One Database for Two Jobs
Databases have traditionally been built for one of two jobs. Analytical databases help businesses spot trends and answer questions, while operational databases store the day-to-day records of transactions, such as orders and payments. But Munich-based database engine CedarDB supports both โ an approach that has gained relevance with the rise of agentic AI, cofounder and CEO Moritz Sichert told TechCrunch.
CedarDB has been a decade in the making, but its current momentum stems from enterprise companies whose existing data stacks may have been well suited to storing years of transactions, yet less suited to quickly answering analytical questions about that data posed in natural language.
Built on Umbra, Tuned for Modern Hardware
CedarDB's questioning of the divide between analytical and operational queries is a byproduct of its focus on efficiency. The company is a Technical University of Munich spinout based on Umbra, a research project built from scratch to beat legacy database engines on processing speed by making the most of modern hardware.
With five Umbra researchers among its cofounders, CedarDB kept that approach, promising faster performance without requiring more CPUs, processing power, or memory โ and requiring few changes from customers. In particular, the team made sure CedarDB was compatible with PostgreSQL, the popular open-source database system.
According to Sichert, this compatibility helps keep switching costs low, since users don't need to rebuild their applications. It carries an added advantage: CedarDB follows the logic that LLMs are most familiar with as a result of their training.
From Academia to Production
That decision is one of many choices that proved lucky in retrospect, Sichert said. But it wasn't all luck. In a blog post, cofounder Christian Winter detailed key priorities the team set after leaving academia, "such as reliability, fault tolerance and compatibility with popular tools."
Not all of the answers were obvious. The team strongly debated whether it should launch a cloud offering right away. In the end, it launched CedarDB Enterprise as an on-premise offering alongside its free version, CedarDB Community Edition, both announced in May of last year, alongside a โฌ5.3 million seed funding round (approximately $6 million).
Cloud Arrives โ and the Competition Heats Up
The cloud offering was only delayed, not abandoned. After conducting beta testing with users who had signed up for its waitlist, the company announced limited availability of CedarDB Cloud on Wednesday, on stage at TechCrunch Disrupt, where it is competing with 20 other startups vying to win TC's prestigious Startup Battlefield competition.
A fully managed, cloud-hosted version will likely help CedarDB expand its user base beyond large enterprises. It will also make it better prepared to compete with the likes of Databricks, a major data company that seems to have reached the same conclusion as the German startup.
At its Data + AI Summit in June, the American company cited "the agentic era" as the reason it was launching LTAP, a new data architecture aiming to "[unify] transactions, analytics, streaming, and operational data on a single copy of storage in the lake."
Why It Matters in 2026
CedarDB's hybrid transactional/analytical (HTAP) pitch has found fresh relevance as AI agents increasingly query enterprise data on users' behalf. Agents need to both read live operational records and run analytical queries in near real time โ a workload that a single unified engine handles far more gracefully than a stack stitched together from separate OLTP and OLAP systems. With the arrival of CedarDB Cloud and growing competition from established players like Databricks, 2026 shapes up to be a pivotal year for the Munich spinout.
via TechCrunch