A global search optimization system is applied for design of a horizontal axis tidal current turbine with shroud. 11 design parameters of the turbine blade and 4 design parameters of the shroud casing are considered for the optimization search by a genetic algorithm. For reducing the simulation cost, a neural network is applied as the meta-model of the RANS solver. Multiobjectives of a power coefficient at different tip speed ratio are applied for giving a function of wide operating range of the turbine. A proposed optimized design of the turbine shows a high output shaft power under a low tip speed ratio. Internal flow of the optimized horizontal axis tidal current turbine is discussed in detail. It is found that the optimized blade generates swirling flow and suppress flow separation at the diffuser wall. The wide angle of the diffuser successfully achieves higher pressure recovery ratio and results in a high suction power at the inlet of the turbine. It is found that the high performance tidal turbine is possible to design if both the blade and the shroud diffuser are optimized in same time.