You have questions.

Traditionally, serverless means you offload server management to a public provider like Google BigQuery or Amazon Athena. You pay-as-you-go, but you lose control over where your data sits and how the underlying compute is managed.

Icebreaker brings that same provider-grade infrastructure directly into your own VPC. We automate the backend IT—specifically using one-query-per-pod containerization to bin-pack your workloads, and take advantage of your negotiated compute and Spot pricing. You get the extraordinary efficiency of a public service, but it runs entirely on your private infrastructure.

It's actually the opposite of overkill—it's the ultimate lean setup. While "right-sizing" traditional warehouses or clusters requires constant tuning, Icebreaker's serverless model optimizes every query automatically.

• Security: Your data never leaves your network.


• Query-Level Economics: Because we execute queries in isolated pods that fill "gaps" in your existing idle capacity, it is significantly less expensive than sizing and maintaining dedicated clusters and warehouses.


• Operational Simplicity: You stop worrying about routing queries or scaling instances; you simply run the query, and the platform handles the micro-scheduling automatically.

The Icebreaker Agent opens a single outbound connection from your cluster to our control plane. Nothing ever connects inbound to reach your cluster, so there's no firewall rule to open, no exposed endpoint to defend, and no direct access to your sensitive data to hand over.

Only what capacity management actually requires, scoped to least privilege. The Agent's permissions are limited to the namespace you choose; cluster-wide access is limited to the handful of resources that capacity management needs.

Yes. Every connection — Agent to control plane, browser to console — runs over TLS.

Sign-in goes through Google OAuth today, with additional enterprise identity providers on the roadmap. Icebreaker never sees or stores your password. See the full breakdown on our Security page.

Icebreaker replaces SQL workloads that use Iceberg tables. Our perpective is there's no point in paying premium pricing for what are now commodity services.

Icebreaker will support Spark for distributed queries in the future. But you may still find value in Icebreaker now. Many jobs that run on distributed architectures can run just fine on large, single instances at the same "compute unit" cost.

Currently, yes. Icebreaker works with Iceberg tables.

You can use Icebreaker in your ETL/ELT pipeline to write Iceberg tables and perform SQL transformations.

Icebreaker currently runs in your AWS cloud account using AWS EC2 instances. You get to prioritize on-demand, reserved instance (RI), or spot pricing for your various workloads.

Icebreaker was inspired by organizations just like yours! Too often, organizations have loads of discounted compute capacity, which they use for non-data applications. They still pay for Snowflake, or Redshift, or BigQuery (the list goes on) for data processing! Instead, you can run your SQL jobs on Icebreaker, which uses available capacity on Reserved Instances (or spot instances for even more savings). See BYOC or DaaS to learn more.

Yes, Icebreaker provides endpoints that work with your BI tool.