The US Jedi project was critical to the future of the US Defense department, far from trivial at $10B dollars, appeared to have been bid correctly, yet it failed due to two problems.
One was the perception of impropriety unacceptable in any government contract; additionally, the delays due to contract challenges have rendered the Jedi approach obsolete.
While it might have appeared that the conflict was between Amazon and Microsoft, there was no evidence of either company doing anything wrong. Still, it did showcase how image is essential in contracts like this and that Microsoft needed to deploy marketing resources if they wanted to keep the deal.
However, doing so, mainly if poorly done, could have created additional strategic problems for Microsoft, making the best path for all parties to re-bid the deal altogether.
Amazon had a problem with the last Administration, which had little to do with the company and more to do with Jeff Bezos, who was then CEO of Amazon, holding ownership of the Washington Post. The Washington Post was an outspoken critic of the last US administration, creating an impressive animosity between that Administration and Jeff Bezos personally.
This apparently led to attempts by the Administration to steer the deal away from Amazon. With Bezos stepping down from Amazon leadership and the recent change in the White House, this should have mitigated this exposure for a future attempt. Still, it creates a cautionary warning for anyone planning to sell to the government and comments at scale about that government’s performance.
The DOD’s Problem
This effort, which probably wasn’t successful, still created a cloud of perceived bias and the related email chains, which could be and likely was subpoenaed. But Amazon’s challenge to the deal would have created enough smoke for a judge to determine that undue influence was at least likely resulting in the re-bid anyway.
In addition, Jedi appeared to be poorly thought through. It relied on a single vendor reducing redundancy in what should have been a far more redundant, multi-vendor solution.
By not approaching this as a multi-vendor solution, the possible ideal mix of IBM providing both the blending of both Amazon and Microsoft into a fully redundant large scale solution, placing IBM’s security-focused cloud as a critical component, didn’t appear to be considered. On the face of the effort, that could have been a configuration that would seem to provide both the redundancy and security requirements more thoroughly than any one of the vendors can provide. While the IBM solution likely didn’t exist when Jedi was conceived, it exists now and would seem to match better what the DOD wants than either Amazon or Microsoft alone could provide.
In addition, a multi-vendor solution is far more palatable because it both spreads the wealth and assures redundancy across cloud vendors should any one of the vendors fail due to physical or electronic threat likely with any Defense-focused offering.
Why Not Google?
In terms of capability, Google should match Amazon and Microsoft, but they were left off the list of acceptable vendors. This unacceptability is likely due to two reasons.
First, Google, unlike the others, sells information for a living, and while it’s highly doubtful they would sell DOD information given the dire consequences, the fact they do sell information makes them undesirable for anyone concerned about security. The other reason is that Google employees have been outspoken about not having Google support any military organization; Google has had employee breaches (and a data breech with Google+) in the past and that certainly didn’t help.
Wrapping Up: Large Cloud Bids
When you are approaching any cloud effort at this scale, the danger of any one cloud vendor having a catastrophic failure, particularly in a Defense Department-focused solution that will be targeted, is excessively great. While I believe a combination of IBM, Amazon, and Microsoft would best match the needs of the DOD and prevent future challenges to the effort, mainly because it isn’t uncommon to see this solution at this scale, it should have always been a multi-vendor solution that was specified.
The result is potentially more robust, more secure, and spreads the related revenue around so that no one area is uniquely benefited, providing better political top-cover than a single vendor solution. Jedi was problematic from the start; we’ll know shortly if the DOD has learned from their mistake and uses a more prudent approach or swaps one mistake for another. The group doing the bidding this time appears to be more capable. Let’s see if they actually are.