Stanhope AI announces $8 million seed funding, sets sights on edge AI revolution

Stanhope AI is a spin-out from the University College London and is developing a new form of AI that could revolutionise AI for edge applications like drones and autonomous vehicles. This funding round positions the company to start partnering in defence. 

By Sam Cranny-Evans, published on February 13, 2026. 

Stanhope AI has announced the successful closing of its seed funding round at $8 million (£6.15 million/€7.43 million), according to a February 12 press release. The funding round was led by Frontline Ventures, with participation from Paladin Capital Fund, Auxxo Female Catalyst Fund, and follow-on investment from UCL Technology Fund and MMC Ventures. 

The funding will help Stanhope AI to advance what it calls its Real World Model, described as: “A next-generation framework for adaptive intelligence, designed to function in dynamic, physical environments beyond the limitations of large language models (LLMs).”

The core of Stanhope AI’s offering is far more adaptive artificial intelligence that is better at responding to its environment. “We’re moving from language-based AI to intelligence that possesses the ability to act to understand its world – a system with a fundamental agency,” said Professor Rosalyn Moran, CEO and Co-founder of Stanhope AI.

What does this mean? Well, in many cases where AI is used for defence (or other complex things), it is a combination of models. This means there might be one set of algorithms for processing imagery, another for controlling flight, and a third for communicating with other systems. Over or under all of that could be another set of models for combining their outputs into a central understanding. The models don’t really understand the world, but are providing the user with the outputs they have been trained to produce. 

Stanhope AI appears to be positioning its AI as something much more intelligent. “Our approach doesn’t just process words, it understands context, uncertainty, and physical reality,” Professor Moran said. The company is already trialling its AI in “autonomous drone and robotics applications with international partners, teaching machines to behave more intelligently in unpredictable, real-world environments,” the press release states. 

Stanhope AI and the AI revolution

Ok, so why is this a revolution of edge AI? Well, according to Stanhope’s press release, there is a move towards AI systems being used at the edge. This is in reference to cloud-based AI and LLMs like Google Gemini. Every time you use Gemini or ChatGPT, you are sending requests to models running on servers somewhere else. If you had edge AI, it would be running directly on your computer. But this is challenging, AI models can be very big and take up a lot of power, both in terms of electricity and computing power. 

Techniques like “minifying” (yes it’s a real word) focus on squeezing the algorithms and making them smaller so they can be used on smaller chips. This is part of what Stanhope AI is referring to, stating that its “lean, explainable models can run efficiently on-device, using minimal data and energy. This enables deployment in autonomous systems, defence technology, industrial automation, and embedded devices, where efficiency and reliability are mission critical.” 

Critically, the company has developed a  brain-inspired paradigm for its models, known as ‘Active Inference.’ The press release states that it “allows machines to learn and adapt on the fly, which is a crucial capability missing from LLM-based systems that rely on large static data sets.” It is worth noting that updates to models can be provided over the air, whilst they are in service, but the update would have to be trained somewhere else. 

Company profile: Stanhope AI

Stanhope AI is a London-based deep-tech startup. It is a spin-out from University College London and King’s College London. Neuroscientists Professor Rosalyn Moran and Professor Karl Friston co-founded the firm in 2023. The team pioneers a new model of artificial intelligence.

They draw from the ‘Free Energy Principle’. This framework explains how intelligent systems minimise uncertainty. Consequently, the company develops AI for autonomous systems. These machines mimic the human brain. This paradigm is known as ‘Active Inference’. It allows machines to learn and adapt on the fly. This capability is missing from LLM-based systems.

In March 2024, the company secured £2.3 million ($3.00 million/€2.83 million) in funding. The UCL Technology Fund led this investment round. Other participants included Creator Fund and MMC Ventures. The team will use the capital to advance ‘agentic’ AI. These models decode environments using real-time data from onboard sensors. Currently, partners like the Royal Navy test the technology but it also has potential in manufacturing and industrial robotics.

Calibre comment: The balance

If you are read-up on AI, you will know that running algorithms at the tactical edge is far from new. There are drones like the Bumblebee and Lancet that notionally have the ability to run machine vision at the edge. And systems like the Nvidia Jetson are designed for edge AI applications. The distinction largely appears to be in terms of where the algorithms are trained and how. 

Stanhope AI’s position is that its algorithms would have the ability to learn whilst on the job. See a new type of Russian tank – no problem, the AI will classify it itself and remember. The current standard would require a new round of training on that data to help the models recognise the tank. Presumably Stanhope AI’s models also need foundational training on existing static data sets, but the follow-on learning is enabled by operational experience. 

If this can be deployed safely and reliably, it could have some benefits for defence. But when AI is being considered as a tool for tracking enemy nuclear submarines, or shooting down aircraft, there is very little room for error. So, it is reasonable to assume that there will be interest in what Stanhope is offering, but there will be caution too. 

Interested in other applications of AI to defence? Look no further, we’ve got a few links below that you might find useful. And as always, thanks for reading and sharing these articles, it really supports Calibre Defence: 

• Swarmbotics AI: Why the US Army is betting on robotic mass – Calibre Defence 
• Arexis: AI powered EW en route for German Eurofighters – Calibre Defence 
• Russia’s V2U autonomous drone is striking targets in Ukraine – Calibre Defence 
• Bumblebee drone to bolster US counter-UAS capabilities – Calibre Defence 

The lead image shows a drone that is understood to be running models from Stanhope AI. Credit: Stanhope AI.