The Defense Department’s strategy to transition to cloud computing. Note both the prominent role for the JEDI project and the parallel system of “Fit For Purpose” clouds.

Recently released court documents reveal that federal judge Eric Bruggink had some doubts about the official justification for awarding the $10 billion JEDI cloud computing contract to a single vendor. But ultimately, he ruled, those doubts didn’t outweigh the Pentagon’s sound reasons for rejecting Oracle’s bid.

While there may have been some “smoke” in Oracle’s allegations of conflict of interest, Bruggink wrote, there was no “fire.”

Microsoft Azure and Amazon Web Services are the only two competitors remaining for the controversial Joint Enterprise Defense Infrastructure contract after the Defense Department disqualified IBM and Oracle. Oracle immediately protested the decision to the Government Accountability Office and then, when rebuffed by GAO, filed suit. On July 12th, Bruggink ruled that Pentagon employees who’d later gone on to work with Amazon had not unduly influenced the contract, and that the Defense Department had the legal right to rule out Oracle on the grounds that it lacked a sufficiently large and dispersed system of data centers to remain operational during a catastrophe.

But the brief summary Judge Bruggink released at the time was silent on the biggest, most contentious concern raised in Congress and elsewhere about JEDI: Was the Defense Department allowed to award the contract to a single vendor — something only Amazon and perhaps Microsoft were big enough to manage — or did it need to break it up among multiple vendors — as smaller competitors like Oracle would prefer.

Bruggink’s full 60-page ruling had been sealed until this past Friday to permit the parties to propose redactions. (The court accepted two redactions proposed by Amazon Web Services to address protected information).

As Bruggink summarized the issue, the Federal Acquisition Regulation (FAR) governing this kind of acquisition — a so-called indefinite-quantity contract where the government may repeatedly order new goods or services off a single contract — states a preference for making multiple awards, so there can be competition for each order.

But the FAR allows for six exceptions. Bruggink ruled that the contracting officer decided on a single award based on three of those six:

• “Based on the CO’s knowledge of the market, more favorable terms and conditions, including pricing, will be provided if a single award is made”;
• “The expected cost of administration of multiple contracts outweighs the expected benefits of making multiple awards;” and
• “Multiple awards would not be in the best interests of the Government” (a classic catch-call escape clause).

“The regulation is unambiguous: even in light of the multiple award preference, [t]he contracting officer must not use a multiple award approach if one of six listed conditions exists,” Bruggink wrote (emphasis ours). “The question is whether the CO rationally determined that any of the three chosen conditions exist. We believe she did. Plaintiff offers us no real no basis for questioning any of these conclusions. They were completely reasonable, and we have no grounds to disturb her conclusion that multiple awards cannot be used.”

The Defense Department issued a statement of support for the judge’s rulings on Sunday evening:

“Read in its entirety, the ruling highlights that DOD’s unique needs for the warfighter have consistently driven decisions related to JEDI,” said DoD spokeswoman Elissa Smith. “The Court examines DOD’s evaluation criteria, particularly the requirement that bidders can meet minimal cyber security standards for handling government data, and determines that they are reasonable and enforceable. The Court also thoroughly reviews Oracle’s allegations about the potential conflicts of interest, which have been the subject of poorly-informed and often manipulative speculation. The Court concurs with DOD’s extensive review – that the individuals at issue were ‘bit players,’ in the Court’s own words, and the alleged conflicts had no impact on the integrity of the procurement.”

In his ruling, however, the judge did acknowledge that the Pentagon’s rationale for issuing a single award for JEDI may run afoul of a different acquisition rule that says task or delivery orders exceeding $112 million can only be awarded to single sources under one of four exceptions. Now, the Defense Department did cite one of these four. Specifically, “the contract provides only for firm, fixed price task orders or delivery orders for products for which unit prices are established in the contract; or services for which prices are established in the contract for the specific tasks to be performed.” In other words, there’s no need to ensure competition on each order because the price is already fixed up front.

In its protest, Oracle argued that nobody can foresee the need for all possible future products and services, let alone price them. The judge’s ruling agreed:

“At first blush, DoD’s (determination and findings) tracks precisely with the chosen exception: the JEDI Cloud RFP provides only for firm, fixed price task orders. It solicits IaaS (Infrastructure as a Service), PaaS (Platform as a Service), and support services for which offerors will propose a catalog of prices; that catalog will be incorporated into the contract, i.e., established, at the time of award. If the prices of all possible tasks were ‘established’ in this fashion, then we would agree that exception could be relied upon. That is not the case, however,” Bruggink wrote. “As Oracle points out, there is a logical disconnect between claiming that prices are ‘established in the contract’ for ‘specific tasks’ while simultaneously acknowledging that those tasks, and their accompanying prices, do not yet exist.”

That disconnect, however, seems to have struck Bruggink as more of a quibble than a dealbreaker, because he decided it did not override an otherwise sound decision against Oracle.