Lockheed Taps OpenAI to Solve F-35 Challenges

Lockheed Martin is deploying 55 different large language models across internal operations and defense applications, including autonomous weapons systems and crewed-uncrewed teaming. OpenAI is collaborating with the F-35 program on complex mathematics and physics challenges related to advanced sensor capabilities. The model-agnostic AI strategy highlights expanding defense-sector adoption of generative AI, though no financial impact or contract value was disclosed.

Analysis

The investable issue is not enterprise-AI adoption; it is whether Lockheed can convert AI-assisted engineering into shorter development cycles, lower rework, and sustained program-margin expansion without creating cyber, export-control, or mission-assurance liabilities. Near term, this is unlikely to move FY guidance because defense contract economics generally pass labor productivity gains to the customer over time through recompetes and negotiated rates. The first financial proof points should instead be lower bid-cycle costs, improved EAC performance on fixed-price development programs, and stronger win rates on classified autonomy and sensor work over the next 1-3 quarters.

The more material 6-18 month effect is competitive differentiation in next-generation air and missile-defense architectures, where software iteration speed increasingly determines platform relevance. LMT could gain share versus RTX and NOC if it demonstrates credible crewed-uncrewed teaming and sensor-fusion capability, but the same model-agnostic strategy risks integration complexity and fragmented security controls. Suppliers with exposure to edge compute, radiation-tolerant processing, secure communications, and autonomy payloads—not generic AI infrastructure—are better second-order beneficiaries; relevant public proxies include AVAV, KTOS, LHX and CRDO.

Consensus may overvalue the AI narrative as a standalone multiple catalyst. Defense primes trade primarily on backlog conversion, appropriations, program execution and cash return; absent disclosed contract awards, milestone acceleration, or measurable margin improvement, this is reputationally positive rather than earnings-changing. A public failure involving hallucination, data leakage, model provenance, or autonomous-system safety could rapidly turn this from a valuation support into a procurement and oversight risk.

Key Decisions for Investors

• Maintain, but do not add aggressively to, LMT on this item alone. Treat it as a 6-18 month execution watch: add only if quarterly commentary shows engineering-hours reduction, improved program EACs, or autonomy/sensor contract wins; otherwise AI claims should not justify multiple expansion.
• Prefer a 3-6 month relative-value expression: long LMT / short RTX in equal dollar amounts if LMT demonstrates defense-AI contract conversion. LMT has greater upside from air-platform and mission-system software differentiation, while RTX retains relatively greater exposure to commercial-aerospace cycle normalization; exit if LMT program-margin guidance falls or RTX wins the relevant autonomy/sensor awards.
• Build a watchlist for AVAV and KTOS rather than chasing LMT: evidence that crewed-uncrewed teaming moves from demonstrations to funded programs would be a more direct revenue catalyst for autonomous-platform providers. Require named program funding or backlog disclosure before entry, given elevated small/mid-cap defense valuation sensitivity.
• Set a downside alert around cybersecurity, export-control, and Pentagon AI-assurance developments over the next 90 days. Any mandated model restrictions, data-governance incident, or autonomous-weapons oversight action would falsify the near-term differentiation thesis and likely favor diversified incumbents NOC and RTX over LMT.