Video Encoding API, Open source video player (video.js)
Turn this teardown into a decision-ready prompt for ChatGPT, Claude, or your agent.
Zencoder was a San Francisco-based video infrastructure company founded in 2007 by Jon Dahl, Steve Heffernan, and Brandon Arbini. It participated in Y Combinator's Winter 2010 batch and launched publicly in May 2010 with a cloud-based video encoding API — a service that let developers transcode video at scale without managing hardware, paying per minute of encoded content rather than per gigabyte. Alongside the core API, the team built Video.js, the first open-source HTML5 video player, which became one of the most widely adopted open-source video projects ever created.
Zencoder did not fail in the conventional sense. Brightcove acquired it in July 2012 for approximately $27.4–$30 million — roughly 15x revenue — a strong outcome for a 10-person team that had raised just over $2 million. The more instructive story is structural: the standalone cloud encoding API model was inherently capped. It was transactional by nature, vulnerable to platform commoditization, and lacked a natural path to the data and analytics integration that would define the next generation of video infrastructure.
The founders sold at precisely the right moment. Less than a year after the acquisition, Amazon launched Elastic Transcoder — a direct copy of Zencoder's API design and pricing model. The exit validated both the founders' timing instincts and the thesis they would later build Mux around: that encoding alone was never the endgame.
The Zencoder story is, by the founders' own description, "built on a string of failures." [1] Understanding those failures is essential to understanding why the eventual product worked.
In 2007, Jon Dahl, Steve Heffernan, and Brandon Arbini attempted to build a consumer video platform — essentially an "HD YouTube" — that never gained traction. [2] The specific name and funding details of that venture are not publicly documented, but the experience gave the founders deep familiarity with the pain of video encoding at scale: the hardware costs, the format fragmentation, the operational burden of keeping encoding pipelines running.
Their second attempt was a self-hosted encoding product — software a company could run on its own servers. It sold exactly one copy. The customer returned it. The signal was unambiguous: the market wanted a service, not software. Infrastructure buyers did not want to manage encoding pipelines any more than they wanted to manage their own email servers.
The third attempt came in November 2008, when the founders partnered with On2 Technologies — the codec company behind VP6 and VP8 — to build Flix Cloud, a high-capacity cloud encoding service running on Amazon EC2. [3] Flix Cloud was, in retrospect, a direct prototype of Zencoder: same infrastructure model, same target customer, same core value proposition. It was gaining traction when Google acquired On2 in February 2010 for approximately $124.6 million. [4] Google had no interest in running a commercial encoding service for third parties. Flix Cloud was shut down in November 2010.
The Google acquisition was both a near-death experience and an unexpected gift. When Google wound down Flix Cloud, it actively encouraged existing customers to migrate to Zencoder — providing the startup with a ready-made early customer base at the moment of public launch. [5]
The founders entered Y Combinator's Winter 2010 batch with this history behind them. During the program, Steve Heffernan retreated to a rented house in the Santa Cruz mountains and wrote the first version of Video.js — not as a strategic product decision, but as a practical solution to their own problem. "We were as much solving our own problem," Heffernan later said. "We need to embed videos ourselves and between HTML5 and Flash, there's no obvious way to do this. Why don't we solve it for everybody else too." [6]
Read the complete post-mortem, the rebuild playbook, and the exact reasons Zencoder is still worth studying now.