Skyfire

» see skyfire.com as it lived, on the wayback machine

~ the last words ~

in 2008 my phone was something like a nokia s40 with a clamshell, the browser was the s40 stub that could open wap pages on a good day. skyfire never showed up on that device. it surfaced for me later, in tech press posts around 2010, when the pitch was a windows mobile and symbian app that could load full youtube and full hulu on a phone by routing the page through a server farm in mountain view, rendering the flash there, and streaming the decoded video back as h.264. i read the writeups, never installed it. by the time it could matter to me the iphone 3gs i was eyeing had already started displaying video by way of html5 and the second skyfire app was on the app store.

dvc labs founded in 2007 by nitin bhandari and erik swenson, both ex-microsoft, with funding from matrix partners and trinity. the company renamed itself skyfire in april '08 alongside a public beta for windows mobile, then symbian, then a much bigger android client. the trick was thin-client mobile rendering. the heavy work happened on skyfire's servers and the phone got a lighter stream back. that was a real engineering pitch in '08 when phones had 200 mhz arm cores and the desktop web was full of flash.

the iphone version, released november 2010, was the one that got the press. apple did not let third-party browsers render their own html, so skyfire could not really be a browser on ios. it shipped as an app that took a url, detected flash video on the page, transcoded it server-side, played back the result inside a video element. it was a workaround dressed as a product. the app sold out of its $2.99 sku within hours of launch, the company had to shut the store entry and queue people up for new server capacity. that was the peak public moment.

opera software acquired skyfire on 15 february 2013 in a deal valued around $155 million, mostly stock. what opera wanted was the server-side video-optimization pipeline, which fit the opera mini and opera max compression playbook. the consumer skyfire browser kept getting builds for about another year, then the android and iphone apps were pulled from their stores around may 2014. by february 2015 the skyfire brand itself was gone, the team had become opera's network solutions division.

~ the record ~

~ what it actually did ~

the trick in 2010 was that mobile safari on the iphone could not play flash video. youtube already had an html5 fallback, almost nobody else did. hulu, espn, the bbc iplayer, most network tv catchup sites, all of them were flash. skyfire's server farm took the requested page, ran flash inside a headless renderer, captured the video element, transcoded it to h.264 and streamed it back to the phone wrapped in a video tag the native player could handle. it worked for maybe 60 percent of the flash video out there. the rest hit drm walls or interactive flash that did not survive transcoding.

on android the model was simpler. android shipped flash player mobile as an official adobe build until late 2012, so skyfire could either use the local flash runtime or fall back to its server pipeline. the value-add there was bandwidth compression more than codec translation. that pivot toward compression was already what opera mini did, which is why the acquisition made sense to opera on paper.

~ the actual end ~

two things killed the consumer brand. one was that adobe announced the end of flash player mobile on 9 november 2011 and shipped the final mobile build a year later, which removed the only reason an android user installed skyfire. the other was that the major flash video sites began shipping html5 video players of their own through 2011 and 2012, so the workaround on ios stopped being necessary site by site. by the time opera closed the deal in february '13 the consumer product was already shrinking.

opera kept the engineers and the data center pipeline. they folded the transcoding work into opera max, the data-saver consumer app for android that ran from 2014 until opera retired it in 2018. the skyfire web browser itself disappeared from the app stores quietly during may 2014. there was no announcement, no migration, no successor product wearing the skyfire badge. the company had become a tech-acquisition footnote inside opera.

~ the long tail ~

the server-side rendering pattern skyfire pioneered for mobile flash came back a decade later in a different form, as cloud-rendered browsers like puffin and as the server-side javascript work that powers cloudflare workers preview and replit's webview. the lineage is real even if nobody credits skyfire for it. opera's own cloud-rendering work for opera mini still uses the same compression and proxy posture skyfire used for flash video, which is how a 2008 startup ends up still running inside a 2026 browser nobody calls skyfire anymore.

the people who remember skyfire mostly remember the iphone launch in november 2010. a $2.99 paid app that sold out its server capacity in the first day, a press cycle about the browser that finally let iphone watch hulu, then a slow fade as the rest of the web stopped needing the trick. that is roughly the dignity owed to a workaround startup that did its workaround better than anyone else and then watched the workaround stop being needed.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »