OLE embedding is the closest you can get. Writing native viewer component for these file formats is overkill, it's just like writing another LibreOffice or WPS Office or StarOffice or any other office suites claiming for compatibility with MS Office (and I believe you know the compatibility sutation with those suites I've mentioned).
I don't think that it's overkill. We already have fpvectorial which provides the basic infrastructure for the tree of objects contained in the slides. What only is needed is a reader for the pptx format. I opened one: like Excel's xlsx or Word's docx this is a renamed zip file of a series of xml files; the slides, for example, are found in folder pptx/slides as files slide1.xml, slide2.xml, etc. I think once the details of fpvectorial are understood it is relatively easy to read the basic information contained, of course, without the transition and effect gimmicks. All we need is a motivated developer who dives into the details...