Chrome supports headless browsing that's exposed through a switch that can be used when starting the process. This enables opening pages without creating a browser window, thus a graphical environment is not required.
When used in conjuction with an appliance that connects to the remote debugging port, it enables interacton with the document, which is particularly useful for testing and CI automation, where a graphical environment is not necessary to get results or isn't available.
Historically, other's have succeeded at using Chrome as a headless browser by running the process in a hidden display.
Invoking Chrome directly is not the only available option for using it as a headless browser. The Embedder API also alows one to use Chrome directly from the application's process.