1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.Nc3 e6 5.e3 Nbd7 6.Qc2 Bd6 7.Be2 0-0 8.0-0 dxc4 9.Bxc4 b5 10.Be2 Bb7 11.a3 Re8 Aronian is an expert of the Semi-Slav with both white and black, which is possibly why Shirov chooses this uncommon continuation instead of the more usual ...a6.
12.Rd1 Qb8 13.e4 e5 14.g3 a6 15.dxe5 Nxe5 16.Bg5 Neg4 17.h3 There can be no doubt Aronian saw Black's next move, which isn't winning per se, but what was the alternative? [White has to try and kick the knight off, and 17.Nh4 also gives his own knight plenty of good squares to choose from. 17...Bc5 18.Rf1 Qa7 19.Bf3 h6 20.Bxf6 Nxf6 21.Nf5 with a dynamically balanced position with a slight tug for Black.]
17...Bxg3! 18.hxg4 [Of course not 18.fxg3?? Qxg3+ 19.Kh1 Nf2# ]
18...Nxg4 19.Be3 Nxe3 20.fxe3 Re6 So how do things stand? Is Black winning? Is White? In all honesty, a draw with some perpetual theme seems likely, since Black does have enough for that, but doesn't seem to have any means of bringing in the other pieces with any ease, though one should never discount Shirov's boundless imagination in these situations.
21.Bf1 Qa7 22.Bh3 Rh6 [22...Qxe3+ would be a serious mistake, since after 23.Kg2 Rg6 24.Bf5 Qh6 25.Rd7 Bc7+ 26.Bxg6 Qxg6+ 27.Kh1 White regroups and is winning.]
23.Kg2 Bc7 24.Bf5 Qxe3 25.Rd7 Qf4 26.Qf2 Bc8 27.Ne2 Rh2+ 28.Nxh2 Qxh2+ 29.Kf3?! An inaccuracy that gives Black chances for more than a draw now. [29.Kf1 was best.]
29...Qh5+ 30.Bg4 Qg5 31.Rad1 Bxd7 32.Rxd7 Be5 33.b3 h5 34.Bf5 g6 35.Bh3 Bb2?! [35...Re8 was stronger. White would not have time for 36.Ra7 because of 36...Rd8 and White has to trade rooks, leaving a nasty endgame.]
36.Nf4 Qf6 37.Kg2 Bxa3 38.e5 Qg5+ 39.Kh1 Qxe5 40.Nxh5 Rf8?! [40...Be7 was preferrable as it cuts off the rook and prevents Nf6+ as in the game.]
41.Nf6+ Kg7 42.Ng4 Qc3 43.Kg2 Bc1 44.Qf3 Qxf3+ 45.Kxf3 a5 46.Rc7 Bb2 47.Rxc6 a4 48.bxa4 bxa4 49.Bf1 Ra8 50.Rc7 a3 51.Bc4 a2 52.Rxf7+ Kh8 53.Bxa2 Rxa2 54.Ke4 1/2-1/2