Arlequin AI’s €28M Series A: Building Topological Neural Networks for High-Stakes Decisions

Arlequin AI has raised a €28 million Series A to build proprietary topological neural networks (TNNs), expand its deployments beyond Europe and grow the research and commercial teams behind its evidence-led analytics platform. The Paris company is targeting decisions where analysing isolated facts is not enough: investigators and operators also need to understand the relationships among people, events, locations, transactions and other evidence. 12

What the €28 million will fund

The Series A was co-led by redalpine and OTB Ventures. Bpifrance’s Defence Innovation Fund participated, while existing investors Vsquared Ventures and 10x Founders increased their backing; Xavier Niel also joined the round. Arlequin says the financing is exclusively European and is intended to accelerate TNN development and international deployment. 1210

The company’s immediate priorities are:

• training and developing its proprietary TNN models;
• expanding the team developing and operating those models;
• growing commercial deployments across Europe and internationally; and
• building out its international presence, including offices in London and Berlin and a planned Silicon Valley AI lab by the end of 2026. 14

Arlequin reported a team of about 50 people, including 35 engineers and 15 PhDs, researchers and data scientists, at the time of the funding announcement. 1

What Arlequin means by topological neural networks

Arlequin’s platform is designed to work with fragmented and heterogeneous information, including documents, transactions, video and operational data. Its proposed TNN architecture is intended to learn from both individual data points and the structures that connect them, including relationships involving multiple elements at once. 1810

That focus is central to the company’s pitch. Rather than approaching a case as a collection of separate records, the system is meant to identify patterns across connected evidence—such as links among people, communications, locations and events. This is particularly relevant when the relationships themselves are material to an investigation or operational decision. 110

Why the company is pursuing an alternative AI architecture

Arlequin argues that simply scaling conventional AI models with more parameters, data and compute is not necessarily the best approach for analysing large, interconnected bodies of evidence. It says TNNs can expose complex, multi-element dynamics while retaining traceability to source material and using less compute. 110

Those advantages should be read as company and investor claims, not as independently validated benchmarks. Publicly available materials describe the architecture and intended benefits, but do not establish comparative performance against other graph analytics, neural-network architectures or large language models. 11014

The decision-making use cases Arlequin is targeting

Arlequin was founded in 2024 by Hugo Micheron, a researcher of jihadism and geopolitical instability, and Antoine Jardin, a former CNRS research engineer focused on data science and human behaviour. The company was built for governments and large organisations making consequential decisions from incomplete, dispersed information. 19

Its stated applications include:

• counterterrorism, security and defence;
• criminal and judicial investigations;
• fraud detection and anti-money-laundering work;
• cybersecurity and information integrity; and
• AI safety and security. 1

One example cited by the company is the ability to connect people, locations, communications and events across millions of data points collected from seized devices. The broader commercial opportunity extends to organisations that need explainable analysis for financial crime, cyber risk and complex operational intelligence. 1

Where the technology is already deployed

Arlequin says its technology is already used by governments and large organisations in Western and Eastern Europe, with activity in four European countries. The company has not publicly detailed every customer or deployment, so the scope and results of those implementations cannot be independently assessed from the announced materials. 1

Its expansion plans point to a wider market: public-sector users dealing with security and defence issues, as well as enterprises handling complex financial, cyber and integrity risks. 113

The larger European AI strategy

The funding round gives Arlequin resources to pursue a distinctly European positioning: an AI architecture owned and developed in Europe for sensitive, high-stakes analysis. The investor group—including Bpifrance’s Defence Innovation Fund—also reflects the strategic relevance of tools for intelligence, security and critical decision-making. 1210

The key test now is execution. Arlequin must demonstrate that its proprietary TNNs provide measurable advantages in accuracy, traceability, deployment cost or operational usefulness over established graph-analysis and AI approaches. Its €28 million raise finances that next stage of model development, hiring and international commercialisation. 114