Arlequin AI Raises €28 Million to Build a European Alternative to LLM-Based Intelligence

Paris-based Arlequin AI has raised €28 million in Series A funding to scale a fundamentally different approach to artificial intelligence—one designed to analyse complex data without relying on large language models (LLMs).

The round was jointly led by redalpine and OTB Ventures, with participation from Bpifrance’s Defence Innovation Fund and French entrepreneur Xavier Niel. Existing investors Vsquared Ventures and 10x Founders also increased their stakes. The latest financing takes Arlequin’s total funding to €32.4 million, following its €4.4 million seed round in 2025.

Building AI Without Depending on LLMs

While much of the current AI industry is centred around increasingly large language models, Arlequin is taking a different route.

The company has developed topological neural network (TNN) technology, designed to identify relationships and connections hidden within huge volumes of data rather than relying primarily on language patterns.

Its systems can analyse text, images, audio, video and data extracted from seized devices, helping users discover connections and patterns that can then be traced and audited. Arlequin does use an LLM, but only for its conversational interface—to help users formulate questions. The underlying analysis is performed by its own unsupervised models.

The company's argument is particularly relevant for sensitive sectors where an AI mistake can have serious consequences. In areas such as counterterrorism, security and energy, a plausible but incorrect answer can be far more damaging than a simple productivity error.

From Academic Research to a Deeptech Startup

Arlequin's story began with Hugo Micheron, whose academic work focused on radicalisation, jihadism and the formation of extremist networks.

Micheron learned Arabic in Syria and conducted extensive field research on European jihadism, including interviews with hundreds of convicted terrorists. During his time as a lecturer at Princeton University from 2020 to 2023, he became increasingly concerned about the limitations and biases of the AI tools available for analysing complex information.

That experience eventually led him to create Arlequin and pursue a different architecture for AI.

Meet the Founders

Hugo Micheron — Co-Founder & CEO

Hugo Micheron is the co-founder and CEO of Arlequin AI. His background is unusual for a technology founder, combining deep academic research, geopolitics and fieldwork.

His research into European jihadism involved extensive work with primary sources and interviews with convicted terrorists. He later brought that experience into the technology world, with the goal of developing AI capable of finding reliable and explainable patterns in complex information.

Micheron's central belief is that critical decisions require AI systems whose conclusions can be understood, tested and audited, rather than simply accepted because a model produces a highly probable answer.

Antoine Jardin — Co-Founder & CTO

Antoine Jardin co-founded Arlequin with Micheron. The two have been friends for around 15 years.

Before Arlequin, Jardin worked as a research engineer at CNRS, France's national scientific research organisation. His expertise includes big data and dimensionality reduction, a mathematical technique that sits at the core of Arlequin's AI models.

He was also involved with Jean Zay, France's national AI supercomputer, giving him extensive experience in large-scale computational research.

Together, Micheron's expertise in geopolitics and complex social systems and Jardin's technical background created the foundation for Arlequin's alternative approach to AI.

A Platform Designed for High-Stakes Decisions

Arlequin is targeting areas where understanding relationships within large datasets can make a significant difference.

Its technology can help organisations analyse massive and unstructured datasets, identify previously hidden connections and produce intelligence that can be traced back to the underlying information.

The company's potential applications extend beyond defence and security into areas such as:

• Counterterrorism and threat intelligence
• Defence and national security
• Banking fraud detection
• Corporate due diligence
• Risk analysis
• Strategic intelligence
• Complex data analysis

Arlequin already has more than 30 clients across four European countries, demonstrating that its technology is moving beyond research and into commercial use.

A 50-Person Team With a Heavy Research Focus

Just two years after its founding, Arlequin has grown to approximately 50 employees.

The team includes more than 35 engineers and around 15 PhDs and postdoctoral researchers, reflecting the highly technical nature of the company's work.

This research-heavy structure is important because Arlequin is operating in a relatively young field. According to Micheron, only around 20 researchers worldwide are currently working specifically on topological neural network approaches.

The company believes it is among the first to turn this research direction into a commercial product.

Why Europe Is Betting on a Different AI Architecture

Arlequin sees its technology as part of a broader European push for AI sovereignty.

The US currently has major advantages in computing infrastructure and access to data, while China is developing its own AI ecosystem. European companies such as Mistral are building competitive foundation models, but the continent still faces a significant scale disadvantage.

Arlequin's strategy is therefore not to simply compete with the US and China by building ever-larger models. Instead, it wants to create a different technological path that could deliver powerful analysis without requiring the same level of computing resources.

The company's investors see this as an important opportunity. OTB Ventures has described Arlequin's approach as a way for Europe to build AI systems it can own, understand and trust, particularly in sensitive areas.

Strong Investor Backing

The €28 million Series A was jointly led by redalpine and OTB Ventures.

Bpifrance's Defence Innovation Fund also participated, while existing investors Vsquared Ventures and 10x Founders increased their investments. Xavier Niel joined the company as a new investor.

Redalpine, which has backed companies including N26, Mistral and Klarna, said Arlequin stood out because of its combination of technical differentiation, founder expertise and focus on a major security challenge.

What Arlequin Plans to Do With the Funding

The new capital will help Arlequin accelerate the development and commercialisation of its AI architecture while expanding its European footprint.

The company plans to:

• Continue expanding its operations in France
• Grow its presence in Germany
• Establish a new office in London
• Increase its engineering and research capabilities
• Expand its platform across defence, security and other high-value sectors
• Further develop its topological neural network technology

Arlequin has already begun building its London presence, with its first employees starting there in September 2026.

A Growing Defence AI Market

The opportunity is substantial. The European AI and analytics market for defence applications is projected to grow from approximately $4.8 billion in 2026 to $19.27 billion by 2035, according to figures cited by Tech Funding News.

However, Arlequin is not limiting itself to defence. Its technology could also be applied to banking fraud, due diligence, corporate intelligence and other situations where organisations need to make decisions based on large, complicated datasets.

The Bigger Bet on AI

Arlequin's funding round represents a different kind of AI bet.

Instead of believing that bigger models and more computing power will solve every AI problem, the company is exploring whether a different architecture can provide more transparent and reliable intelligence for high-stakes decisions.

That approach is still relatively young, and the company faces significant competition from established AI and defence technology players. But with €28 million in new capital, a team of around 50 people and more than 30 European clients, Arlequin now has the resources to test whether its alternative approach can scale.

If successful, the company could help establish a new category of AI—one focused less on generating convincing answers and more on understanding complex relationships, uncovering hidden patterns and producing intelligence that people can verify.