NestAI Eyes European Defense Autonomy with New AI Models

NestAI, a military-focused AI lab founded by Peter Sarlin in 2025, has unveiled its inaugural models designed specifically for defense applications. This initiative aims to empower Europe to decrease its reliance on foreign technology providers for military capabilities.

Before founding NestAI, Sarlin established Silo, an AI lab that became notable in Europe after being acquired by AMD in 2024, marking one of the largest AI acquisitions on the continent. Since its inception, NestAI has rapidly scaled to a team of 200, attracting talent from renowned companies such as Tesla, Nokia, and Intel. The company secured €100 million in funding from Nokia and the Finnish investment firm Tesi last November.

NestAI is developing two primary functionalities: one focusing on autonomy for drones, allowing them to operate effectively in various scenarios, and another for battlefield orchestration, which enables military forces to manage entire operations through its NestOS platform. The training datasets for these models blend synthetic data with real-world information, increasing their operational relevance.

The urgency of NestAI’s mission became especially clear following a recent US government restriction on exporting advanced models from Anthropic, highlighting Europe’s dependence on external tech providers. Sarlin emphasizes the critical need for European sovereignty in controlling the foundational models that enhance military operational capabilities. “There’s been significant concern about owning and controlling the model layer in defense,” he remarks, adding, “this is basically a solution to that.”

According to Sarlin, NestAI does not intend to compete against general purpose models like those from Anthropic or OpenAI. Instead, the strategy hinges on creating domain-specific models tailored to defense requirements, a choice he believes will enhance their competitiveness. He draws a parallel between their technology and self-driving cars, noting how environmental changes require adaptability—an issue magnified on a battlefield where conditions can shift dramatically due to adversarial actions.

“If your adversary aggressively bombed the battlefield, the maps you once relied upon for autonomy could become obsolete,” Sarlin explains. “We optimize for a foundational model that adapts to ever-changing conditions.”

Ongoing Military Pilots

The company is currently engaged in pilot programs with the armed forces of Estonia and Finland. With the foundational drone model, these military bodies can initiate and complete missions autonomously using fleets of drones. Sarlin notes that the orchestration model allows them to comprehensively plan and execute missions involving drones tailored to specific tasks.

Looking ahead, NestAI plans to collaborate with military organizations in other allied nations, utilizing the data generated from these engagements to refine their models continuously. Sarlin remarks that model development is an ongoing process, stating, “It's something that starts and never really ends.”

The Urgency of Modern Warfare

The landscape of ground warfare has noticeably evolved since the onset of the Ukraine conflict. According to reports from the Ukrainian Air Force, 2024 witnessed around 13,300 air attacks from Russia—encompassing drones, missiles, and cruise missiles—targeting civilians. By 2025, that figure escalated to approximately 56,700, largely due to an uptick in drone deployments.

With such rapid technological advancements on the battlefield, Sarlin indicates that European nations must embrace “war-time capability” to adapt effectively to new demands. He points out that many national defense systems still operate under “peace-time capability,” as few countries are experiencing immediate wartime needs. “We’re optimizing for the Ukraine model, focusing on how they operate and the speed at which they can adapt,” he states.

Sarlin further emphasizes the need for a framework that enables defense R&D to be directly linked to battlefield experiences. The challenge for military units, he says, is the pace at which they can adjust as adversarial tactics evolve.

Strategic Partnerships

In its mission, NestAI has established partnerships that enhance its tech capabilities. It’s collaborating with AMD to increase its computational resources and is working with Finland’s LUMI AI factory to utilize its supercomputer for model training. Another significant collaboration exists with Qutwo, a quantum startup where Sarlin is a co-founder. Qutwo is developing methods to simulate quantum computing on GPU clusters, facilitating the compression of large AI models while maintaining performance, thus enabling deployment on edge devices. “This is key to delivering capable AI on affordable hardware,” Sarlin concludes.