Not making a comparison in terms of contributions to science. But the same could be said about Newton and a host of other great (mostly) men of history.
One of the most useful emotional filters I've learned to drop is the predisposition against zealous enthusiasm. Automatically dismissing it is as reductive as immediately submitting to it.
I will push back a bit and say that the arrogance of current and past scientists and engineers mostly comes across in dealing with them personally and, for the most part, they try to put their egos to the side to better communicate their ideas. Wolfram puts his ego front and center in his more recent scientific communications (the CA collected papers didn't have this problem, as far as I can remember). It's been a while since I looked at ANKoS but, as I remember, the scientific ideas are intertwined with his editorializing. The editorializing gets in the way of understanding what actual fundamental scientific ideas he's trying to push forward.
Could this be a function of the eras' SNRs? When the principal mode of public and scientific discourse was curated print media and letters, one needed only so much pomp to decide you could deduce something grand about the world, and then upon having deduced it, convince your peers that you had in fact done so. In our modern era, to attract people and funding and attention, something more extreme may be needed. (Very, very loose hypothesis. It's probably just a personality thing.)
To me, Wolfram treats science and discovery as a kind of business opportunity while pretending it's for the greater good. Open source, peer review, transparency, etc. all give way to his "strong man" personality.
Not to fall afoul of Wittgenstein's ruler, I think Wolfram got right the principle of Turing Machine Equivalence. The idea that Turing machine completeness is the norm, rather than the exception, is something that I think people at large don't really understand or accept (and which I think is correct and profound). I will point out that there's work to be done here in trying to prove TME for other 1D cellular automata, not just rule 110, to further cement this point in these toy systems.
A fundamental thing that I think Wolfram got wrong is that something fundamentally new needs to supplant scientific discovery and mathematics. Often times discrete systems turn into continuous systems, and vice versa, so throwing out everything and starting over seems counterproductive. I'm in favor of bringing computational tools, both symbolic and numerical, in helping discovery but this is already happening. I also don't understand how this can be done if scientific discovery and tooling isn't democratized in the way that open source focuses on.
I think framing Wolfram solely in terms of his marketing for Mathematica and his ANKoS book is overly reductive but I also think it perverts his research, the way he presents his research and explains a lot of why people can come away with that feeling that he's a con artist.