Lockheed Martin — the Maryland-based defense contractor that produces fighter jets, missile systems, satellites and military helicopters — said in early March it was in the process of quadrupling production of “critical munitions,” as global demand rises, including needs tied to the U.S. war with Iran.
That ramp-up is being supported by artificial intelligence tools developed over several years at the company’s Shelton-based center, where Lockheed has built a centralized system to deploy AI across its global operations. The platform allows teams to share data, automate manufacturing processes and identify supply chain constraints, helping speed production.
According to an industry analyst, that coordinated approach has made Lockheed Martin a leader in AI adoption in the defense sector.
The Lockheed AI Center, or LAIC, traces its origins to an engineering team formed about a decade ago at Sikorsky Aircraft in Stratford.
“We were working on what we call prognostics and health management,” said Greg Forrest, now the vice president of AI foundations and commercialization for Lockheed. “Essentially using data to predict when components on a helicopter may fail.”
The team was in the right place at the right time. When Lockheed acquired Sikorsky in 2015, their expertise in AI — combined with the increasing availability of big data sets and advances in computing power — made the technology useful across more parts of the company.
Forrest and his team proposed what became the AI Center in 2021. It now encompasses about 350 engineers distributed across the company, including 25 in Connecticut.
“We wanted to drive a hub-and-spoke model across the business,” Forrest said. “We provide underlying foundational technologies — the tools that teams use to build AI — and then we proliferate that across all of our business areas.”
Together, those tools form what Lockheed calls its “AI Factory.” The defense giant’s focus has since shifted from building those tools to expanding their use across the company.
‘Own how we build it’
Ensuring the technology is used across different programs is the job of Grace Scanlon, a former Sikorsky engineer who has worked at the AI Center since its inception.
“We’re opening the door to new technologies in a central way that makes it feasible and cost-effective for a lot of programs where it otherwise wouldn’t be,” said Scanlon, who is an AI systems engineering manager.
That centralized approach is critical given the high cost of computing power needed to run AI models. By housing that infrastructure in one place, Lockheed Martin — which employs more than 120,000 people worldwide and reported about $71 billion in revenue in 2025 — allows engineers across the company to access those capabilities without duplicating expensive systems within individual teams or programs.
Those shared systems are used in several ways across the organization.
At the broadest level, the company’s Navigator tool — a generative AI chatbot — is used by more than 80,000 employees. The system includes about 20,000 AI agents performing tasks on Lockheed’s hardware and software infrastructure.
Another program, Genesis, allows users to access large language models hosted on a central cluster, allowing teams to customize AI agents for specific tasks or areas of expertise.
At the most specialized level, about 7,000 engineers use a tool called Panel to rapidly train machine learning models.
“I improve the documentation about these tools, and then I interface with all of our different users,” Scanlon said. “It’s an interesting combination of teaching and upskilling and technical troubleshooting.”
She said much of the early hesitation around AI has faded as employees increasingly look for ways to learn and adopt the technology.
“There’s a desire to move forward as fast as possible,” she said. “We’re starting to see programs factor in the use of artificial intelligence in accelerating their timelines to delivery.”
For Lockheed Martin, building its own AI infrastructure is also important for compliance with defense contracting requirements.
“It’s really important for us to own how we build it,” Forrest explains. “Because we have to show the customer the lineage of how we built these models from start to finish, especially when we deliver into (Department of Defense) and customer contracts.”
At the Shelton AI Center, that work includes a range of testing environments, including a hardware-in-the-loop (HIL) farm used to test new hardware, and a cognitive electronic warfare lab where algorithms are evaluated.
“We’re building electronic warfare and computer vision algorithms, running across data centers in Connecticut and other locations that Lockheed Martin has. And then we are also testing them in that Connecticut facility,” Forrest said.
Real world impact
Forrest said AI is central to the company’s “21st Century Security” strategy, which focuses on integrating digital technology into Lockheed Martin’s products and strengthening its systems, including the supply chain.
Part of that strategy includes integrating AI into manufacturing. The technology is already used for several processes, including controlling CNC machining, generating written work instructions from 3D digital models and using computer vision tools to detect foreign object debris.
While the company declined to comment on how AI is being used in its current production ramp-up amid heightened global tensions, Forrest did describe its role in a recent conflict.
During missile attacks in the Red Sea by Houthi rebels a few years ago, Lockheed partnered with the U.S. Navy to develop updates for destroyers tasked with intercepting missiles and drones.
Radar operators were being overwhelmed by false alarms. Lockheed engineers used real-time radar data to train machine learning models, which were deployed back to the ship within 24 hours. The system helped filter false alarms, improving accuracy and reducing operator workload.
Forrest said development of the AI Factory has also pushed Lockheed to evolve its business model. Last year, the company launched a wholly owned subsidiary, Astris AI, allowing it to more easily sell AI Factory tools to government and commercial clients.
‘Comprehensive approach’
Defense analyst David Pring-Mill, founder of tech policy and research firm Policy2050, said the latest Middle East conflict is likely to accelerate AI adoption across the defense sector.
He said the U.S.-Iran conflict has exposed limits in the U.S. defense manufacturing base, which is designed for steady peacetime production rather than rapid increases in wartime demand.
Pring-Mill said Lockheed’s coordinated approach — developed through its AI Center — has helped give the company an edge over competitors.
“That kind of centralized AI infrastructure is what allows a company with over 120,000 employees to apply AI at scale, rather than running disconnected experiments in individual factories,” he told the Hartford Business Journal.
Lockheed, Pring-Mill said, is effectively pursuing three AI strategies at once.
In addition to the AI Center, the company has committed more than $330 million to internal digital factory capabilities — including connected factories, smart tools and augmented reality systems on production floors.
It has also partnered with AI-native startups to expand manufacturing capacity and invested venture capital in emerging AI technologies it could acquire later.
“Based on what’s publicly visible, that’s the most comprehensive approach of any defense prime right now,” Pring-Mill said.
