These microdisplays and microlenses and holographic waveguides are complicated. Snap seem to use technology from WaveOptics it acquired, and it require entire glass(not just display area) to be micromachined like silicon chips, of this size. The microdisplay is LCoS based, which needs to be front lit and front viewed through a tiny prism as well. And they would all have to go together with micro scale precision without defects.
You’re making it sounds like the margins on this are probably not great. Do you know if it’s likely that these aren’t meant to be directly profitable, or is it possible to get all of this technology together for around $1000–$1500 USD?
I don't know, I just know that these components used are both individually and collectively expensive to make. But a hypothetical $1500 per unit cost with delta between it and $2200 covering initial costs do sound somewhat plausible to me as an enthusiastic layperson.
The most costly part would be the transparent waveguide plates. They look like thin sheets of glass, but also virtually work as series of lenses and mirrors against light entering a small designated spot. WaveOptics[1] that reportedly supply that device seem to build them by stacking three plates for RGB made with electron beam lithography and nanoimprinting, which are both relatively slow, low volume semiconductor processes. The PR reel shows 10 monochrome waveguide plate patterns on one 200 or 300mm wafer, so theoretical wafer to RGB stack ratio is only 1:3. Nowhere in product pages of both Snap and WaveOptics says the wafers are glass, so I assume it's something else.
R&D costs are also not trivial considering that this isn't first gen product, but 5th or 6th gen. They'd have to be recoup costs of those devices as well, so I wouldn't be surprised if this was technically sold at small losses or at cost from accounting perspective.