Why Transcoding, Not Storage, Is the Real Cost of Open Video
Why transcoding, not storage, is the real cost of open video
Transcoding — not storage — is the line item that actually breaks the economics of open video platforms. A single hour of source footage encoded into the multiple resolutions and bitrates a modern player expects (360p through 4K, H.264 and AV1) can consume 10-50x more compute than the original file takes to store, and that cost repeats on centralized platforms every time a creator re-uploads or a format spec changes. Understanding where that cost actually lives — and how protocols like Livepeer route around it — explains why “just put video on IPFS” was never the whole story. It’s also the lens this publication covers video infrastructure through — see what this site is and isn’t for the full framing.
Key takeaways
- Storage is cheap and getting cheaper; transcoding is compute-bound and does not follow the same cost curve.
- Centralized platforms subsidize transcoding with ad revenue — a cost creators never see, but also never control.
- Decentralized transcoding networks (Livepeer being the most established) turn spare GPU capacity into a priced, competitive market instead of a platform-owned cost center.
- Open video’s real trade-off isn’t storage vs. platform lock-in — it’s who pays for transcoding, and whether that cost is transparent.
What actually happens when a video is “uploaded”
When a viewer hits play on a platform video, they’re rarely watching the exact file the creator uploaded. Adaptive bitrate streaming requires the source to be re-encoded into a ladder of resolutions and bitrates so the player can switch smoothly as network conditions change — a process that runs on GPU or specialized encoding hardware, not the cheap spinning disks or object storage used for the source file itself.
This is the part of the video pipeline platforms rarely make visible. A creator sees “upload” and “publish” as one step. Underneath, that upload triggers a transcoding job that has to run before the first viewer can watch — and that job’s cost scales with video length and resolution ladder depth, not with how many times the video gets watched afterward.
The cost centralized platforms hide
YouTube, TikTok, and similar platforms absorb transcoding cost entirely and recover it through advertising and, increasingly, subscription revenue shares. That subsidy is real, but it comes with a structural trade: the platform that pays for your transcoding also controls your distribution, your monetization terms, and — as many creators have learned during demonetization waves or account suspensions — your continued access to an audience you built on someone else’s infrastructure.
The subsidy is also invisible by design. A creator has no way to see what their upload actually cost to encode, no way to shop that cost to a cheaper provider, and no leverage if the platform decides to change the terms of the trade.
How decentralized transcoding changes the equation
Livepeer’s approach — and it’s the most mature example in this space — treats transcoding as a market rather than a platform cost center. Orchestrators run transcoding jobs on spare GPU capacity and get paid in the protocol’s token for completed work, verified through a probabilistic proof system rather than trusting a single centralized operator. The effect is a transcoding cost that’s priced by supply and demand across a network, instead of set unilaterally by whichever platform happens to host the video.
| Model | Who pays for transcoding | Who sets the price | Creator visibility into cost |
|---|---|---|---|
| Centralized platform (YouTube-style) | The platform, subsidized by ads | The platform, unilaterally | None |
| Self-hosted (FFmpeg on your own server) | The creator directly | Cloud provider’s compute pricing | Full, but requires ops expertise |
| Decentralized network (Livepeer-style) | The creator, via protocol fees | Market-set by orchestrator competition | Priced per job, visible on-chain |
This doesn’t mean decentralized transcoding is automatically cheaper — orchestrator competition can push prices down, but a thin network with few orchestrators can be more expensive than a hyperscale platform’s economies of scale. What it changes is legibility: the cost exists as a priced, observable transaction instead of a subsidy bundled into a platform relationship a creator can’t audit or exit.
What “creator ownership” actually costs in this model
Protocol-native video — where a creator’s content and audience relationship live in a wallet-addressable, portable format rather than a platform account — trades the invisible subsidy for a visible, ongoing cost. That’s not automatically a win. A creator who can’t cover transcoding fees on a decentralized network faces a real service interruption in a way that platform-hosted video, subsidized indefinitely by ad revenue, does not.
The honest framing is a trade-off, not a strict upgrade: platform hosting is free at the point of use but the platform owns the relationship; protocol-native hosting costs money directly but the creator owns the asset. Which side of that trade makes sense depends entirely on whether the value of portability and independence, for a given creator’s audience size and monetization model, exceeds the transcoding fees they’d otherwise never see.
FAQ
Why is transcoding more expensive than storage for video?
Storage cost scales with file size and is served by cheap, commodity infrastructure (object storage, spinning disks). Transcoding cost scales with video duration and resolution-ladder depth and requires GPU or dedicated encoding hardware, which is a fundamentally more expensive resource per unit of work.
Does IPFS solve the transcoding cost problem?
No — IPFS solves storage and content-addressing, not compute. A video pinned to IPFS still needs to be transcoded into a playable adaptive-bitrate ladder before a player can stream it smoothly, which is why decentralized video stacks pair IPFS-style storage with a separate transcoding layer like Livepeer rather than treating storage alone as the solution.
Is decentralized transcoding always cheaper than platform-hosted video?
Not necessarily. Centralized platforms benefit from hyperscale economies of scale that a thin decentralized orchestrator network may not match yet. The difference is less about guaranteed lower cost and more about that cost being priced, visible, and creator-controlled rather than an opaque subsidy.
What happens to a creator’s video if they can’t pay ongoing transcoding costs on a decentralized network?
Unlike ad-subsidized platform hosting, a decentralized transcoding network’s continued service is generally tied to continued payment for the work — a real economic obligation that platform-hosted video, at least while it remains ad-monetized, does not impose directly on the creator.