Earlier this year, the Defense Department Office of Industrial Base Resilience (OIBR) released a notice for public comment regarding actions the DoD can take to enable the Defense Industrial Base (DIB) to continue adopting artificial intelligence for defense applications.
The DoD acknowledged the DIB’s critical role in creating operational advantages and military superiority over adversaries, highlighting in the notice that “DoD collaborates with the DIB to rapidly adopt AI for defense applications and maintain the nation’s tactical advantage.”
Over the past several months, contractors and IT providers have responded to the DoD’s call to action. However, some concerning — but correctable — red flags have emerged regarding how AI, particularly generative AI, is being developed and deployed governmentwide. The DoD now finds itself at a pivotal juncture, where it can affirm commitment to DIB by embracing commercial technology or risk disillusionment among defense startups if military branches continue to leverage R&D money to build in-house competitor products.
Buy vs. build: Nip NIPRGPT in the bud?
When evaluating whether to buy vs. build, government agencies will rightly assess several key factors such as the superiority, cost, scalability, security and availability of commercial technology relative to what can be built in-house. At least, that’s how the process should work.
Those aforementioned red flags? One can be found with the launch of NIPRGPT. The Air Force introduced its generative AI platform in June 2024 to enable Guardians, Airmen, civilian employees and contractors to test and adopt generative AI in a secure environment. Advancing NIPRGPT offers an instructive test case on how building in-house when proven commercial options are available can disincentivize the defense industrial base from investing in the technologies, processes and manpower required to bring innovation to market.
Advancing an inferior product forward sets the technology back
Commercial products succeed from being battle-tested on hundreds, if not thousands, of use cases across customers. As a result, agencies benefit from the product evolution that comes from rolling feedback and enhancements. While cases exist where internal builds make sense (i.e., lack of superior commercial solutions), there are too many examples of products built by a single team in a vacuum leading to operational, feature/functionality and scalability issues if that product is rolled out agencywide.
In this case, if an inferior generative AI product is put forward, no matter how well-intentioned, it sets adoption back for the entire technology as initial users are left with a poor first impression that is difficult to overcome.
The devil is in the dollar details
Incentives to spur user adoption of an in-house generative AI product on the front end can come back to bite teams on the back end. With NIPRGPT, the government has asked teams to help fund the platform at a higher price than it would take to fund a DoD-wide license from proven commercial products. In other words, while an in-house solution may appear cost-competitive at the time of launch, by the time all is said and done, teams end up paying more for a generative AI product when the in-house product is actually years behind the capabilities and models of more advanced commercial products.
Disincentivizing first movers halts innovation
Defense startups with proven generative AI products lose first-mover advantage when they invest significant financial and human resources in technologies the DoD needs (and communicates they need), only to get burned when an agency opts to build rather than buy. Security and compliance costs alone for defense startups to go through FedRAMP and DISA Authorization processes can reach millions of dollars — a level of investment they will be unable to justify if DoD increasingly opts to build instead of buy.
What’s worse is that government teams are then required to justify why they won’t use an in-house solution like NIPRGPT, and justify why they prefer using a commercial product, which adds months to the commercial solution adoption process and creates revenue loss for those first-mover defense startups, potentially making them give up on the defense market altogether or worse, go bankrupt.
Restraining DIB drives users to the shadows
While agency leaders can direct teams toward an in-house product, users will ultimately gravitate to technologies that help them do their jobs better and faster. Suppose the preferred AI solution for airmen and guardians is not available enterprise-wide. In that case, they will turn to Shadow AI platforms operating outside secure networks and without the required security clearance for government work. Government employees using unauthorized AI applications may unintentionally share or upload confidential information, increasing the damage that data leaks or breaches would do. Unsecured solutions leave government agencies vulnerable to security hacks, compliance and regulatory risks, data exposure, and privacy issues.
Avoid stacking the deck against commercial
DoD must also focus on the upper stack with tangible mission outcomes, augment commercial stacks, and refrain from trying to shut them out and build over them. The ultimate goals are cost efficiency, superior products, and accelerated time to market. For example, instead of AFRL reinventing the scaffolding wheels of generative AI with NIPRGPT, they could have instead focused on building complex agents for specific DoD missions on top of a commercial solution. This would have brought tangible value to the warfighters and met the definition of true research and development that a research lab is supposed to focus on.
DoD must enable, rather than inhibit, DIB innovation
Federal Acquisition Regulations (FARS) 12 mandates government agencies procure commercially available products rather than develop their own. Beyond regulatory requirements, DoD and the nation benefit when the commercial partners are incentivized to advance technology innovation, and DoD stays true to its mission to invest in these efforts while continuing to strengthen public-private collaboration. The DoD cannot continue to cry that the DIB is shrinking and ask for startups and companies to join the fight if the DoD can’t put its money where its mouth is.
Nicolas Chaillan is the founder of Ask Sage.
Copyright
© 2025 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.