I unfortunately can't agree with that sentiment. Composites fail in very complex ways, with part of a tested sample failing in tension, another portion in compression and a third portion of that same sample in torque (possibly other ways too, depending on layup etc).

To bolster my point, listen to the video you've linked. Long before the steel has started obviously yielding, you can hear some of the CF fibers failing. It'll take it, but your rated static strength is actually decreasing as those fibers in the downtube cross-section progressively fail. When failure in that video finally happens, it's indeed at a high value... But it was only tested to failure once. And therein lies the insidiousness.

You've taken a jump on your mtb several times, jump, jump, jump, not hearing the gentle pinging in your bike's headset area every time you land. One day you take the jump, as usual, but the headset and thus the forks & handlebars snap off. That's a problem.