Three years. That's how long the Pentagon now waits for certain critical munitions, according to the Center for Strategic and International Studies—a timeline that has shifted from bureaucratic inconvenience to strategic liability. By June of this year, unfilled orders for defense aircraft and parts had climbed to $79.6 billion, according to Federal Reserve data—a backlog that isn't shrinking so much as metastasizing.
The crisis in America's defense industrial base has moved past the point where it can be politely described as a bottleneck. This is structural collapse, dressed in procurement language.
Into this strained landscape has arrived a cohort of startups with an audacious premise: that artificial intelligence can do more than optimize the manufacturing processes we already have. They want to reimagine production itself, treating the orchestration of aerospace parts not as a series of human-mediated handoffs but as a computational problem waiting to be solved. GUILD, a two-person company founded in 2025 and backed by Y Combinator, is among the newest entrants. It joins better-capitalized players like Hadrian and Machina Labs in what might be called the AI-native manufacturing movement—systems where algorithms handle everything from contract interpretation to supplier vetting to delivery logistics.
Whether automation will reach aerospace production is no longer the question. Major CAD software providers already ship generative AI assistants as standard features. The real question is whether this new generation of companies can navigate the regulatory thicket that has kept the defense supply chain concentrated among a shrinking club of established contractors. Certifications, export controls, quality management protocols—these aren't just bureaucratic hurdles. They're the moat that protects incumbents and, perhaps more importantly, keeps out newcomers.
When Capacity Meets Its Limit
The numbers tell a story of an industrial base stretched past capacity and still accelerating. Global military spending hit $2.887 trillion in 2025, up 2.9% year-over-year, according to the Stockholm International Peace Research Institute. European defense budgets surged 14%. Asia-Oceania climbed 8.1%. The Pentagon's fiscal 2026 budget request emphasized shipbuilding, missile defense, and munitions—precisely the categories where production constraints bite hardest.
A progress report from the Center for Strategic and International Studies in July laid out the scale of the challenge. The Department of Defense has resorted to multiyear procurement contracts and direct supplier investments to catalyze capacity for weapons systems that, in some cases, take longer to manufacture than to design. The Air Force has adopted AI toolkits for condition-based maintenance and supply-chain management for propulsion overhauls involving "thousands of parts," according to a July report in Janes.
Meanwhile, small-business participation in the defense industrial base has declined more than 40% over the past decade, per the Pentagon's 2023 Small Business Strategy. The supplier base is aging, consolidating, and—most critically—operating at full tilt with limited ability to absorb new demand.
Commercial aerospace faces parallel constraints, which suggests this isn't purely a defense procurement problem. The International Air Transport Association warned in December 2025 that aircraft and engine supply mismatches were unlikely to normalize before 2031 to 2034. By June 2026, IATA was calling for urgent action on engine maintenance, repair, and overhaul bottlenecks, citing durability issues, parts shortages, and limited spare engine availability.
Convergence of Urgency and Capability
Two forces have converged to make this moment distinct, perhaps even pivotal.
First, the Pentagon's capacity requirements have become impossible to ignore. The Associated Press reported in August on the Department's push to accelerate weapons production amid stockpile concerns—conflicts in recent years have depleted interceptor inventories faster than factories could replenish them. The Navy's partnership in March with Hadrian to mass-produce critical submarine components illustrates the urgency. Roughly $900 million in Navy funding combined with more than $1.5 billion in private capital backed a facility in Cherokee, Alabama, designed to relieve bottlenecks in Virginia- and Columbia-class submarine production.
Second, the technology stack required to automate aerospace manufacturing has matured in ways that seemed speculative just five years ago. Model-Based Definition standards like ASME Y14.41-2019 create machine-readable technical packages—blueprints that algorithms can interpret. PTC's Creo 13, released in June, added an AI assistant for design guidance and enhanced generative capabilities. Siemens NX Manufacturing 2606 introduced generative AI machining suggestions the same month. Autodesk's Fusion roadmap now centers its Autodesk Assistant to automate sketch constraints, drawings, and toolpaths.
These aren't peripheral features. They represent a shift from CAD as documentation to CAD as the starting point for algorithmic production planning.
At the same time, export controls, quality management systems, and cybersecurity requirements create what amounts to a regulatory moat. ITAR governs defense articles. EAR covers dual-use aerospace components. DFARS specialty metals restrictions require domestic melting and production. CMMC 2.0 mandates cybersecurity self-assessments, though the Pentagon suspended Phase II third-party assessments in July for a 60-day review. AS9100 quality management, AS9102 First Article Inspection, and Nadcap accreditation for special processes define the table stakes.
These barriers protect incumbents, certainly. But they also create opportunity for companies that can bake compliance into their operating models from inception.
The New Entrants
GUILD positions itself as what co-founder Erim Gurlemis describes as an "AI-native defense contractor to make procurement and production faster, more efficient, more cost-effective." The company operates in two connected modes. As a prime contractor, GUILD wins government manufacturing contracts and delivers finished parts. Simultaneously, GUILD Forge offers production-planning software that translates designs, bills of materials, and requirements into executable procurement and production workflows. The focus is aerospace precision metals: jet and engine components, structural parts, housings, brackets, fittings.
