In the scenario you describe, it will likely be slower. This is because (as @Jaxidian points out) that the caching does not happen until the content is requested.
Without CDN: user requests content from data center where blobs live (there are 8 data center options today). The content is returned directly.
With CDN: user pulls from "closest" (closest in networking sense) data center (there are 24 CDN points of presence today). The local CDN cache is checked. If the data is not in the cache, the content is requested from the source data center and the cache is populated. Then the content is returned to the user. For first retrieval request, this obviously hurts latency.
However, with a blog, there may be some fairly static content (stylesheets, javascript, logos, other "blog-wide" content) common either across pages on a single user's blog - and some other content common to all users on the platform - that would be more valuable to serve up via CDN for repeated access.
Note that there is no way (today) with Windows Azure CDN to force any CDN node to load data (only external requests do this) and no way to bump an item from the CDN (only HTTP cache headers are used for this), so you need to be careful about how long you cache things for in the CDN (consider when a published blog post is modified).