This is typically because you have chosen an inappropriate alignment algorithm for your dataset. The standard Geneious aligner is based on dynamic programming and will be slow when presented with long sequences or large sets of sequences. In the case of large multiple alignments, you should look at MUSCLE or MAFFT rather than the standard Geneious aligner. These are much faster and still quite accurate in most cases. For aligning genomes you should use the Mauve Genome Alignment plugin, which is designed to deal with the inverted and duplicated regions that a genome alignment is likely to have. See this article for further information about multiple alignment algorithms.
Aligning many primers against a set of sequences can also be slow. ‘Test with Saved Primers’ should be used for this, but this can also be really slow if primers have high levels of degeneracy and you are testing against lots of sequences. See the section on Primers for possible solutions.