The Livepeer Bet: Allocative Capacity

After successfully livestreaming hundreds of conferences and events on their alpha network, Livepeer is gearing up to support scaled users — hundreds and thousands of concurrent livestreams. It’s still a bit early, but the team is seeing promising results, and the plan is to go to market and mainnet this year.

If you’ve seen the numbers for existing decentralized applications (dapps) on mainnet, you’d be right to be skeptical about Livepeer achieving broader adoption. Very few non-gambling dapps have a substantial number of active users, and Web 3.0 still has a long way to go. Perhaps the biggest question for Livepeer is this: how can decentralization improve livestreaming, anyway? In other words, will anyone (besides crypto nerds) actually use this?

Delivering Before Web 3.0

According Doug Petkanics of Livepeer, cloud providers charge ~$3 per livestream per hour. With thousands of streams at any given time, that’s thousands of dollars per hour, which is prohibitive for streaming service startups.

The Livepeer team believes that they can deliver huge cost savings and increase reliability. Even if you don’t buy Web 3.0’s promise for widespread adoption of decentralized infrastructure, Livepeer’s value proposition is still worth some attention.


Case Study: encoding

When miners hash cryptocurrencies, they use the entire capacity of the GPU’s CUDA cores while their two video encoding ASICs sit idle. Livepeer’s software can harness idle video encoding power in the miners’ GPUs, and without disrupting their mining. The cost comparison? About $0.74 per day (electricity and bandwidth) for miners versus $144 for a cloud provider (encoding 2 HD videos concurrently).


This example may be particularly fortuitous, but regardless of how GPU architecture evolves, Livepeer still looks like it could be a winning bet.

The Bet

A bet on Livepeer is a bet on allocative capacity. This is the premise: idle compute capacity will always be cheaper than dedicated capacity. Machines provisioned to handle peak traffic (thousands of concurrent livestreams) are idle at hundreds of livestreams, whereas machines running in a functioning business model are often already paid for. Thus, dedicated capacity will always be more expensive to build, deploy and operate than to participate in what Livepeer is building: an open market place that lets anyone put idle capacity on the network.

At Figment, investing in the Livepeer network has turned out to be a lucrative bet. It’s a good time to be staking on Livepeer, particularly because we’re benefiting from rewards that are relatively high when compared with those of other proof of stake (PoS) networks.

Staking on Livepeer

When you stake, you have the right to participate and do incentivized work. You can also delegate to others who will do the work on your behalf, and then share the rewards. The easiest way to participate in the Livepeer network is by delegating LPT (Livepeer tokens) to an existing transcoder, such as Figment.

Here’s our delegation guide. Figment is one of the few transcoders offering a service-level agreement (SLA) that guarantees compensation for missed rewards arising from any failure in our in performance. If you’d like to learn more about Livepeer, Will Little’s January article explores Livepeer more broadly.

Hopefully you found this useful. Feedback is always welcome!