The FY2026 NDAA Is Rewriting Defense Manufacturing. Most Factories Aren’t Ready.
Walk onto a production floor building components for a Virginia-class submarine. The hull section was designed in a 3D CAD model with full geometric dimensioning. The engineering data lives in a PLM system. The digital thread, as far as engineering is concerned, is intact.
Now watch what happens when that design reaches the technician who actually builds it. The 3Dmodel gets flattened into a 2D drawing. The drawing gets printed. Someone pastes screenshots into a Word document and adds handwritten annotations. A technician pulls a three-ring binderoff a shelf, flips to a tab marked with masking tape, and starts work on a hull section worth more than the Mona Lisa.
That binder is the real engineering deliverable on most defense production floors. Not the CAD model. Not the PLM system. A binder, in a plastic sleeve, with someone’s handwriting in the margins explaining what the drawing doesn’t show. The digital thread breaks between engineering and manufacturing. And it breaks the same way almost everywhere.
What Congress Got Right
The FY2026 NDAA[1] reflects a serious diagnosis.[2] Section 221 mandates a digital engineering standards review across all military departments within 180 days.[3] That review will document a gap most program offices already know exists but have never been required to quantify. Section 1842 directs product support managers to assess which critical readiness items can be transitioned to advanced manufacturing and submit transition plans by April 2026.[4] This is among the most ambitious provisions in the bill, because it assumes the existence of production data that the vast majority of defense manufacturers do not generate. Section 1841 creates a Civil Reserve Manufacturing Network so commercial factories can surge into defense production, and requires DoD to establish an “advanced manufacturing crisis qualification framework” to streamline the qualification of new manufacturing sources and processes.[5] That framework only works if those factories can receive and execute defense manufacturing process documentation without rebuilding them from scratch. And for the first time, Congress legally defined “advanced manufacturing” to include software-controlled production.[6]
The bill tripled its AI provisions over the prior year.[7] Section 1535 stands up an AI Futures Steering Committee. Section 1512 requires department-wide AI governance and cybersecurity policy within 180 days. Section 347 directs integration of commercial AI tools into logistics, prioritizing nontraditional vendors.[8]
The intent is clear: digitize the industrial base, instrument it for AI, and build governance to sustain both. The diagnosis is correct. The question is whether the industrial base can fill the prescriptions.
Where the Theory Meets the Floor
The mandate is right, but it rests on incomplete assumptions.
Section 221 assumes a digital thread that runs from engineering through manufacturing. On most shop floors, it doesn’t. Engineering outputs a 3D model. Manufacturing consumes a printed drawing. The data changes format, loses fidelity, and exits the digital environment entirely. A standards review will surface this gap. It will not close it.
Section 1842 assumes manufacturing processes that can be qualified, versioned, and audited as structured data. But if work instructions live in binders and PowerPoint documents, there is nothing to qualify against. You cannot easily version a binder. You cannot fully audit a screenshot. Process-based qualification requires process-based data, and that data does not exist.
The AI provisions assume this most of all. Every AI application Congress envisions, from predictive quality to automated compliance to logistics optimization, requires structured, machine-readable production data as input. If your work instructions are printed, you have an archive, not a dataset. And no amount of AI funding changes that.
The Workforce Multiplier
This gap compounds the workforce crisis the NDAA is also trying to address. 82% of manufacturing workers who recently left the industrial base did so to retire.[9] They did not document their years of expertise because there was never a system to capture it.
Picture a senior welder at a submarine yard on his last day. He has thirty years of knowledge about which weld sequence avoids distortion on a particular hull section, which adjustments compensate for tooling that has drifted slightly out of spec, and which techniques he developed over decades through trial and error. None of it is written down. None of it is in any system. He shakes a few hands, walks to the parking lot, and three decades of institutional knowledge drive home with him.
The new hire who replaces him will learn from whoever is available, if anyone is. The NDAA funds recruitment and training. But you cannot train someone on knowledge that was never recorded. Digital production infrastructure is the prerequisite for workforce continuity.
What “Ready” Actually Looks Like
A production floor that can comply with these mandates looks fundamentally different from what most defense contractors operate today. Work instructions are digital, interactive, and derived directly from engineering models so that design changes propagate to the floor automatically. Process data is structured and traceable from the first step. Technician knowledge is captured in the system as it is applied, not reconstructed after the fact. Every step generates auditable records that feed quality, compliance, and AI-driven analytics without manual data entry.
