The Department of Defense's push to adopt AI at scale, formalized in three coordinated memoranda in January 2026, has been reshaping defense contracting for more than a year.
The speed of that shift is easier to grasp through a single number than through any policy document: in the six months after its December 2025 launch, the Pentagon's enterprise AI platform GenAI.mil grew from 80,000 daily users to 1.5 million.
For Margarita Howard, who founded defense contractor HX5 in 2004 and now leads a workforce of close to 1,000 people, the pace of that pressure is not so unfamiliar. Howard has been investing in AI tools and preparing for AI-driven procurement changes for years. Throughout the process, she says, the driving question her teams face daily as they work on sensitive government contract work is simply where does automation actually improve the work, and where does it compromise it?
That distinction, between augmenting human expertise and substituting for it, has become central to how HX5 structures its work across research and development, engineering, information technology, and mission operations.
The company has invested in internal AI development and is preparing for AI-driven procurement changes ahead of government mandates. “If you don’t embrace AI, you’re just going to be gone,” she says. “We’re adopting it heavily.”
HX5 operates across a broad set of service categories. In research and development, the company supports government customers running technical programs where outputs are often one of a kind. In engineering, teams address systems that are operational, often fielded, and where errors carry steep consequences. In IT, HX5 manages infrastructure under compliance regimes with specific performance and documentation requirements. In mission operations, which includes supporting NASA and defense programs, the work sits closest to real-time consequences.
There is no uniform answer to where automation helps across those four categories, and Margarita Howard has been careful to treat the question as one that must be answered by service line rather than settled in advance.
In IT and compliance-adjacent work, Howard says the company has seen direct productivity gains. “We’re actually developing AI tools internally that we’re using and seeing benefits from,” she says. That category of work is largely repeatable and rule-bound: running consistent checks, tracking system configurations, flagging deviations from required standards.
Research and development work runs in a different direction. The value HX5 brings in supporting customer R&D settings requires domain knowledge and contextual understanding: knowing why a result matters, what its implications are for program objectives, and where it points the customer's thinking next. A model can process large volumes of data and deliver that processing to a knowledgeable worker, but it can’t necessarily replicate the institutional knowledge a cleared engineer has built over years working with a specific program office. That synthesis, and the judgment behind it, is what the government is contracting for.
Mission operations present the hardest case. When HX5 teams support time-sensitive, operationally consequential work, the margin for model error is set by the mission, not by software tolerances. The cleared personnel who perform that work are the accountability mechanism the government relies on. No automation framework changes that fact.
The way HX5 recruits reflects this framework. More than 30% of the company's roughly 1,000 employees are veterans. That proportion is not incidental to HX5's service model. People with military backgrounds already understand the operating environment: the culture of accountability, the stakes attached to mission work, and the judgment that comes from having performed consequential tasks in conditions where errors were not abstract.
“Experience in their respective fields, while supporting these [government] agencies’ respective programs and missions, is very different from experience gained from working in the commercial world,” Howard says.
HX5 has participated in the Department of Defense's SkillBridge program and the Hiring Our Heroes Corporate Fellowship Program since 2021. Transitioning service members who move through those programs arrive with security credentials already in process and direct experience in the government environment HX5 serves. The company's 2025 HIRE Vets Gold Medallion Award from the Department of Labor recognized that pipeline as one answer to a persistent labor-market problem.
As of 2023, the most recent comprehensive estimate placed more than 70,000 security-cleared positions unfilled across the defense sector. The average end-to-end processing time for a Top Secret clearance ran to 249 days as of late 2024. Together, those figures make the cleared workforce look less like an abundant resource and more like one that takes most of a year to build, one credential at a time. A clearance can’t be issued by an algorithm. The judgment that builds over years of performing mission-critical work at a specific location can’t be replicated by a model. The professional trust that develops between a cleared team and a program office takes time and can’t be replaced by simply deploying software.
The FY2026 NDAA's AI provisions add a layer of compliance complexity that Howard views as a natural extension of the discipline HX5 already maintains. Section 1513 directs the DoD to develop a cybersecurity framework specifically for AI and machine learning systems acquired by the Pentagon, with explicit attention to supply chain risks: data poisoning, adversarial tampering, and unintentional data exposure. That framework will be incorporated into CMMC requirements and the Defense Federal Acquisition Regulation Supplement, meaning contractors that develop, deploy, store, or host AI tools for DoD work will eventually need to demonstrate compliance with it.
The urgency behind those requirements is easier to understand given the pace of what has already happened inside the DoD. Within two weeks of gaining access to agentic AI tools on GenAI.mil, Pentagon personnel had created more than 100,000 semi-autonomous agents. The existing compliance architecture was not built to handle that rate of deployment, and the NDAA's push for a dedicated AI security framework addresses exactly that gap.
For HX5, the trajectory of these requirements is not new in kind. Howard has been tracking CMMC’s development since before its formal rulemaking cycle closed. “We try to stay ahead of changing technologies like artificial intelligence and cybersecurity,” she says. “It’s expensive to ensure it’s done right, but it’s worth it.”
That posture extends to how Howard reads the NDAA’s AI framework. “There are heightened cybersecurity requirements,” she says, “and contractors will not have a choice but to implement them if they want to be a government contractor.”
The Pentagon’s June 2026 status report deadline to Congress marks how quickly that logic is now reaching AI. The AI security framework follows the same arc as CMMC: NDAA provision, acquisition regulation rollout, and eventually a disqualifying condition for contractors who have not kept pace.
Howard's view of the environment is that the contractors best positioned in this period are those who can answer a foundational question before it becomes a requirement: what is the human role in this work, and what are the consequences of removing it?
Suggested reading:
Subscribe to our channels on YouTube and WhatsApp
Download our app on Play Store