Cosmic-ray (CR) antihelium is an important observable for dark matter (DM) indirect searches due to extremely low secondary backgrounds towards low energies. In a given DM model, the predicted CR antihelium flux is expected to be strongly correlated with that of CR antiprotons. In this work, we use the AMS-02 p̄/p data to constrain the DM annihilation cross section, and the ALICE data on the 3H̄ē and T̄ productions to determine the parameters in the coalescence model for anti-nucleus formation. The hadronic cross sections are estimated using Monte-Carlo event generators including EPOS-LHC and DPMJET. Based on these constraints, we make predictions for the maximal antihelium flux for typical DM annihilation final states, and perform a detailed analysis on the uncertainties due to the DM density profiles and CR propagation models. We find that the results are highly insensitive to both of them, but still significantly depend on the hadronization models in event generators. The prospects of detecting antihelium for the AMS-02 experiment is discussed. We show that with very optimistic assumptions, CR 3H̄ē is within the reach of the AMS-02 experiment. The 3H̄ē events which can be detected by AMS-02 are likely to have kinetic energy T ≳ 30 GeV, which is consistent with the preliminary AMS-02 search results. The events which can be observed by AMS-02 are likely to arise dominantly from secondary backgrounds rather than DM interactions.