Tinybird

What Is Tinybird?

The data warehouse market has consolidated around a handful of American giants — Snowflake, Databricks, BigQuery. All are capable platforms for batch analytics. None of them were designed to answer a user's dashboard query in under 100 milliseconds against a dataset that is still growing in real time.

That gap is where Tinybird operates. Founded in 2019 in Madrid by a team with backgrounds in large-scale data infrastructure, Tinybird built a developer platform on top of ClickHouse — the open-source column-oriented database originally developed at Yandex that has become the de facto standard for high-speed analytical queries. Tinybird manages the cluster, handles ingestion from Kafka and S3, and does something no competing product does as cleanly: it lets you publish the results of a SQL query as a production-grade REST API endpoint in minutes, with authentication, caching, and rate limiting included.

The target user is an engineering team building a product with embedded analytics — a SaaS dashboard showing customers their own usage, a financial platform displaying real-time transaction trends, a logistics tool tracking fleet status. In every case, the pattern is the same: high-volume event data arrives continuously, users query it interactively, and any latency above a few hundred milliseconds degrades the user experience. Tinybird is purpose-built for this pattern.

The company has raised funding to Series A stage and operates exclusively from Madrid, with data hosted in AWS eu-west-1 (Ireland). For European teams, that geographic and legal grounding matters alongside the technical capabilities.

Key Features

ClickHouse-Native Analytics Engine

Tinybird's performance story rests on ClickHouse, which can scan billions of rows per second using columnar storage and vectorised query execution. Tinybird provisions and manages this infrastructure, so teams get the performance without the operational overhead of running clusters. In practice, this means queries over 100 million rows execute in under a second — a threshold that Snowflake or Redshift can rarely match on the same data volume without expensive caching layers or materialised views.

Data arrives via HTTP event endpoints (supporting up to 100,000 events per second), Kafka connectors for streaming sources, or S3 imports for batch loading historical data. Schema inference handles unstructured JSON payloads, and a quarantine system isolates malformed records without halting the pipeline — a practical detail that matters when ingesting data from multiple upstream producers of varying quality.

SQL Pipes and API Publishing

The core workflow in Tinybird is the Pipe: a chain of SQL transformations applied to one or more Data Sources. At the end of the chain, a Pipe can be published as an API endpoint with a single click. The endpoint accepts query parameters, applies token-based authentication, enforces rate limits, and returns JSON — no application code required.

This architecture compresses a workflow that traditionally requires a data warehouse query layer, a backend API service, a caching layer, and an authentication system into a single managed primitive. For small engineering teams, that compression translates directly into shipping time. For larger teams, it means analytics APIs can be owned and versioned by the data team without requiring backend engineering involvement.

Git-Native Deployment

Tinybird has invested heavily in developer experience at the deployment layer. Data pipelines are defined as code in version-controlled repositories, and the platform supports branch-based environments — a staging branch for testing changes, a production branch for live traffic. CI/CD integration means a pull request to a data pipeline triggers automated tests before deployment, the same workflow engineers use for application code.

This approach stands in contrast to the point-and-click pipeline builders common in other BI tools. It is explicitly optimised for engineering teams rather than business analysts, which is both Tinybird's strength and its constraint.

Copy Pipes and Materialised Views

For queries that are too complex or expensive to run in real time even with ClickHouse's speed, Tinybird supports Copy Pipes — scheduled SQL transformations that pre-aggregate data into new tables. This is effectively a materialised view system, and it allows teams to build hierarchical pipelines where expensive computations run on a schedule and fast-path queries read from pre-computed results.

The practical use case is time-series aggregation: raw events arrive at sub-second intervals, a Copy Pipe aggregates them into hourly or daily summaries every few minutes, and the API endpoint queries the summary table rather than the raw event stream. This pattern handles virtually any scale with sub-100ms query times.

Workspace Collaboration

Tinybird supports multi-member workspaces with shared access to Data Sources and Pipes. The Business tier adds SAML SSO and audit logs for enterprise deployments. Teams working across time zones can iterate on pipelines independently, with the Git-based deployment model preventing conflicts and providing a clear history of changes.

Pricing

Tinybird's pricing is structured around data volume processed. The Free tier covers 10 GB per month and 1,000 API requests per day — enough to prototype and validate a use case, but not production-capable for any meaningful traffic.

The Pro tier at $249/month (billed monthly) includes 500 GB processed and unlimited API requests, which covers most early-stage product deployments. The Business tier at $999/month adds 2 TB, unlimited members, and a dedicated Slack support channel — suitable for mid-sized production workloads.

