Who needs whom?
It will be very interesting indeed to see how Oracle can deliver Fusion in any reasonable form to the market without this acquisition. They may not have the circumspection of someone like an SAP (or better still my old alma mater Baan) to have its set of APIs well (I mean very well documented a la War and Peace type detail) to publicly publish their APIs is some systematic way. Even if they did have it for Oracle (ERP) I doubt that you would find the same generousity found in the R & D budgets for JDE (who tested then refused to license Progress an RDBMS made to work on both the AS-400 and anything else you can think of (in 1995)). Or Peoplesoft home of THE Human Resources application (payroll only need apply).
What they may have difficulty with as does BEA (apparently) is packaging an offering that makes sense to the up and down-market segments thereby creating a captive and growing market. Unfortunately EAI (SOA, integration framework et. al.) never made it to the desktop and there is still no one in this town that can explain systematic implementation of one (it).
If I remember correctly (and please don't jump in the 'if') once a uniform standard of role definition and BPM can be arrived at (objects, subjects, and predicates), BEA would own the market. That hasn't happened.
The best kit to launch this from, and to profit from it, would be the BEA platform. I say kill BizTalk and long live the fully abstracted interchange framework.
Oracle should upgrade and deliver Fusion using BEA.