CedarDB launches managed cloud database for combined transactional and analytical AI workloads

CedarDB Cloud arrives in limited availability

CedarDB, a database startup based in Munich, has taken the wraps off a managed, cloud-hosted version of its database engine, announcing limited availability of CedarDB Cloud on stage at TechCrunch's Disrupt event, where it is competing against 20 other startups in the Startup Battlefield contest. According to TechCrunch, the rollout follows beta testing with users who had joined the company's waitlist.

The cloud product joins a portfolio that went public in May 2025, when CedarDB announced its paid on-premises edition, CedarDB Enterprise, alongside a free Community Edition and a €5.3 million (roughly $6 million) seed round. TechCrunch reports the team debated launching a cloud service at that point but held off. In a blog post, cofounder Christian Winter described the priorities the group set after leaving academia, including reliability, fault tolerance and compatibility with widely used tooling.

One engine for two jobs

Most databases have historically specialised in one of two workloads: operational systems that record day-to-day transactions such as orders and payments, and analytical systems built to surface trends. CedarDB claims to handle both in a single engine, and CEO Moritz Sichert told TechCrunch that the rise of agentic AI has made that combination more relevant, because enterprise data stacks built to store years of transactions often struggle to quickly answer analytical questions phrased in natural language — exactly the kind of request an AI agent generates.

The engine itself is not a newcomer. CedarDB is a spinout from the Technical University of Munich, built on a research project called Umbra that was written from scratch to outpace legacy database engines by using modern hardware more efficiently. Five Umbra researchers — Moritz Sichert, Lukas Vogel, Philipp Fent, Dominik Durner and Christian Winter — cofounded the company, and the commercial product promises better performance without demanding extra CPUs, processing power or memory, and with few changes required on the customer side.

PostgreSQL compatibility as an AI bet

A central design decision was compatibility with PostgreSQL. Sichert told TechCrunch this keeps switching costs low, since customers do not need to rebuild their applications to move to CedarDB. He also pointed to a less obvious advantage: because large language models absorbed large amounts of PostgreSQL-related material during training, CedarDB's dialect matches the database logic those models already know well — useful when AI agents are the ones composing queries.

Databricks is converging on the same idea

CedarDB is not alone in questioning the divide between transactional and analytical systems. Databricks used its Data + AI Summit in June to unveil LTAP, a data architecture motivated by what the company called "the agentic era", aiming to bring transactions, analytics, streaming and operational data together on a single copy of data in the lake. Sichert reads that move as evidence that demand exists for what CedarDB does, though TechCrunch notes it is too early to tell how disruptive the shift will be.

For now, CedarDB plans to serve its first cloud customers with a team of 14 people. Sichert expects headcount to grow alongside adoption, and told TechCrunch that expansion is the point at which the company would raise its next funding round.

Why it matters

The split between operational and analytical databases has shaped enterprise data architecture for decades, typically forcing organisations to maintain separate systems and shuttle data between them. AI agents blur that line: they both record actions and ask open-ended questions about accumulated data, ideally against a single consistent copy. If that usage pattern becomes standard, engines that serve both access patterns in one system — whether CedarDB's approach or Databricks' lakehouse architecture — could simplify data stacks considerably. For CedarDB, the cloud launch is also the step that carries a decade of university research beyond large enterprises running their own infrastructure, opening the door to a much broader set of customers.