Your browser is not supported. Please update it.

13 January 2022

Livepeer goes silent on scene classification, as another $20m rolls in

Less than half a year since closing a $20 million Series B funding round, distributed transcoding platform Livepeer has raised another $20 million in Series B-1 financing. This rapid cash injection shows that investors are waking up to the idea of web 3.0 – with Livepeer’s total funding ballooning by 500% to $48 million in a matter of months.

In retrospect, Faultline’s interview with Livepeer CEO Doug Petkanics in July 2021 was perhaps ill-timed. Had it taken place only a week later, we would have landed a first-look insight into the company’s first $20 million spending plans, some two years on from its $8 million Series A round. While Petkanics successfully kept that investment under wraps, what we did report exclusively at the time is that Livepeer was on track to launch a pilot for scene classification technology – one of five future video processing opportunities identified by the US start-up.

Petkanics has not responded to our request for comment, while we haven’t seen or heard anything about Livepeer’s scene classification angle, which we were told would relate to adult content and soccer matches. The pitch is that streaming platforms dreaming of scale face challenges when identifying stream classifications in real-time, such as rights-infringing content. Livepeer claims companies can end up burdened with huge bills from human moderation when classifying scenes without an AI-based approach.

Adding additional features to the Livepeer portfolio such as scene classification is how the company plans to ultimately grow its customer base, by winning over clients that might be tentative about joining the Livepeer network – those with concerns about leaving behind the comfort of traditional transcoding partners. Livepeer’s four other future video processing opportunities are object recognition, song-title detection, video fingerprinting, and video-stack expansion. We were hoping to hear from Petkanics about how the latest financing round changes this roadmap, if at all, but we’ll have to save that conversation for another time.

In the meantime, Livepeer says the latest funds will help expand functionality for broadcasters, while positioning itself as the leading blockchain-enabled video infrastructure for future web 3.0-enabled applications to be built upon. As for which functions Livepeer plans to expand for broadcasters, it doesn’t say, but presumably the immediate priority is expanding the Livepeer transcoding network itself beyond adaptive bitrate transcoding, while the five future video processing opportunities are added extras further down the line. We see these more as nice-to-haves than must-haves.

For readers not already familiar with Livepeer, the firm has been making early waves with its decentralized approach to ABR video transcoding, by paying out in cryptocurrency to anyone who deploys a Livepeer node on their compute resources – either CPU, GPU, or bandwidth – to transcode and distribute video at reduced costs.

Those costs can be significant. Livepeer states that live video transcoding today costs around $3 per stream per hour for a cloud service such as AWS, easily racking up bills of $4500 per month for one media server, and up to $1500 per month before even factoring in bandwidth for a CDN. Livepeer’s core business is about increasing the reliability of video streaming while reducing costs associated with it by up to 50x.

This segues us nicely to the time of Livepeer’s Series B in July, when the company said funds would be put towards scaling teams and expanding use cases of its core technology. One way to achieve that is via acquisition, and Livepeer did just that in October with MistServer so that it could expand into distributed CDN too.

MistServer’s contribution to Livepeer’s roadmap is what the pair claim is the industry-leading media server, which effectively manages incoming video streams and ensures that they are outputted to the correct destination over the network. Conventionally, this would mean firing this video off to a transcoding provider, who would then handle the live compression of the input feed, who would then hand that transcoded video off to a CDN vendor (or run your own using MistServer), which can pipe those packets through to the viewer.

Now, however, Livepeer’s transcoding offering will be integrated closely with MistServer, which should place it front and center for MistServer users. The pitch here is that when these users see the difference in price between Livepeer and the alternatives, their choice will be a no-brainer.

The term web 3.0 is cropping up a lot. It describes a future where the centralized cloud platforms of our current web 2.0 era have been fully decentralized. Livepeer is already headed down that road, delivering decentralized transcoding among hundreds to thousands of third-parties, instead of relying on today’s handful of major transcoding providers.