The National Science Foundation has to take a second look at part of the evaluation behind an $8 billion, 20-year contract to run science and engineering support for the U.S. Antarctic Program, after the Government Accountability Office sustained a protest from Battelle Memorial Institute, the losing bidder, over how NSF scored the winning bidder's experience.
GAO ruled on September 10, 2026, and posted the decision to its public list around September 22. The watchdog agency found that NSF unreasonably credited KBR Services, LLC with the experience and past performance of an affiliated company when it evaluated KBR's proposal for the Antarctic Science and Engineering Support Contract, a single-award IDIQ supporting the U.S. Antarctic Program, even though KBR's proposal did not spell out what resources or role the affiliated company would actually contribute to performing the contract. In its decision resolving protest B-424575 and B-424575.2, GAO wrote plainly: "We sustain the protest in part and deny it in part."
The dispute traces back to NSF Solicitation No. 49100425R0006, which sought a single indefinite-delivery, indefinite-quantity contractor to support the U.S. Antarctic Program for two decades, with a ceiling value of $8 billion. NSF selected KBR over Battelle Memorial Institute and made award on June 2, 2026. Battelle filed a protest challenging seven separate aspects of that decision. GAO agreed with Battelle on one of them and rejected the rest.
Why GAO Found NSF's Affiliate Credit Improper
The sustained ground centered on how NSF treated an affiliated firm's track record as if it were KBR's own. Under the standard GAO applied, drawn from the Court of Federal Claims' 2026 decision in Noblis MSD, LLC v. United States, an agency may only credit a proposal with a related company's experience if the proposal spells out concretely what that company will actually do on the contract. KBR's proposal, according to GAO, did not clear that bar. The decision faulted the agency for accepting general assertions of corporate relationship rather than documented specifics tied to contract performance, treating a bare reference to a corporate affiliate as insufficient on its own.
That mattered because the affiliate credit was not a minor scoring footnote. GAO found NSF's source selection authority had leaned on the combined KBR-affiliate record when closing the gap between the two offerors on the experience and past performance factor, and that this narrowed evaluation disparity fed directly into the tradeoff decision favoring KBR. GAO concluded Battelle was competitively prejudiced as a result, meaning the error was not harmless — it plausibly changed the outcome of the award decision. That prejudice finding is what turned a technical evaluation flaw into a sustained protest ground rather than a denied one.
What Battelle's Other Six Challenges Couldn't Prove
Battelle threw a wide net beyond the affiliate-credit issue, and GAO turned down every other argument. The protester challenged a corporate restructuring and spinoff involving KBR, arguing it undercut the reliability of KBR's proposed team; GAO found NSF reasonably relied on KBR's own representations about how the restructuring would affect performance. Battelle also argued NSF's cost realism evaluation let KBR get away with unrealistically low labor rates, but GAO found the agency's judgment reasonable given the support KBR provided for its proposed rates.
The remaining grounds fared no better. Battelle claimed NSF applied a relaxed key-personnel or cooling-off requirement to KBR that wasn't extended evenly to all offerors, that a former NSF employee hired by KBR had improper access to competitive information, and that NSF's management-approach and comparative evaluations were flawed or unequal. GAO denied all of these, finding the agency's underlying judgments reasonable and adequately documented. In bid protest practice, denying six of seven grounds while sustaining one still counts as a loss for the awardee on that one point — and it's the point that determines what happens next.
What NSF Has to Do Now
GAO's recommendation is specific and narrow rather than a wholesale do-over. The agency directed NSF to reevaluate KBR's proposal under the prior experience and past performance factor without the improperly credited affiliate experience, then conduct a new tradeoff analysis and issue a new source selection decision based on that corrected record. GAO also recommended NSF reimburse Battelle for the costs of pursuing the sustained protest ground, including reasonable attorneys' fees. The award to KBR is not automatically overturned; NSF could still select KBR again once it redoes the affected part of the evaluation, or the corrected scoring could tip the tradeoff toward Battelle. Either way, the $8 billion program's award decision is not final until NSF completes that reevaluation.
What It Means for Contractors
The ruling is a reminder that "corporate affiliate" credit is one of the more heavily scrutinized moves in a proposal, particularly on large IDIQ competitions where past performance and experience factors can decide close calls. Offerors that want an affiliate's track record counted toward their own need to document, in the proposal itself, exactly which resources, personnel, facilities or services that affiliate will contribute to contract performance — not just assert a corporate relationship and expect evaluators to infer the rest. Agencies face the mirror-image lesson: crediting affiliate experience without that specificity in the record is a well-established basis for a sustained protest, and GAO has now reinforced it by applying a recent Court of Federal Claims standard directly to a bid protest decision.
For companies chasing large multi-decade IDIQs like the USAP support contract, the case also underscores how much weight sits on a single evaluation factor. Six of Battelle's seven arguments failed outright, yet the one that succeeded was enough to force NSF back to the drawing board on an $8 billion award. Protesters and awardees alike should expect GAO to keep treating affiliate-experience claims as a discrete, checkable fact question rather than a matter of evaluator discretion.