Compared to Snowflake, where compute costs alone for real-time query patterns can reach several thousand dollars per month at equivalent data volumes, Tinybird's pricing is substantially lower for analytics API use cases. The critical caveat is that Tinybird is not a general-purpose data warehouse — if the requirement is ad hoc SQL over historical data with no latency constraint, Snowflake remains more capable.

An annual subscription reduces Pro pricing by approximately 20%. Enterprise contracts are negotiated directly and include custom data volumes, dedicated infrastructure, and formal SLAs.

EU Compliance & Privacy

Tinybird is a Spanish company — Tinybird Technologies SL, headquartered in Madrid — and hosts customer data in AWS eu-west-1 (Dublin, Ireland). Data does not leave the EU by default. The company holds SOC 2 Type II certification and offers Data Processing Agreements to paying customers, which satisfies the contractual requirements of most GDPR compliance programmes.

There is no option for US-region hosting, which is a constraint for multinational teams but an advantage for EU-focused deployments. The absence of a non-EU hosting region means the architecture cannot accidentally route European personal data outside the jurisdiction.

Role-based access controls govern who can read, modify, or publish Data Sources and Pipes within a workspace. Token-based API authentication allows granular control over which endpoints are accessible and at what rate. For organisations processing personal data through analytics pipelines, these controls support data minimisation and access restriction requirements under GDPR Article 25.

Who It's Best For

Engineering teams building product analytics with user-facing dashboards that require sub-second query responses. If a dashboard's load time is a product quality metric, Tinybird's architecture addresses it directly.

Data teams in EU-regulated industries — fintech, healthtech, government services — where data residency is a legal requirement rather than a preference. The combination of Spanish HQ, Irish hosting, and SOC 2 certification satisfies most regulatory procurement checklists.

Startups moving off ad hoc analytics who want to productionise their data pipelines without hiring infrastructure engineers. The Free tier allows exploration; the Pro tier supports early production deployments at a predictable monthly cost.

Teams with streaming data sources — Kafka-backed event systems, high-frequency sensor data, real-time transaction logs — where batch-oriented warehouses introduce unacceptable latency between data arrival and query availability.

The Verdict

Tinybird solves a specific problem exceptionally well: making high-volume, real-time analytical data queryable via an API with minimal engineering overhead. The managed ClickHouse foundation delivers genuine performance advantages over general-purpose warehouses for low-latency use cases. The SQL-plus-API-endpoint workflow is genuinely novel and well-executed. EU hosting and SOC 2 certification make it deployable within GDPR-governed organisations without additional compliance work.

The trade-offs are real: it is not a Snowflake replacement for broad data warehousing, it has no visual front-end for business analysts, and the free tier is too constrained for meaningful production use. Teams whose analytics requirements stay within the OLAP-API pattern will find Tinybird difficult to displace once deployed. Teams with broader or more varied data requirements will need to treat it as a specialist component rather than a platform.

Frequently Asked Questions

Is Tinybird GDPR compliant?

Yes. Tinybird is headquartered in Madrid, Spain, and stores all customer data in AWS eu-west-1 (Dublin, Ireland). The company is SOC 2 Type II certified and offers Data Processing Agreements to paying customers. Data does not leave the EU.

How does Tinybird compare to Snowflake?

Tinybird is purpose-built for real-time, sub-second analytics API workloads using ClickHouse. Snowflake is a general-purpose data warehouse optimised for batch analytics and broad SQL compatibility. Tinybird is faster and cheaper for low-latency API patterns; Snowflake supports broader use cases including data sharing, ML integrations, and ad hoc exploration.

Can I self-host Tinybird?

No. Tinybird is a fully managed SaaS platform. ClickHouse itself is open-source and self-hostable, but Tinybird's pipeline management, API publishing, branch-based environments, and CI/CD integrations are only available on the managed platform.

What counts as "processed data" in Tinybird's pricing?

Processed data refers to the volume of data scanned or written during ingestion and query execution. Running a query over a 1 GB table counts as 1 GB of processed data. Materialised views via Copy Pipes count data written to the target table. Tinybird's dashboard shows real-time consumption so teams can monitor usage against their plan limits.

Does Tinybird support real-time data from Kafka?

Yes. Tinybird provides native Kafka connectors, including support for Confluent Cloud. Events published to a Kafka topic can be ingested into a Tinybird Data Source with end-to-end latency under one second, making the data immediately queryable via API endpoints.