Lockheed’s AI Factory: Turning a “Digital Thread” Into a Battle-Ready Advantage

ORLANDO — A new AI platform that provides defense-grade security across environments is now available from Lockheed Martin, which has created a subsidiary to manage it.

The subsidiary, Astris AI, provides the company’s AI Factory, offering an all-in-one operating model for all users. Companies or agencies that use it get access to the full arsenal of AI components available at the nation’s largest defense contractor, according to Greg Forrest, Lockheed Martin Vice President of AI Foundations and Commercialization. 

The platform, which can process massive amounts of data securely, fills a void for security across defense, aerospace, finance, healthcare, energy, and manufacturing industries. Forrest described the platform in a HUB session at AIAA SciTech Forum 2026 in January. 

While Lockheed has used its AI networks internally for years, Astris AI makes the same tools available to a wide range of users. Forrest explained, “There are many disparate tactical networks across the government. One of the big advantages of the AI factory is that we have the programs and the access to those networks to be able to go in and give them a really high level of AI and machine learning. We can go deep into those tactical networks to provide capability at the edge.”

Astris AI combines all Lockheed’s AI agents and machine learning operations under one tent. It has available a suite of AI solutions, and customers can pick and choose modules that fit their needs, Forrest said. 

Together, these pieces form a self-contained ecosystem that can ingest raw data, curate and label it, train models, validate them against tests, and monitor drift once the models are deployed. 

Forrest illustrated the AI Factory’s impact for the AIAA audience with case studies.

1. Orion Space Capsule – As the crewed vehicle for NASA’s Artemis II mission prepares for launch, Lockheed is using AI to accelerate processes: natural language processing parses design documents and operation manuals, surfacing actionable items that shave weeks off the build schedule. Further, an AI-generated suite of tests scans newly written code for defects, catching bugs before the capsule ever leaves the ground.
2. Joint Simulation Environment – The U.S. Air Force and Navy jointly operate a high-fidelity virtual battlefield where new fighter jet capabilities are vetted. Rather than wade through dense manuals, operators now converse with a knowledge agent that understands the simulation’s physics and can instantly spin up test scenarios.
3. Radar False Alarm Reduction – In the Red Sea, the U.S. Navy and Lockheed Martin used AI to develop and field updates for destroyers shooting down missiles and drones via a process called Aegis Speed To Capability. Lockheed software helped operators filter false alarms, reducing operator workload on naval ships. 
4. Edge retraining for vision models on a small unmanned aerial system – When a forward deployed customer needed to identify objects hidden behind terrain, engineers used the AI Factory to “see” beyond line of sight and relay actionable intel to ground operators.

The success of these pilots revealed a market gap, Forrest noted, and Astris AI was born. By exporting the same modular stack that powers, say, new Aegis radar algorithms, the company sells a compliant, classified‑ready solution without reinventing the wheel for each client. Its benefit is threefold: It’s modular, scalable, and secure.

In December, the company launched Astris AI for Government™ to aid in operationalizing America’s AI Action Plan through an integrated, turnkey AI platform that enables government agencies to build, deploy, and sustain trusted AI solutions. Teaming with Oracle, NVIDIA, and Meta, these solutions are now available through commercial federal procurement vehicles. 

X-Bow became the first to deploy Lockheed Martin’s secure AI for rocket production, and this week, Albany Engineered Composites announced a deal with Astris AI to accelerate AI adoption across its advanced aerospace manufacturing operations. 

The Astris AI Factory platform can be deployed on the HPE ProLiant Compute DL380a, an AI fine-tuning and inferencing platform, or on the HPE Edgeline EL8000, an edge server optimized for high-speed data analytics and AI/ML capable computing.

Forrest emphasized that technology alone does not deliver value. “The best tool in the world is useless if engineers don’t know how to wield it,” he said, outlining an internal consulting team that conducts workshops within program offices. This ecosystem ensures that AI tools are not just deployed but adopted – which can translate to increased productivity.

Forrest showed the audience that a digital thread could be expanded to span requirements, design, test, and sustainment – through a modular, open source, and customer-friendly AI Factory.