As a number of people have said, the article applies the triangle to a process for which it was never intended. The triangle is about research and development processes, not about the manufacturing process for a known product with a working production line.
So the cost measurement is the wrong one: for a laptop (which is the example used in the article) the cost that is involved in the triangle is the cost of designing the product, working out how to manufacture it, and setting up the production line. It's already been pointed out pretty clearly that "fast" doesn't refer to the end product but to the development, validation, and mandufacturing setup process. The cost of the final product and its speed of operation are both part of "good", not the "cheap" and "fast" of the triangle.
The cause of the triangle is quite straightforward. There are various ways of doing research, development, and manufacturing setup. Some of these ways are "right" in the sense that they work well. Othere are wrong in that they don't work well, indeed don't really work at all. For any given project with a given set of aims, doing it right is essential. If that implies unacceptable timescale or unacceptable development cost or both, there are two options: (i) scrap the project and (ii) relax the requirements on the end product. Doing it wrong in an attempt to reduce costs and/or timescale is a fools game, the effect will be to increase both timescale and cost (or lead to the project being scrapped after a lot of effort has been wasted on it). Of course it may be possible to make trades between cost and timescale, but this tends to be more difficult than people expect (although people know that nine-fold bigamy doesn't deliver the 1-month baby they don't believe that that applies to their required miracle); more than 40 years ago the famous "you want it when?" poster was all over the place and clearly indicated that most of the technical community actually understood that management tred to ignore reality and deny the alidity of the triangle, so it's shocking to see today an article that gets it so wrong.