Anduril has fully embraced this transition. Anduril runs more than 35 product lines with constant design iteration across multiple factories. They had over 100 manufacturing engineers spending half their time manually authoring and updating work instructions.
After evaluating enterprise incumbents and internal builds, they deployed Dirac’s BuildOS platform to generate interactive 3D work instructions directly from CAD files. Authoring time dropped from twelve hours to 90 minutes.[11] Engineering changes propagate automatically. Tribal knowledge gets captured in structured, searchable form rather than lost to turnover. The results went beyond production. As COO Matt Grimm put it: “One of the most unexpected outcomes of working with Dirac: it’s accelerated our sales.”[12]
The value extends beyond any single factory. The NDAA’s Civil Reserve Network requires commercial manufacturers that can convert to defense production. That conversion is impossible if every work instruction must be rebuilt from scratch. A platform that generates structured production data directly from engineering models is what makes surge capacity operational rather than aspirational.
The Numbers
The Navy needs 66 attack submarines and has 47.[13] Virginia-class production runs at 60% of the required rate,[14] and AUKUS commitments will widen the gap.[15] Columbia is 17 months behind schedule.[16] Munitions production sits at half its target.[17]
Congress wrote the mandate. Congress set the deadlines. But mandates do not build submarines. Production floors do. And until those floors are digital, automated, and generating the structured data that AI requires, the mandates describe a future the industrial base is not equipped to deliver.
The technology exists. The proof points exist. What remains is adoption at the pace the moment demands.
You cannot have digital engineering without digital production.
References
1. S.1071 — FY2026 National Defense Authorization Act
2. Miller & Chevalier: Key Provisions for Government Contractors in the 2026 NDAA
4. FY2026 NDAA § 1842, “Transition to Advanced Manufacturing for Certain Critical Readiness Items of Supply.” Requires product support managers to assess critical readiness items producible via advanced manufacturing within 24 months and submit transition plans. https://www.congress.gov/bill/119th-congress/senate-bill/1071/text
5. FY2026 NDAA § 1841, “Civil Reserve Manufacturing Network.” Directs DoD to establish “a process to streamline and expedite the qualification of advanced manufacturing sources, processes, or products prior to or during wartime or upon activation of the Civil Reserve Manufacturing Network.” https://www.gtlaw.com/en/insights/2026/2/fy2026ndaa/fy-2026-ndaa-the-substantial-impact-of-the-fiscal-year-2026-national-defense-authorization-act-on-federal-procurement-law
6. FY2026 NDAA § 1841 (codified at 10 U.S.C. § 4841), defining advanced manufacturing as “manufacturing through the use of interconnected, advanced technologies throughout the design and manufacturing process that enables modular, adaptable, and efficient manufacturing, including software-controlled subtractive manufacturing, additive manufacturing, powder-bed-fusion manufacturing, and other similar manufacturing techniques.”
7. GovWin IQ: AI Provisions in the Final FY 2026 NDAA
8. Akin Gump: Congress Moves Forward with AI Measures in Key Defense Legislation, Dec 2025
9. National Association of Manufacturers workforce survey
10. Justin Katz, “Amid shortage, Navy recruiting program struggles to keep half first-year shipbuilders,” Breaking Defense, Mar 2025. Brett Seidle testimony before SASC subcommittee. https://breakingdefense.com/2025/03/amid-shortage-navy-recruiting-program-struggles-to-keep-half-first-year-shipbuilders-official/
11. Anduril Selects Dirac to Power AI-Driven Work Instructions Across Its Factories, PR Newswire, 2025
12. Dirac Inc., post on X, featuring Anduril COO Matt Grimm
13. SECNAV Phelan testimony, USNI News, May 2025
14. CRS Report on Virginia-Class Submarine Program and AUKUS Pillar I, USNI News, Jan 2026
15. CRS Report on Virginia-Class Submarine Program and AUKUS Pillar I, USNI News, Jan 2026
16. Columbia-class delivery delays, USNI News, Apr 2025
17. Army 155mm artillery production, Defense One, Jun 2025
About Dirac
Dirac is a manufacturing technology company building the AI-driven system of record for production orchestration. Their product BuildOS is the first AI-driven work instruction platform that replaces document-driven manufacturing with an AI-driven, dynamic, model-based system. Dirac’s mission is to rebuild the industrial capacity of the West by turning manufacturing facilities into context-aware, adaptive, dynamic environments.
