G-cluster

» see gcluster.jp as it lived, on the wayback machine

~ the obituary ~

Erik Piehl founded G-cluster in Finland in 2000. The proposition was the one every cloud gaming company has made since: the game runs on somebody else's hardware and your button presses travel to it over the network. What made G-cluster unusual is that it was doing this in the era of the IPTV set-top box, so the screen it aimed at was a television and the pipe it used was whatever the phone company had already run into the house.

That made the customer a carrier rather than a player. The first commercial rollout was for CYTA on the Cypriot IPTV network in 2005. HD support arrived in March 2008 on an Amino box. In November 2010, after a long beta, SFR launched a commercial service in France on G-cluster technology, and the company signed content deals with Electronic Arts and Ubisoft among others. Between 2010 and 2016 the money came from telecoms, with SFR and NTT among the investors. SoftBank had owned the company through a subsidiary since 2004.

None of that produced a business. The global company went into bankruptcy, and in July 2016 G-cluster was bought by Broadmedia Corporation, a Japanese operator that had been a partner for years. Broadmedia had already launched a dedicated G-cluster game machine in Japan in 2013 alongside Hikari TV support, which it marketed as the country's first cloud gaming service, and the brand continued there on Sharp AQUOS sets and LG smart TVs.

On 8 August 2024 Broadmedia published the notice. New registrations and all game sales would end at noon on 30 August 2024, and the whole service, television and app versions together, would end at noon on 28 February 2025. The reason given was that it had become difficult to keep operating at the quality the service required. Anyone who had already bought a game kept playing it until that last noon.

~ the case file ~

~ the carrier was the customer ~

The thing that kept G-cluster alive for twenty-five years is the same thing that kept it small. Selling to a phone company means one signature buys you a whole country, and it also means you never speak to a player. The service appeared in a Cypriot living room as a menu item on the IPTV box next to the video rental, with no brand of its own worth remembering and no reason for anybody to go looking for it.

OnLive and Gaikai went the other way in the same years, selling direct and spending enormously on marketing, and both died first. G-cluster outlived them by more than a decade while never selling to a player at all.

~ the six-month goodbye ~

The gap between the two 2024 dates is the part worth noticing. Sales stopped on 30 August and the service ran on until the end of February, which means Broadmedia operated a games platform for six months knowing nobody new could join it. Japanese press covering the announcement put the numbers plainly: ten years of service, twenty years of technology.

~ what it had right too early ~

Every assumption in the G-cluster design was correct and about fifteen years premature. Put the expensive hardware in a datacenter and send the frames to a cheap box by the television. Charge monthly instead of selling a console. Google announced Stadia in 2019 using that exact argument, ran it for three years, and refunded everybody. G-cluster had been shipping it commercially since 2005.

Turkey never had the carrier side of this. The TTNet era ran on ADSL where the upstream was an afterthought, and the set-top boxes that did arrive were sold to show football rather than to run Ubisoft titles. Cloud gaming reached here as a browser tab much later, which is a strange way to meet a technology that spent two decades being delivered as part of a television subscription somewhere else.

@}-->--

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

« previous  ·  random  ·  next »