In the AVC world putting together MCU streams is very straight forward but compute intensive. The streams from each client are decoded, combined into the appropriate images, re-encoded and shipped out. As resolutions increase, this transcoding process becomes more compute intensive. In essence every video pixel gets touched by the hardware three times; once when it is decoded, once when it is combined with other images to form the final image and lastly when it is re-encoded.
In the SVC world, video streams are composed of several layers. SVC encoders compress the video into temporal streams (streams with different frame rates), spatial streams (streams with different resolutions) and quality (streams with different details). These streams are reassembled into a single high frame rate high resolution high quality video by the decoder. Working strictly within the parameters of the H.264 SVC specification, Vidyo developed a router architecture which does not rely on transcoding to provide continuous presence. Essentially what the VidyoRouter™ does is route the encoded packets so that the appropriate layers are delivered to each participant in a call. The router never decodes a single pixel. This allows the router to be constructed using a medium end server with no special acceleration hardware because there is no extremely compute intensive transcoding function.
It is the fundamental difference in how these streams are processed in the AVC vs. the standards compliant approach Vidyo was able to implement in the SVC world that accounts for the difference in MCU vs. VidyoRouter™ costs and power consumption.