Chromecast and How Google Cracked the Code of Video
When Google launched Chromecast I assumed it was based on the open source version of Airplay. Seemed logical as a bunch of people are working on standardized cross-platform approaches to Airplay, and the Wifi chip supports these variants. But Chromecast is not using mirroring or Miracast or Wifi Direct. As others have reported, all data to your TV is streamed directly to Chromecast from the cloud.
To really appreciate what this means you need to look at the diagram in this post and showcased on Google’s dev portal. It explicitly shows how Google approached the channelization of wireless video that is coming to your PC, and how they separated the control plane (the second screen device controlling the video) from the data plane (the actual video coming down from the cloud to Chromecast via WiFi).
Out of band control and data is non-trivial and there are a lot of pieces at work here: firmware & driver level (embedded in the SoC—more on this below), OS layer, app layer, etc.
Separating data and control is commonly done in wireless networks, but often both are sent through the same pipe. There are always tradeoffs between user experience, latency, and resources because control data can’t be dropped—you’ll lose things like billing info, channel number, connection status, etc… whereas it’s actually okay for some data to be dropped and lost (e.g. a pixel in a frame of a video—no one will see it or miss it).
After making some calls to friends in the valley I’ve found out more about Chromecast’s internals—the chip inside is the Marvell dual-core ARM7-based DE3005. Very similar to this one, which was in the previous Google TV. Google pays at most $10 for this chip. Because this SoC includes Wi-Fi, the GPU, and the dual core application processor itself, there aren’t many other expensive components on the board.
Believe it or not, for $35 resale Google will be making money on Chromecast. In fact the total BOM of Chromecast is likely less than $20!
This is because the major difference in the bill of materials between Chromecast and the Apple TV is the 8GB of internal Flash memory on the Apple TV, which is very expensive. Teardowns will reveal how much RAM Chromecast has, but it can likely get away with 512MB, just enough to buffer video, not store it for retrieval—again there is no Flash.
Not chasing the dream of being the hub of the TV (a la set top box and every previous Google attempt) is possibly a very smart move on their part. It’s Google’s most compelling entrée yet to the living room, because it demonstrates firsthand how their service and device ecosystem serve as satellites and controls for the TV.
By separating data and control planes, Google is essentially handing off the control and delivery of the content to application and services up the stack. And they couldn’t do that without control over distribution.
Google’s last few years of decisions across services and apps have given power to ordinary users (you) who want to stream video between screens, and taken power away from the central hub or set top box. However, this only works if Google can distribute its apps and services across devices and platforms.
The biggest example of how this strategy is coming together for Google is the iOS YouTube app, which could very well end up being the most used “client” for Chromecast (iPhones, iPod Touches and iPads number in the tens of millions across US living rooms).
The genius yet subtle win here for Google dates to when they unbundled Youtube from being baked into iOS and made it a separate app. If it was still built in, they would NOT have been able to add a “send to Chromecast” button and iOS devices would not be able to natively play YouTube via Chromecast, which they’ll obviously be able to do when Google releases a new rev of the app.
This fact that Google now controls YouTube on the iPhone (remember it was only about a year ago when YouTube was a “stock” app and Apple limited distribution apart from yearly iOS revisions) gives them immense power. It also makes the proliferation of the Chrome browser play smart. Extends playback from PCs overnight.
All of this points to Chromecast as the most elegant solution yet for “video redistribution” among screens in the home. The previous GoogleTV stuff was just a big joke. This is the right way to do this. Google can start steamrolling a lot of folks like Roku quickly if they get traction with this device.
Indeed as a wedge in to the living room, there is little doubt to me that we’re looking at the first major winner from Google. A lot of Chromecast’s unique tweaks—to decouple data and control in separate channels and enable the user to queue and buffer videos while not encumbering the 2nd screen device from doing other things could very well be part of the elegance Steve Jobs was referencing when he claimed he cracked the code for video. But Apple isn’t standing still either, and it’s going to be a very fun and competitive next push into television and video for us all to sit back and watch.