Leidos will keep a roughly $331 million Defense Threat Reduction Agency task order for counter-WMD targeting work after the Government Accountability Office rejected a protest from incumbent rival Applied Research Associates. The decision, B-424731 and B-424731.2, is dated October 28, 2026, and GAO added it to its public list of recent bid protest decisions on November 12. ARA had the higher technical and management ratings, but DTRA chose the cheaper offer, and GAO found nothing in the record that would overturn that choice.
ARA, of Albuquerque, New Mexico, challenged DTRA's technical evaluation, its cost realism analysis, its past performance assessment and its best-value tradeoff. GAO's answer was brief: "We deny the protest."
What DTRA Is Buying With the HDTRA1-25-R-0015 Order
DTRA issued the task order request for proposals on March 4, 2025, under FAR subpart 16.5. The order was open to holders of the agency's Combating Weapons of Mass Destruction IDIQ, Mission Pool 3, covering Targeting, Information Operations, and Irregular Warfare. According to the decision, the awardee will develop methodologies, models and tools for finding, characterizing, assessing and defeating underground facilities, hard and deeply buried targets, and foreign WMD. It will also support combatant commands and the intelligence community in defeat operations.
The single cost-plus-fixed-fee order has a one-year base period, four one-year options and a maximum performance period of 66 months. The solicitation ranked four factors in descending order of importance: technical approach, management approach, past performance and most probable cost. It stated that the non-cost factors combined were "significantly more important than cost or price."
ARA and Leidos, of Reston, Virginia, both submitted proposals. After multiple rounds of interchanges, DTRA issued the order to Leidos on July 31, 2026. ARA received Blue/Outstanding ratings for both technical and management approach, and Leidos received Purple/Good for both. Both firms were rated Acceptable on past performance. DTRA calculated a most probable cost of $348,732,095 for ARA and $331,864,900 for Leidos, a difference of about $16.9 million, or roughly 4.8 percent of ARA's figure. The source selection official concluded that ARA's unique strengths "do not provide sufficient benefit to the Government that would justify paying ARA’s higher evaluated cost."
How Leidos Turned AI Tools From Risk Into "Accelerators"
ARA's first argument focused on a weakness that DTRA assigned to Leidos and later removed. In its initial evaluation, the agency faulted Leidos for relying on four tools still in development: Shredder, a generative AI tool; CEDAR, a product for optimization and uncertainty quantification; TRFCTA, a customized AI tool developed for DTRA; and HyperCube, a repository of building standards, manufacturer information and part numbers. DTRA wrote that "it is not feasible to utilize these technologies within 18 months of contract award as it will take more time to fully develop them." It also found that the reliance presented a moderate risk to the program.
Leidos revised its proposal during interchanges. It removed the staff reductions it had planned at the 18-month mark and committed to existing commercial and government off-the-shelf tools and experienced personnel. It described the new technologies as "accelerators, not dependencies." DTRA accepted that revision and dropped the weakness.
ARA argued that Leidos had not shown when, or even whether, the tools would ever be ready. GAO found it reasonable for DTRA to conclude that the amended approach, which used proven methods while the new technologies matured, resolved the agency's concern.
Why an $18M Cost Realism Adjustment Was Enough
The central dispute was over surge staffing. The solicitation gave level-of-effort estimates for research and development and for operation and maintenance surge CLINs, but offerors were not required to bid those figures. Leidos proposed between 11.4 and 13 FTEs for the R&D surge CLINs, based on seven years of historical data from the predecessor contract and on a "scalable staffing approach." In its realism analysis, DTRA raised those CLINs to the independent government cost estimate of 28.6 FTEs, which increased Leidos' most probable cost by about $18 million. For the O&M surge CLINs, Leidos also proposed fewer FTEs than the agency's 10.5 FTE estimate at some points. DTRA left those costs unchanged and described the variance as "manageable and technically justified rather than a risk to mission execution."
ARA argued that DTRA "allowed Leidos to lowball its [LOE] estimates without consequence." GAO disagreed. It found that any risk of understaffing was effectively mitigated by DTRA's upward adjustment of Leidos' costs. ARA also argued that, because it had bid closer to the government's estimates, it was entitled to a downward adjustment of its own once DTRA accepted Leidos' leaner staffing. GAO rejected that argument too: "how ARA staffs its proposal is entirely a function of its business judgment," and a protester is not entitled to a cost adjustment merely because another offeror proposed a viable leaner solution.
A "Digital Twin" Drafting Error That Did Not Change the Outcome
ARA's strongest point may have been a mismatch in the tradeoff memo. DTRA initially gave Leidos a weakness for proposing digital twins. Leidos resolved it, in DTRA's words, "by rebranding digital twins as [multi-resolution models] ‘MRMs’." The final tradeoff nonetheless credited Leidos with a comprehensive understanding of future-state requirements "through Digital Twin Stage 2." The evaluation team lead stated in a declaration that the reference was a mistake and should have been to multi-resolution models. GAO found that explanation reasonable and concluded the reference was not a prejudicial error, noting that the evaluation elsewhere identified the underlying Leidos strength without the digital-twin wording.
GAO also rejected the past performance claim. DTRA acknowledged in the tradeoff that ARA provided more past performance references than Leidos. Because both companies were "performing exceptionally" as incumbents, the agency concluded that the volume of references did "not provide a qualitative advantage or a lower risk profile over Leidos." GAO found that conclusion reasonable.
Stinson LLP represented ARA. Wiley Rein LLP represented Leidos as intervenor, and Department of Defense counsel represented DTRA.
What It Means for Contractors
The decision shows that a higher-rated incumbent cannot count on its ratings to justify a higher evaluated cost when the solicitation says cost will be a consideration and the non-cost factors, while weightier, are not the only basis for award. Three practical lessons stand out. First, government level-of-effort tables in task order solicitations are only estimates. A competitor that bids below them with a documented rationale will face a realism adjustment but not further penalty, and bidding close to the government's numbers does not entitle an offeror to relief later. Second, interchanges let an agency remove a weakness when an offeror substantively changes its approach. Recasting unproven technology as an accelerator rather than a dependency, backed by proven staffing, can be enough. Third, drafting errors in a tradeoff memo rarely win a protest when the underlying evaluation supports the same strength. Contractors competing for task orders under DTRA's Mission Pool 3 IDIQ should read this decision as a reminder that a credible lean staffing plan can beat higher adjectival ratings.
