Twin Prime: A New Frontier AI Lab for Defense and Security

Defense and security systems increasingly rely on large volumes of sensor data—from drones, radar, electro‑optical systems, and battlefield devices. Twin Prime is a new AI startup focused on turning that data into fast, actionable decisions.

Founded in 2025, the US‑based company is building specialized multimodal frontier AI models designed for defense and national‑security environments, where reacting quickly to complex signals can be critical. The startup has already raised early funding and formed a strategic partnership with a defense manufacturer to bring its technology into operational systems.

What Twin Prime Is

Twin Prime is a frontier AI lab dedicated to defense and security applications. It was founded by George Lentzas, Stephane Sezer, Drew Calcagno, and Michael Leite‑Garcia, whose backgrounds span frontier AI research, quantitative finance, and national security institutions.

The founders collectively bring experience from organizations including Google Research, Hudson River Trading, Lawrence Livermore National Laboratory, Columbia University, the White House, the Pentagon, and U.S. and European armed forces, reflecting a mix of technical AI expertise and operational security knowledge.

The company’s core focus is developing AI systems designed for high‑stakes operational environments, where accuracy, speed, and the ability to synthesize multiple data sources are essential.

Multimodal AI Models for Sensor Fusion

Twin Prime’s central technology is a class of multimodal AI models that can reason across many types of physical‑world sensor data at once.

Rather than analyzing one data stream at a time, these models are designed to ingest and interpret signals from multiple sensor modalities—such as imaging systems, environmental sensors, or other surveillance inputs—and combine them into a unified understanding of a situation.

The goal is to compress what the company describes as the “perception‑to‑decision” cycle: the time between detecting a signal and taking action. By shortening this loop, operators can respond to potential threats faster and with more contextual information.

This capability is particularly important in defense settings, where modern systems generate vast streams of data but human operators must still make rapid decisions. AI models that can fuse and interpret that information in real time may significantly improve situational awareness.

Funding: $10 Million Pre‑Seed Round

Twin Prime launched with $10 million in pre‑seed funding, led by the defense‑focused venture firm Expeditions.

The round also included participation from:

• American and European venture capital firms
• Defense manufacturer Theon
• Family offices
• Angel investors affiliated with companies such as Palantir, Anduril, and Quorum

The funding will support development of the company’s multimodal AI models and their deployment in national‑security and defense use cases.

Strategic Investment and Joint Venture With Theon

Alongside the venture funding, Theon International Plc, a defense technology company, announced a $3 million strategic investment in Twin Prime for a single‑digit minority equity stake.

The two companies also plan to form a joint venture incorporated in Greece, structured with:

• 60% ownership by Theon
• 40% ownership by Twin Prime

The joint venture will develop, commercialize, and deploy AI solutions based on Twin Prime’s platform across Theon’s defense product portfolio.

Why the Partnership Matters

The partnership provides a path for Twin Prime’s AI to move from research into real defense systems.

Theon produces a range of defense technologies, including electro‑optical devices and soldier‑system equipment. Integrating Twin Prime’s AI into those platforms could allow sensor data from deployed equipment to be processed and interpreted in real time.

For Theon, the collaboration is intended to accelerate its internal AI development efforts as part of its Theon Next innovation initiative. For Twin Prime, it offers access to established defense products and customers—an important step for any startup aiming to deploy AI in operational military environments.

The Bigger Picture: AI for High‑Stakes Decisions

Twin Prime represents a growing wave of startups building AI systems specifically for defense and national‑security applications, rather than adapting general‑purpose AI models.

By focusing on multimodal sensor fusion and rapid decision‑making, the company is targeting one of the most challenging problems in modern security: converting overwhelming streams of real‑world data into clear, timely actions.

If successful, such systems could reshape how defense organizations process information, enabling faster and more coordinated responses to complex threats.