Atmos is not a codec. Atmos is an object-based authoring workflow for media where, instead of discretely-mixed channels, you have audio objects with positional information. It is the job of the decoder (Dolby Digital Plus, or E-AC-3, Dolby True HD, or Meridian Lossless Packing with metadata, and AC-4, a new format extending some of the tools of HE AAC v1 and v2 with some new goodies) to decode these objects with the correct loudness and compression characteristics per the metadata that accompanies them. A renderer that understand the actual speaker configuration present in the listening environment, rather than the CEA-861 standard baseline configurations for typical surround, then mixes these objects.
In Dolby Digital Plus, which is the most common way to do this today, these objects are stored as a different substream type that goes along with the channelized audio, so the audio renderer in an Atmos-enabled Dolby Digital Plus has the job of dealing with and mixing into the speakers - typically, this means a 5.1 + 4 or 7.1 + 4 configuration, where the +4 implies four ceiling-mounted speakers. In AC-4, it is possible to have an entirely object-based render with no channel information so that the renderer bears the entire burden of channelizing, whereas in Dolby Digital Plus it is normally accompanied by a fixed channelized mix of anything from mono center only to 7.1.
Here is where the Dolby narrative breaks down: it is wholly impractical to have ceiling speakers for the vast majority of installations. About 30% of homes have surround of any kind, and less than a sixth of those have 7.1. Even ceiling-fired speakers that Dolby touts are very limited in their efficacy. The most shocking thing of all here is that upmixed content (i.e. content that uses an interpretation of a channelized mix to fill ceiling speakers) sounds better to most listeners because it is inherently louder than the native mix! People want to hear more coming out of their surround and ceiling speakers even though the guys in Hollywood mixing their movies have a more nuanced approach to this, so they're better off with an A/V receiver that supports upmixing with ceiling speakers and sending a non-Atmos mix to the A/V receiver.
Even if you virtualize surround to headphones - a thing that has been done since the Aureal 3D days back in the '90s and very cheap technology fundamentally as Realtek and the rest of the cheap guys have shown - it is still of limited value to have an Atmos mix because you can still upmix and virtualize the content in the same way, and most people can't A/B test this. And, of course, any game or native PC content that has linear PCM out can be channelized and output to HDMI without touching an Atmos encoder or decoder.
Dolby has some big challenges ahead of it. Its patents on AC-3, its bread and butter, run out next year. While E-AC-3 has taken up some of that slack, it is not the entirety of it. Audio coding efficiency is no longer a problem as of HE AAC v2 for stereo and HE AAC v1 for multichannel, which is why AC-4 was really just an add-on to these. Many countries, such as Korea, are adopting the competing MPEG-H standard because Dolby has stuck it to their largest conglomerates for so many years in terms of the licensing money they have paid. Their other technologies such as Dolby Vision and Contrast are being put to the wayside in favor of HDR10 and other standards. All the parlor tricks and announcements about Atmos are not worth the time and effort for most of us, especially in the PC world, when there are very good alternatives. All this because the companies that Dolby sells to are fundamentally ignorant about audio technology.
In short, this is much ado about almost nothing.