Oops. yes. Another query1, query2, not your query1, query2. Sorry about that.

First most likely issue, the two different tables in the query, DIM_MATERIAL vs. DIM2_MATERIAL are showing different statistics right from the start. The cardinality for one is 82000 the other is 81999. That's a miniscule difference, but if that statistic is different, that might explain why the two plans are different. All the other estimates are a little off. The principal difference is in where the Compute Scalar operation is located. I'd suggest getting the statistics up to date as a starting point.

The second plan timed out during compile, so that might explain the difference too. Again, possibly caused by the differences in structures, possibly relating to statistics or something else around DIM2_MATERIAL.