Is the King's Indian Defence refuted by stockfish?

No, by a big shot. Whoever claimed this, is illiterate.

Not really, 7...Nc6 is still 100% relevant in ICCF.

The bayonet should be answered either by 9.b4 a5, or by 9.b4 Nh5 10.Re1 f5 11.a4!? a5!, when Black is {insert 20 moves of anaysis here} just OK. 

Nefedov's 9.b4 Ne8 might not be a good idea, but this does not mean that the KID is "refuted".

And also add

- 7.0-0 exd4!? 8.Nxd4 Re8 9.f3 Nc6! (which I believe is just equal, although not ambitious) and

- 7.0-0 h6!? which is a new idea, with a very sound positional basis. In the main line, Stockfish after proper analysis claims a huge advantage to white, of the scale +0.15 or so.

The Kramnik v Kasparov game was from over twenty years, there were many Loek van Wely vs Radjabov games in the 9...Nh5 afterwards. It is really weird logic that 9...Ne8 was an attempt at improvement\soundness, it is an extreme line completely leaving queen-side to fate. Somebody playing 9...Ne8 isn't looking for soundness. 

                              Personally play 6...Nb-d7 because there isn't so many forcing lines.

#26
"there were many Loek van Wely vs Radjabov games in the 9...Nh5 afterwards"
++ You mean this game?
https://www.chessgames.com/perl/chessgame?gid=1531189 
Not very encouraging for white.
Radjabov also gave up on the King's Indian Defence.

This game was played recently, between Stockfish15 and LCZero:

As you can see, Stockfish 15 (+NNUE) was unable to crack the King's Indian Defense.

I believe the KID is likely a draw with perfect play from both sides. But in OTB play, black sometimes has to make harder practical decisions.

That doesn't mean it's refuted, though - it just means that it's sometimes sharp and treacherous. The same can be said of the Sicilian.

Actually, and contrary to what a patzer thinks, this Van Wely- Radjabov game is still State Of Art, and was repeated lots of times.

Radjabov's mistake is the dubious 21...gxh4. After 21...Nxe6 the game is completely equal.

I have deviated at move 20 (20.c5), but it this isn't a new move, nor it changes the overall evaluation of equality.

The opponent blundered at move 29 (I think on purpose, he just wanted to abandon the tournament without picking a suspend), while this very position has been played at no less than one dozen and a half other CC games (of course 29...Bf8 was played instead), and all of them were drawn.

I see you insist pulling evaluations out of the butt, without even bothering to do a tiny bit or research. That's great, you are well on your way to become the best entertainer of this forum.

Looks very much like one of my games as white, except that I don't play the bayonet, because I don't know it. I almost always play Ne1. I find that when I put my N on e6 like that, usually forcing the opponent to take it, in practice it works out very well for white. I love playing against the KID but black only loses if black makes a losing mistake.

Beginners see an engine evaluation on move, I don't know, 8, and think it means something.

The only real evaluations are win and draw... and so when you explore openings more deeply they'll either tends toward win or 0.00

By exploring deeply I don't just mean letting it reach a high depth, I mean putting the moves on the board... sometimes for dozens of moves.

Nearly ALL openings tends towards 0.00 with proper analysis -- but this means very little to human play.

Therefore, chess is a draw with best play.

Well, or it's just that high level players (engines) that are equally skilled will draw against each other in spite of "high" evaluations that occur due to incomplete methods of evaluating early positions.

I like different arguments for chess being a draw such as comparing the drawing margin of endgames to the advantage of the first move, and also noting common elements among zugzwang positions and then noting the opening position has none of these elements.

who cares about some deep theory and stockfish opinions.. you really doesn't need to play a good opening to be good in chess..
and KID is even one of the better openings what you can play as black I believe..

It helps a lot if you know your openings. Saves a lot of time and energy working out openings, so opening knowledge wins you many games.

Yes there are dozens of good arguments that chess is a draw, plus all the evidence one needs.

knowing openings doesn't mean playing a good one.. you can play opening well even when stockfish doesn't like it.. and also if your theory is twenty moves deep and your oponent will play different first move I don't know much about the time you have saved.. 

Although the idea that black can possibly force a zugzwang from the opening position, that wins, has to count as desperation? I mean as a desperate notion, where all else fails.

Or, in fact, when it's clear that every pair of moves equalises the position more, how come all of a sudden that trend can reverse? It's like the cosmological idea of the eternally expanding and contracting universe. There's no possible cause that could make it stop expanding and start contracting. There's no cause that can inevitably reverse the trend away from equalisation in a chess game. We know that, because openings and middlegames can each be of very different durations. You can have an opening of 20 moves, followed by a 50 move middle game and, presumably, an 80 move ending. Sure, it's an intuitive or inductive thing but people who imagine that all mankind's knowledge is deduced are very wrong.

That's the whole point of knowing openings. If you open 1. d4 and get a King's Indian, and you had to work it out from first principles, the first 10 moves could take half an hour to an hour. Yet in an over the board game with long time controls, like in a tournament or in a county match, if you have an interesting and unusual antidote to the King's Indian, you can blitz it out, even while pretending to think for five minutes on a random move to give the idea that you don't know the opening. You take ten minutes on the opening 15 moves even after deliberately wasting five, and maybe your opponent does think for 45 minutes. In the middle game you can often turn your half hour time advantage into a complex position, where there are many pitfalls for black.

Yet some more proof that the KID is unsound.

Mercenary Levon Aronian going down to  a young Greek GM, rated some 200 points lower. The game finished a few minutes ago.

#40
This says nothing about the soundness of the King's Indian Defence.
12 c5 is new but bad.
24 Rg5 and 26 Qf7 play for the loss.

Brilliant. Loved the way white won. I would have preferred it if white had lost, because that would have meant that my method of playing white all these years would have been right.

Brilliant play. It shows that soundness doesn't enter into the equation, where black has such dynamic potential.