Cognee, a Berlin-based AI infrastructure company, today announced a €7.5 million funding round to accelerate the development of its structured memory layer for AI systems and agents.
The round was led by Pebblebed, with participation from 42CAP as a follow-on investor. Existing and new angel investors also participated.
“AI systems today don’t fail because they aren’t powerful enough,” says Vasilije Markovic, Founder and CEO of Cognee. “They fail because they don’t remember. We’re building the memory layer that allows AI to understand context, not just retrieve text.”
Recent reporting by EU-Startups shows continued investor appetite for AI infrastructure and adjacent layers across Europe in 2025–2026.
In the UK, SurrealDB secured €19 million to scale its multi-model database for AI applications, while London-based Overmind raised €2.3 million to develop a supervision and security layer for AI agents, and Toyo attracted €3.6 million to build secure AI agent workflows. In Germany, Offenburg-based happyhotel raised €6.5 million to develop AI agents for hotel revenue management, providing a same-country comparison point for Berlin-based Cognee.
At a larger scale, Finland’s DataCrunch secured €55 million to expand AI cloud infrastructure aimed at strengthening Europe’s compute capacity.
Together, these disclosed rounds represent approximately €86.4 million in funding across AI databases, agent supervision and security, workflow orchestration, vertical AI agents and cloud infrastructure.
Within this broader 2025–2026 context, Cognee’s €7.5 million raise positions it among a cohort of European startups focused on foundational AI layers, as capital continues to flow into infrastructure that supports production-grade deployment rather than solely end-user applications.
“Europe has a strong tradition in building deep infrastructure,” Vasilije adds. “Our ambition is to contribute a core building block to the global AI stack – not another application, but infrastructure that many AI systems can rely on.”
Founded in 2024, Cognee is building core AI infrastructure designed for a global developer and enterprise audience. Its technology aims to address one of the most fundamental limitations of today’s AI systems: the lack of long-term, structured memory.
The company explains that AI systems rely on stateless approaches such as file-based retrieval or short-term context windows. While effective for simple use cases, these methods break down as workflows become more complex, long-running, and business-critical.
Cognee addresses this gap by transforming unstructured data into a persistent, structured memory layer, built on knowledge graphs and semantic representations. This enables AI systems and agents to retain context over time, reason across connected information, and significantly reduce hallucinations in production environments.
Cognee originated as an open-source project and has seen rapid adoption within the global developer community. Thousands of engineers already use Cognee to build and operate AI-native applications that require reliable memory and context handling.
Today, more than 70 companies are running Cognee in live environments, particularly in knowledge-intensive and regulated domains.
The new funding will be used to:
- further develop Cognee’s core memory and graph technology
- expand enterprise-grade capabilities and reliability features
- support growing commercial demand as AI systems move from experimentation into production
While Cognee’s roots are in Berlin, the company is focused on building foundational AI infrastructure for a global market. The team operates across Europe and the United States and is working closely with developers and enterprises worldwide.
Read the orginal article: https://www.eu-startups.com/2026/02/german-ai-infrastructure-startup-cognee-lands-e7-5-million-to-scale-enterprise-grade-memory-technology/
