IBM first derives the expected profit and takes that. Pays the sales commissions (or just a portion if they can) and the leftover is what they budget for project expenses. That is goods and materials. Hardware on the cloud is essentially a fixed low margin cost these days. So that leaves human resources as the last bastion to be squeezed. If what is left doesn't cover the project then offshore is a good solution in this situation. Ignore the fact you painted yourself into a corner where offshore is a good solution. It is ultimately a shell game run by managers to hide and shift problems long enough to bolt to the getaway car after robbing the bank.
This is a reply from another thread but @Wk4yXGT-1vda is on point.