Co-founder Kaya Celebi holds a master's in computer science from Columbia and previously served as AI Tech Lead at Hartree Partners, with machine learning experience at Morgan Stanley. The company describes what it calls a "company brain"—an AI-native operating model connecting requirements, pricing, supplier capability, compliance, manufacturing, finance, and delivery. In theory, every contract executed compounds the system's data and learning.
GUILD is early-stage—very early. Founded in 2025, the two-person New York team joined Y Combinator's Summer 2026 batch. Funding amounts and specific contracts remain undisclosed. But the company's framing as prime contractor plus software platform differentiates it from digital procurement marketplaces that simply connect buyers to existing suppliers without owning execution risk.
Hadrian has pursued a different path, and at a different scale. The company raised a $1.37 billion Series D in August, according to Axios, at a post-money valuation just under $8 billion. It operates 3 million square feet of highly automated factory space across four sites, serving traditional primes like Lockheed Martin and Raytheon as well as newer defense companies. Hadrian's proprietary Opus software orchestrates what CEO Chris Power calls "factories-as-a-service" for precision parts. "This is a no-fail mission to make sure that we have the production capability domestically," Power told Axios. "Being able to produce—and have advanced factories—is deterrence."
Machina Labs raised $124 million in February, according to Axios, to scale what it calls its "Intelligent Factory" for missile structures and airframes. The company's RoboCraftsman system performs robotic sheet-metal forming and assembly. In July, Machina formed a defense subsidiary, Machina Bellator. Lockheed Martin appears as a customer in industry coverage. The pitch centers on software-defined robotic forming that can scale to thousands of assemblies.
GrayMatter Robotics pursues what it terms Physical AI for surface finishing and coatings. An AFWERX Phase II contract awarded in February funds development of an AI-powered robotic system for canopy sanding automation. The company emphasizes air-gapped, edge deployments suitable for defense sites—a critical requirement in classified environments.
Seurat Technologies was selected in late July for America Makes' JAQS-SQ program—a $10.5 million ManTech-funded effort to qualify and expand the metal additive manufacturing supplier base. Seurat's Area Printing technology represents high-throughput metal AM, a potential alternative to traditional machining for certain components.
Established players are adapting too. Xometry markets ITAR-registered, AS9100D-certified, CMMC Level 2 capabilities with what it pitches as AI-enabled sourcing. Protolabs operates U.S. factories with ITAR compliance and AS9100 certification. Fictiv expanded export-control services for EAR-regulated programs in March, though the company does not support ITAR builds.
The Air Force, meanwhile, has moved digital sustainment from pilot to operational deployment. The Rapid Sustainment Office demonstrated a LITE Digital Maintenance Binder at Nellis Air Force Base in January, cutting documentation time and errors. In May, the Air Force Sustainment Center's RAPID Lab delivered an additively manufactured F-35 canopy frame scanned and produced at Hill Air Force Base.
What Comes Next
The modernization underway in the defense industrial base isn't optional. A July analysis from EY noted that the global aerospace and defense backlog represents well over a decade of production at current rates. Elevated defense spending isn't a spike—it's a sustained new baseline, driven by geopolitical realities that show no sign of abating.
For founders, the opportunity lies less in building better versions of existing manufacturing processes and more in reconceiving production as an integrated computational system. Model-Based Definition creates machine-readable technical packages. Generative design tools automate geometry optimization. AI assistants in CAM software suggest toolpaths. The remaining question is orchestration: how to stitch procurement, compliance documentation, supplier qualification, production planning, and delivery into a system that compounds learning with each iteration.
For defense executives and supply chain operators, the choice is whether to partner with this emerging ecosystem or risk being bypassed. The Navy's Hadrian collaboration signals openness to public-private models that would have been unthinkable a decade ago. GE Aerospace's partnership expansion with Palantir in March to deploy "agentic AI" for military aircraft readiness suggests even traditional primes see value in algorithmic approaches, though perhaps reluctantly.
Investors evaluating this space should watch for differentiation beyond automation. Every factory can add robots; that's table stakes. The companies that will matter are those building institutional knowledge graphs—systems that capture not just how to make a part but how to navigate AS9102 First Article Inspection, DFARS specialty metals sourcing, Nadcap accreditation, and contract compliance in a way that gets faster and cheaper with scale. That's a different proposition than deploying more cobots on a factory floor.
Policymakers face a harder problem. The small-business supplier base has hollowed out, and reconstituting it requires more than funding. It demands redesigning procurement processes to reward speed and cost over incumbency—a shift that runs counter to decades of institutional inertia. CMMC's suspension and review in July suggests even well-intentioned cybersecurity requirements can become barriers to entry, choking off the very dynamism the system needs.
The question for 2027 and beyond is whether AI-native manufacturing represents a genuine architectural shift or simply a rebranding of existing digital manufacturing trends with better marketing. GUILD and its cohort remain unproven. But the capacity crisis is real, the technology is maturing, and the Pentagon is running out of alternatives.
Perhaps the most telling metric will be how quickly these startups move from prototype contracts to programs of record. In defense, that transition separates ambitious pitches from actual production capacity. Right now, the industrial base needs the latter far more than the former.
