Relational Database Theory would suggest Company >> has many >> Projects >> has many >> Tasks.
When the Company information can be kept and maintained in one place, (one record per company in the Company table), you greatly reduce the risk of redundancy and versioning.
When you open a new project record and manually enter company information there is always the risk of mis-spelling the company name -- then you have a duplicate error.
If you can 'select' the Company record from a drop down list that comes from the Company parent table, you can bring into the Project record lookup information on that Company, i.e. address, contact phone #, current manager, etc. without having to manually enter it. This way makes for a cleaner database as well as other efficiencies server side.
My two cents worth of guidance. Have a good week.
