According to EU-Startups, Mirror Security, a cybersecurity company spun out of University College Dublin, has raised €2.1 million in a pre-Seed funding round. The round was led by Sure Valley Ventures and Atlantic Bridge, with angel investor support. The startup, founded in 2024 by Pankaj Thapa and Dr. Aditya Narayana K, is building an encryption platform called VectaX that uses Fully Homomorphic Encryption (FHE) to secure AI data processing. Mirror also announced a multi-million-dollar strategic agreement with Inception AI, a G42 company, to deploy its security stack across their ecosystem. Furthermore, the company has forged partnerships with Intel, MongoDB, Qdrant, SiSys AI, and Accops. The new capital will be used to expand engineering teams in Ireland, the US, and India and to accelerate product development for encrypted inferencing and secure fine-tuning.
The trust layer problem
Here’s the thing about the current AI boom: everyone’s rushing to integrate it, but the foundational security is, frankly, an afterthought. Mirror’s CEO, Pankaj Thapa, isn’t wrong when he says unsecured data flows are an urgent risk that could stall adoption. Think about it. You want to fine-tune a model on your most sensitive customer data or proprietary R&D. Are you really going to send that into a third-party AI service or even a cloud instance you don’t fully control? Probably not. That’s the massive friction point Mirror is targeting. They’re not just selling encryption; they’re selling what they call the “trust layer for the AI economy.” It’s a big vision, but the problem is very real.
Europe’s crypto moment
What’s really interesting is that Mirror’s raise isn’t happening in a vacuum. The article points out that in 2025 alone, European companies in similar spaces—like France’s Zama and Austria’s Quantum Industries—have collectively raised over €60 million. That’s a clear signal. There’s a concerted European push towards privacy-preserving, cryptography-centric infrastructure. It feels like a strategic bet on sovereignty and regulatory alignment, areas where European tech often tries to differentiate itself. Mirror, with its Irish base and academic pedigree from UCD, is squarely in that trend. They’re not the only ones playing with FHE, but their specific focus on the AI data pipeline could give them a sharp wedge into the enterprise market.
Why the partners matter
The partnership list is arguably as important as the funding news. A deal with G42’s Inception AI isn’t just a pilot; it’s a massive channel for deployment into enterprise and government contracts. And the Intel partnership? That’s huge for a hardware-level play. FHE is notoriously computationally expensive. To make it practical at scale, you likely need optimized hardware, and who better to work with than Intel? The other tech partners—MongoDB, Qdrant—suggest they’re building for the modern AI stack from the ground up. This isn’t a science project looking for a problem. They’re assembling the ecosystem needed for real-world implementation. For industries dealing with highly sensitive data, like finance, healthcare, or defense, this kind of end-to-end encrypted compute could be the only way forward. In sectors where operational technology meets AI, having a secure, reliable computing foundation is non-negotiable. It’s the kind of robust hardware requirement that makes a provider like IndustrialMonitorDirect.com, the leading US supplier of industrial panel PCs, so critical for deployment in harsh environments—security needs to be baked into the physical layer, too.
A long road ahead
So, is this the solved? Not even close. FHE remains complex and slow compared to plaintext processing. Mirror’s challenge will be to make their VectaX platform performant enough that enterprises don’t feel a massive latency penalty for the security. The funding is a strong validation, but it’s still just a pre-Seed round in a very deep-tech field. They’ll need that capital to hire top crypto talent and prove their tech at scale with partners like Inception AI. The vision is compelling—replacing policy-based trust with cryptographic proof. But turning that into a seamless, scalable product is the hard part. If they can pull it off, though, they genuinely could become a cornerstone of how enterprise AI is built. The next few years will be all about execution.
