Oracle Autonomous AI Lakehouse

Overview

The service queries Apache Iceberg tables in place across clouds, avoiding the need to move that data. It combines lake access with SQL analytics, transactions, streaming ingestion, and support for structured, semi-structured, and unstructured data. Separate storage and compute and a hybrid deployment model suit teams balancing scalable cloud resources with customer infrastructure.

Oracle offers the service on OCI, AWS, Azure, Google Cloud, and Exadata Cloud@Customer. That reach may help organizations working across providers, but the Oracle database foundation makes it a more natural fit for teams already invested in Oracle than for those seeking only a focused lake query layer.

Key features

Integration and sharing

Catalog integrations include OCI Data Catalog, AWS Glue, and Apache Iceberg catalogs in Databricks and Snowflake. Data Studio provides drag-and-drop workflows to integrate data from more than 100 application, cloud service, and database sources. That range can reduce the number of separate ingestion paths teams need to manage, though the breadth will matter less to buyers with a small, stable set of sources.

Data Studio also supports bidirectional sharing with services such as Power BI and Tableau through Delta Sharing. This gives teams a route to share data with those tools in both directions rather than treating the lakehouse as an isolated repository.

AI and automation

Select AI can use models from Cohere, Azure OpenAI, OpenAI, OCI Generative AI, Google, Anthropic, Hugging Face, and AWS, among others. Machine learning, graph analytics, spatial features, and AI Vector Search support semantic search and retrieval-augmented generation. This combination is relevant to organizations seeking analytics and AI capabilities within the same platform; it may be more scope than teams that only need conventional SQL analytics require.

Oracle says autonomous management handles provisioning, configuration, security, tuning, and scaling. Autonomous AI Database encrypts data at rest and in transit by default, applies security patches and updates automatically, and meets a broad set of international and industry-specific compliance standards. These managed controls can reduce operational work, while organizations should still assess whether the deployment and usage model suits their own requirements.

Pricing

The pricing model is freemium, with a capacity-limited free option and usage-based paid deployments. Costs depend on the selected infrastructure and consumption, so buyers with variable workloads should account for ECPU and storage usage rather than treating the service as a fixed-price subscription.

• Always Free Autonomous AI Lakehouse: Free. It provides two Autonomous AI Database instances through Oracle Cloud Free Tier for unlimited time, subject to capacity limits. This is a way to explore the service without a time-based expiry, but capacity constraints limit its role for sustained or demanding production workloads.
• Oracle Autonomous AI Lakehouse Serverless: custom pricing, billed by ECPU per hour and storage and backup storage per gigabyte per month. The pricing structure also includes a developer instance. This usage-based option suits teams that want serverless deployment, but consumption can vary with workload.
• Oracle Autonomous AI Lakehouse on Exadata Cloud@Customer: custom pricing, billed by ECPU per hour and developer instance per hour. It is aimed at deployment on Exadata Cloud@Customer, with hourly charges tied to use.
• Oracle Autonomous AI Lakehouse on Dedicated Infrastructure: custom pricing, billed by ECPU per hour and developer instance per hour. The Database Exadata Infrastructure subscription has a 48-hour minimum term, a material constraint for short-lived workloads.
• Bring Your Own License: custom pricing, billed by ECPU per hour, and listed for Serverless, Dedicated, and Exadata Cloud@Customer deployments. This is the relevant route for teams bringing a license rather than using the other listed plan structure.

Oracle also offers US$300 in cloud credits for up to 30 days. The credit expires when it is spent or when 30 days pass, whichever comes first, making this a short evaluation window rather than a lasting free allowance. For offline development, Oracle provides an unlimited-time container image with Database Actions, ORDS, APEX, and the Database API for MongoDB.

Platforms

The platform is available through API and web, and supports Linux and self-hosted environments. Its cloud availability spans OCI, AWS, Azure, Google Cloud, and Exadata Cloud@Customer; the local container image also supports offline development. JSON and OSON document formats are supported, with a maximum document size of 32 MB.

Who it's for

Oracle Autonomous AI Lakehouse is a strong candidate for organizations that need to query Iceberg data across clouds while combining Oracle database capabilities, integrated data engineering, data sharing, and AI tools. It is less compelling for buyers who need a narrowly scoped query service, cannot accommodate consumption-based costs, or need free capacity without limits.

Pros and cons

• Pro: Queries Apache Iceberg tables in place across clouds, avoiding data movement for this access pattern.
• Pro: Data Studio connects more than 100 source types and supports bidirectional sharing with tools such as Power BI and Tableau.
• Pro: A broad choice of AI models and analytics capabilities supports varied AI and retrieval workloads within the platform.
• Pro: Autonomous management and default encryption reduce routine administration and provide built-in security controls.
• Con: The Always Free offer is limited by capacity, so it cannot be treated as unrestricted production capacity.
• Con: Paid usage is metered across ECPU and, for Serverless, storage and backup storage, which makes the bill depend on consumption.
• Con: Dedicated Infrastructure carries a 48-hour minimum subscription term, limiting its fit for very short deployments.

Alternatives

Starburst Data Platform is worth considering when a forever-free option with up to three clusters for ad hoc queries is the priority.

Starburst Galaxy offers a forever-free plan with up to three clusters and standard ad hoc query execution for teams comparing free options.

Amazon SageMaker Autopilot is a paid, pay-as-you-go choice for buyers seeking an alternative focused on that product's machine-learning offering.

Bauplan has a free sandbox with public datasets or user uploads, a CLI, SDK, API, and community Slack support; it may fit teams comfortable with a shared environment where everything is public.

Databricks Notebooks offers a Free Edition with one serverless workspace and limited compute size and usage, for readers comparing a limited free workspace.

IOMETE offers a free self-hosted, on-premises plan capped at 100 vCPUs, which may suit teams prioritizing that deployment model.

Apache Hudi is an open-source data lakehouse platform with source releases and Maven artifacts.

Cloudera Data Lake Service is another paid data lake service.

Compare more options in Data Lakehouse Platforms, Document Databases, Data Warehouse Software, OLAP Databases, and OLAP Software.

Verdict

Choose Oracle Autonomous AI Lakehouse if your organization wants cross-cloud Iceberg access combined with Oracle database, integration, and AI capabilities, especially when automated management is valuable. Look elsewhere if you need predictable fixed pricing or substantial free capacity: paid charges follow usage, and the free instances are capacity-limited.