Every YouTube app on a Smart TV or set-top box runs on Cobalt, Google’s deliberately minimal HTML5 engine built for devices where a full browser was never an option. Look at Settings → App Version and 25.lts is classic Cobalt; 27.lts or later is Chrobalt, the same product rebuilt as a Chromium embedder. Collabora spent over a year porting it to an RDK reference platform on Amlogic chipsets, and their write-up is a useful map of where that kind of work actually goes.

The architectural argument is simple: maintaining an independent rendering engine against a moving web has no completion state.

  • Cobalt shipped a web-standards subset optimized for media streaming, talking to hardware through an abstraction layer called Starboard
  • Keeping that engine current alongside the modern web “is a massive undertaking” — you are permanently behind
  • Chrobalt merges Cobalt’s code into Chromium: upstream web features and security updates below, Starboard and low-footprint media pipelines above
  • Desktop and Android reference ports already existed; device makers needed one they could copy

The Chrome swap is Google’s work. The year went into the unglamorous layer:

  • Decoupling and restructuring legacy Starboard recipes across RDK’s Yocto layers, modernizing toolchains for the newer C++ standard Chromium requires
  • Running Chromium’s unit tests alongside Starboard’s NPLB (“No Platform Left Behind”) conformance suite as the compatibility baseline
  • Passing the YouTube Test Suite — 4K/UHD playback, DRM, low-latency input
  • Fitting Chrobalt into RDK’s modular plugin architecture (RDK7): plugin lifecycle, memory sandboxing, unified audio/video pipeline
  • Making the Device Automation Bus trustworthy enough to drive a full YTS run reliably and reproducibly, extending the Rust dab-adapter Collabora co-authored
  • Upstreaming fixes back to Cobalt and Starboard

The transferable lesson is about where the durable interface lives. The engine is replaceable; the porting layer, the conformance suites, and the automation bus are what make a port repeatable across silicon vendors. Certification infrastructure, not the public API, is the real contract between an engine owner and the device makers who have to ship it.

Worth reading with a grain of salt: this is a services vendor’s case study with a capability pitch attached. The technical detail is real, but treat the framing as marketing and the checklist as the useful part — especially if you ship media UI on hardware where you cannot just install a browser.