A neural network with a hidden layer has universality: given enough hidden units, it can approximate any function. This is a frequently quoted – and even more frequently, misunderstood and applied – theorem. It’s true, essentially, because the hidden layer can be used as a lookup table.

We show that BigBird is a universal approximator of sequence functions and is Turing complete,

mixing the turing complete of javascript with the markup of HTML eliminates the readability of JSX so that it is actually harder to parse than a solution like hyperscript

A program somewhat controlling it's own flow. That's the key feature that makes a programming language Turing complete.

Here is how you could emulate an if-else-block:

A programming language is Turing complete if you can implement any possible algorithm with it.

Because templates are not Turing-complete, and if they were, they would require vDOM.