ClickHouse launches AI agents as ARR tops USD $250m

ClickHouse has launched ClickHouse Agents, a managed analytics service built with Anthropic's Claude, and said it has passed USD $250 million in annual run-rate revenue.

The service, part of ClickHouse Cloud, is designed as a no-code tool for building data-focused agents. Users can define and deploy agents that work from data held in ClickHouse's database system, with a chat interface, code interpreter, memory, skills management and multi-agent workflows.

Alongside the launch, ClickHouse published CostBench, an open benchmark comparing cost-performance across major cloud data warehouse providers. It covers ClickHouse Cloud, Snowflake, Databricks, BigQuery and Redshift, using the vendors' own compute billing models on the same analytical workload.

ClickHouse said its cloud service was the only system in the benchmark to remain in what it described as the "Fast & Low-Cost" zone across datasets of different sizes. It added that the nearest competitor was 23 times worse on cost-performance.

The announcements came as ClickHouse reported sharp commercial growth in the latest quarter. Annual run-rate revenue has more than tripled from a year earlier, and the company has added more than 1,000 net new customers since January, bringing its total to 4,000.

When ClickHouse completed a USD $400 million Series D funding round in January, it had just passed 3,000 customers. It said recent customer additions and expansions include Capital One, Lovable, Decagon, Polymarket and Airwallex. Existing users include Anthropic, Meta, Cursor, Sony, Tesla, Memorial Sloan Kettering, Lyft and Instacart.

Artificial intelligence workloads are central to the company's latest product push. ClickHouse said high query volumes and low latency requirements are changing what customers want from data infrastructure, particularly for analytics, observability and model-related telemetry.

"AI workloads demand the performance and cost efficiency ClickHouse was built for, and the last quarter has made that clearer than ever," said Aaron Katz, chief executive officer of ClickHouse.

"More than 1,000 new customers and a tripling of ARR within months of our Series D tell us this isn't a cycle, it's a structural shift in what data infrastructure has to do. What we're announcing this week is built to extend that lead as AI moves from experiment to production," Katz said.

Product expansion

ClickHouse Agents is part of a broader product expansion around AI-related workloads. The service can connect directly to ClickHouse and to third-party systems compatible with the Model Context Protocol, including through a native integration with the AWS Agent Registry.

Other updates include Managed Postgres in public beta, with native links to ClickHouse analytics, and Managed ClickStack, a managed observability service for infrastructure and model-training workloads. The company also pointed to Langfuse, which it acquired in January, as part of its work on observability for production AI systems, including evaluation, correctness and model-cost tracking.

ClickHouse said full-text search is now generally available. It also said automatic query optimisation has improved its position on TPC-H and other standard join benchmarks.

ClickHouse Cloud now also includes agentic onboarding, aimed at moving users from sign-up to a first production query without manual schema work. Cross-region replication has been introduced for customers that need greater resilience across deployments.

Cost focus

The publication of CostBench reflects a wider debate in the cloud data market over whether raw speed or overall cost per query matters more for AI-driven usage. ClickHouse argued that high-concurrency, low-latency AI workloads make cost-performance the more relevant measure.

Open benchmarks have become a common way for infrastructure companies to challenge larger rivals, though methodology and workload selection often draw scrutiny from competitors and customers. ClickHouse said its benchmark is open and reproducible, with an interactive explorer and published methodology.

Founded around an open-source columnar database for real-time analytics, ClickHouse has increasingly positioned itself around AI applications and observability as it scales its cloud business. The latest figures suggest that strategy is translating into customer growth as demand rises for systems that can manage large volumes of machine-generated and application data.