The canvas element is great but not for the reasons described in this article. In all of those examples a web developer would be better off saving a static image. This is guaranteed to work in every web browser while the canvas element is not (yet). Also, why create these graphics programmatically on the client-side with each page load? If we are talking about the user customizing the graphics on the client side (graphics with different color schemes for example) then that makes some sense. Otherwise I think this article misses the mark. The canvas element is most useful when you need to generate image content dynamically in RESPONSE to USER INPUT.
Read step #2. The author is pitching the use of canvas as a rapid prototyping tool for common graphics requirements that come up in web design. There are a lot of interesting things you could do without having to do a lot of manual manipulation that would be required in a tool like Photoshop.
Programming in canvas is not automatic (as opposed to manual). For graphic designers I still say canvas offers marginal benefit.