1. e4 c5 2. Nf3 d6 3. d4 Nf6 4. Nc3 cxd4 5. Nxd4 g6 6. Be3 Bg7 7. f3 O-O 8. Bc4 Nc6 9. Qd2 Bd7 10. O-O-O Qa5 11. h4
11 Kb1 is probably the most often-played move, but it is not the most challenging or interesting move. With 11 h4 White announces that he is going to attack at every opportunity, and challenge Black to repel the attack with accurate play.
Rfc8 12. Bb3 Ne5 13. h5
Proceeding in standard fashion, White sacrifices his h-Pawn to open up the h-file for attack against the Black King.
Nxh5 14. g4 Nf6 15. Bh6 Bxh6
White was threatening 16 BxB KxB 17 Qh6ch followed by 18 Nd5, forcing the removal of Black's King Knight which is guarding h7. Theory recommends the immediate 16...Rxc3, eliminating the White Queen Knight which was the threat against the Black Knight on f6, but I prefer the Bishop trade first. This induces the Black Queen to go to h6, whereas after 16...Rxc3 17 bxc3 Bxh6, theory holds that White should recapture on h6 with the rook, as Black gets the better game with White's Queen on h6. Therefore, why not induce the Queen to h6 a move earlier, while White has the hope that perhaps Black will ignore the threat and fail to take on c3, giving White a winning attack due to his h-file pressure. In chess, as in life, conventional wisdom is sometimes wrong!
16 Qxh6 Rxc3 17 bxc3 Qxc3
Sometimes Black will delay this capture, but I see no point to this. Black now has a Knight and 2 Pawns for his Rook, which is plenty of compensation in this position. In fact, the endgame, if Black should get to it, greatly favors Black as White's Rook has no open files to operate on.
Black was threatening the check on a1 followed by taking the Knight on d4 with check. Other White tries here have involved moving the Knight, either to e2 or f5, but these tries have turned out badly for White. His best move is clearly the Kb1 try.
Out of 10 games in the database, 18...Nc4 was played an equal number of times to this. But I see no reason to play ...Nc4 now, as White has no threat yet which requires this. At such time as the Bishop on b3 presents a credible threat to Black's Pawn on f7, certainly Black should then play ...Nc4 to block off the Bishop. By bringing the rook into play, Black opens up the option of taking with the rook on c4 when White plays BxN.
Also, by delaying ...Nc4 and playing ...Qxc3 immediately, Black opens up a threat against White's Pawn on f3, which gives White an additional worry to ponder. The Pawn on f3 is the only thing holding White's fragile King-side Pawn structure together, and when it falls, White's position is in serious danger of structural collapse.
Sometimes Black has also played 18...b5 here, which seems like a pointless waste of time to me.
After 18...Rc8, Black has placed White at a crossroads. White has entered into this position in attacking mode, but now there is no good way to continue his attack. f4 and Nf5 are the only attacking moves available, but these have proved futile. In fact, of 6 games in the database, the only move which produced a White win was Qd2, a defensive move which most White's would be loathe to play. (And that win only came as a result of a tactical trick at the end.) Attempts by White to guard the Pawn on f3 by Rhf1 or Rh3 also were not effective.
So, the conclusion is that Black has more than equalized, and the Dragon has served Black well. If it gets to an ending, Black's Knight and 2 Pawns will have a clear advantage over White's Rook which has nowhere to go. There is also the basic question of whether White is up to the challenge of losing his Pawn on f3, which Qd2 does.