified
Rights and DRM	A studio agreement requires studio DRM, which is a licence server this platform does not run yet; independent content is served by AES-128, which it does	Matching the scheme to the agreement rather than to the brochure
Who operates it	A managed engagement carries people; a handover does not	Taking operation in-house once the runbooks are proven
Deployment model	On-premises is capital then flat; AWS is operational and moves with traffic	Placing each tier where its cost shape is cheapest, which is what hybrid is

Questions

What buyers ask about price

Why is there no price list?

Because it would be wrong for almost everyone who read it. The inputs that decide a deployment — catalogue size, peak concurrency, existing hardware, rights constraints — vary by more than an order of magnitude between two businesses that look similar from outside. A published figure would be a number to argue about rather than a number to plan against.

What does discovery cost?

It is scoped as a short consulting engagement and credited against the deployment if one follows. It ends in an architecture and an honest date, including — sometimes — the answer that this pattern does not fit you.

Is there a free trial?

No, and it would be dishonest to offer one. This is infrastructure deployed into your estate, not an account on a hosted platform. What exists instead is a live consumer service you can watch, measured numbers you can interrogate, and a discovery engagement that ends in a document whether or not you proceed.

Do you charge per subscriber, per gigabyte or per transcoding minute?

No. Those meters exist because a vendor is carrying infrastructure on your behalf. Here the infrastructure becomes yours, so the only recurring cost is the one you would pay anyway for power, transit and hardware — plus a managed fee if you choose that.

How does this compare to a rented platform at our size?

Put your own numbers into the calculator on this page. It uses AWS's published rates, shows every assumption, and will tell you if renting is cheaper at your volume — which below roughly a terabyte a month it usually is.

What happens to pricing if we grow?

Delivery and encoding stop tracking your success, which is the entire point. Growth costs hardware you plan for rather than a bill that arrives afterwards, and the step changes are visible months ahead in the capacity plan.

Send us your numbers and get a real one back

Catalogue size, peak concurrency, and what you run today. What comes back is an architecture and a quote, or an honest reason not to proceed.

Talk to us →See what a deployment includes →