"The short answer is that if you lazily, short-sightedly, greedily allow your multi-million or -billion enteprise to become too dependent on any third party, whether outsourcer or cloud, they will contrive to entrap you and bleed you dry."
May I point you to the following statement in the DoD goals for this contract:
DoD currently has about 2,500 data centers.
“Our intent is to migrate as much as we can to a cloud environment and then regionalize those data centers,” Kinney said. “Maybe we only have 50 regional data centers across the department to support those folks that need to have a data center, a lot of legacy stuff.”
Many of the existing data centres are already outsourced, I believe amongst around 60 different vendors with all of the big DoD suppliers in there.
AWS already have a significant contract with the DoD for Milnet since 2013 and the DoD has been pushing for further consolidation.
The arguments around cloud bleeding the DoD dry are already present in the existing supplier model - the AWS model provides a way of at least addressing some of the cost issues. In terms of USP's, already having provided services to the DoD for 5 years, having a number of DC's already built for Milnet with the ability to expand and operating at a scale that the DoD has requested all seem like pretty key things.
Lookup Milnet and what the DoD have been trying to do to address costs and security concerns - for non-classified material, an AWS/Azure/Google solution (all three have big government data centres in the required areas) seem to match DoD's requirements a lot more closely than Oracle/IBM's proposals to start building DC's as soon as they win the contracts and to charge the DoD a significant amount in year one to meet those costs.
While I'm unsure of the validity of Oracles case to challenge the awarding of contracts over conflicts of interest that appear to have been attempted to be managed, Oracles key problem is that they don't have submission that comes close to other vendors regardless of the conflicts of interest, but maybe the case forces a re-tender. If the re-tender requires a rapid build out, I suspect Oracle loses again for the exact same reasons they lost the first time.