# Find the right HyperLake capabilities

## Batch and real-time analytics without competing for one engine

Velocity Lakehouse combines open tables with a ClickHouse hot tier and separate transformation capacity. Plan how batch, interactive and existing warehouse work will be divided.

- Features: Iceberg, Trino and ClickHouse in one stack; Separate transformation and interactive engines; LakeGate policy and query routing
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/batch-realtime-lakehouse
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Serve existing ML models beside your data: plan the endpoint

Predict serves classic ML models and records predictions with your data systems. Evaluate compatibility, feature freshness and operational behavior before moving an existing endpoint.

- Features: Serving for supported classic ML model formats; dbt feature models and prediction logs in ClickHouse; Data-quality monitors
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/classic-ml-serving
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Federated SQL across existing systems: plan a governed query

Unify uses managed Trino for queries across existing sources. Test one cross-system decision before deciding which copies or sync jobs you can retire.

- Features: Trino federation over connected catalogs; Row filters, column masks and a semantic layer; Governed access through LakeGate
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/federated-sql
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Agent sandboxes with governed data access: plan the boundaries

AgentBay provides isolated code and tool sandboxes with data access through LakeGate. Define what an agent may read, execute and change before giving it a real workflow.

- Features: Isolated sandboxes for agent code and tools; Scoped data access through LakeGate; MCP access alongside existing agent toolkits
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/governed-agent-sandboxes
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## An AI analyst on governed company data: start with one decision

Analyst combines the lakehouse, private models and agent sandboxes. Scope one business question, its permitted evidence and the checks a human reviewer will apply.

- Features: Velocity Lakehouse, Enclave and AgentBay together; Semantic-layer and policy-governed data; Analysis code in sandboxes and data-quality monitors
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/governed-ai-analyst
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Governed database access for people and agents

LakeGate places identity and policy in front of databases and data engines. Start by replacing one shared-credential workflow with scoped, reviewable access.

- Features: Identity-provider sign-in and role-based access; OPA policy decisions and query guardrails; Query logging across integrated engines
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/governed-database-access
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## LLM API vs self-hosted cost calculator

Compare API spend with self-hosted inference using your own prices, request volume and GPU assumptions.

- Features: Workload-based monthly cost comparison; Break-even volume and GPU utilization; Calculator runs in your browser
- Economics: Use your actual API rates, cloud prices and expected utilization. Include software and operational costs when making the buying decision.
- Switching: Verify throughput, model quality, latency and operational requirements before replacing an API service.
- Read more: https://hyperlake.cloud/solutions/inference-cost
- Sources: https://hyperlake.cloud/tools/llm-api-vs-self-hosted-cost-calculator, https://hyperlake.cloud/

## Managed Kubernetes in your own cloud: plan the operating model

Foundry packages a Kubernetes foundation in your AWS or Google Cloud account. Decide who owns access, upgrades, recovery and day-two operations before adding production workloads.

- Features: Managed Kubernetes foundation; Observability, single sign-on, private access, backups and autoscaling; Foundation for additional HyperLake stacks
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/managed-kubernetes
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Managed Postgres and MySQL: protect the app, separate reporting

Fusion combines application databases with reporting on ClickHouse. Start with availability and recovery needs, then assess which analytical work should leave the primary.

- Features: Managed Postgres and MySQL; High availability and point-in-time recovery; ClickHouse reporting from Postgres, behind LakeGate
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/managed-postgres-mysql
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## An open Iceberg lakehouse: plan tables, access and migration

Openhouse combines Iceberg, Trino and dbt on your own object storage. Make table ownership, interoperability and workload fit explicit before migrating an existing lake.

- Features: Iceberg tables on customer object storage; Trino, dbt modeling and a semantic layer; Dashboards and data monitors behind LakeGate
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/open-iceberg-lakehouse
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Physical AI infrastructure and open-source stack map

Map the software layers needed for physical AI and robotics infrastructure.

- Features: Projects grouped by infrastructure layer; Official project references; Connections to data and AI infrastructure
- Economics: Evaluate infrastructure capacity and the work required to operate the selected components. Open-source licensing alone does not determine total cost.
- Switching: Check hardware, data formats, integration needs and operational ownership for your specific deployment.
- Read more: https://hyperlake.cloud/solutions/physical-ai
- Sources: https://hyperlake.cloud/physical-ai/open-source-stack-map, https://hyperlake.cloud/physical-ai

## Private AI infrastructure: features, cost and control

Compare deployment boundaries, model control and who operates the stack. HyperLake manages open-source stacks in your own cloud account.

- Features: Customer-controlled cloud account and region; Managed data engines, model serving and agent workspaces; Expert support included in the subscription
- Economics: Your cloud compute is billed at 0% HyperLake markup, plus the HyperLake subscription. Compare equivalent capacity, utilization and operating effort; total savings depend on the workload.
- Switching: Check supported cloud accounts, data boundaries, model requirements and migration scope with the team. On-premises availability must be confirmed separately.
- Read more: https://hyperlake.cloud/solutions/private-ai
- Sources: https://hyperlake.cloud/, https://hyperlake.cloud/blog/private-ai-infrastructure-for-enterprises-in-2026-options-compared

## Private model serving and fine-tuning: plan quality and capacity

Enclave runs open or custom models on GPUs in your cloud. Compare model quality, workload utilization and operational responsibility before replacing a hosted endpoint.

- Features: Model serving and fine-tuning on customer GPUs; pgvector retrieval and governed data access; Eligible endpoints can scale down when idle
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/private-model-serving
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Real-time analytics in your cloud: scope a ClickHouse workload

Velocity provides managed ClickHouse for dashboards, product analytics and data APIs. Use your own ingestion and query patterns to evaluate a move from hosted analytics.

- Features: Managed ClickHouse behind LakeGate; Backups, sharding and day-two actions; Role-based access for analytics workloads
- Economics: Compare the cloud bill, platform charges, operating effort and migration work for the same workload. Cloud capacity is not free. Charges and scope need agreement before paid deployment.
- Switching: AWS and Google Cloud are available today; other deployment environments are roadmap. Preserve the existing workload and access path until compatibility, recovery and cutover checks pass.
- Read more: https://hyperlake.cloud/solutions/realtime-analytics
- Sources: https://hyperlake.cloud/stacks, https://hyperlake.cloud/faq, https://hyperlake.cloud/get-started

## Compare warehouse platforms and workload economics

Compare the platform to the workload, including cases where another architecture fits better.

- Features: Workload and architecture comparison; Managed open-source data stacks; Option to run alongside an existing platform
- Economics: Compare compute, storage, licensing, support and migration effort at the same workload and service level. 0% compute markup does not mean zero platform or operating cost.
- Switching: Identify which workloads can move and which capabilities should stay on the current platform. Confirm compatibility before migration.
- Read more: https://hyperlake.cloud/solutions/warehouse-cost
- Sources: https://hyperlake.cloud/blog/cloudera-vs-snowflake-in-2026-and-when-neither-fits, https://hyperlake.cloud/

Pricing depends on the agreed scope, usage and support requirements.

[Get free expert help with your use case](https://hyperlake.cloud/contact)