Breaking News: AWS Launches Next-Gen Redshift Instances

AWS today announced Amazon Redshift RG instances, a new instance family powered by its custom AWS Graviton processors, delivering up to 2.2x faster data warehouse performance than existing RA3 instances at a 30% lower price per vCPU. The new instances also include an integrated data lake query engine that accelerates SQL analytics on Apache Iceberg by 2.4x and on Apache Parquet by 1.5x over RA3.

“With RG instances, we’re giving customers the ability to run high-velocity analytics—whether from BI dashboards or autonomous AI agents—at a fraction of the previous cost,” said Rajiv Gupta, Vice President of Analytics at AWS. “This marks a major step in making cloud data warehousing both faster and more economical for the age of agentic AI.”

Industry analyst Martha Chen of CloudTech Insights added: “The combination of Graviton’s efficiency and a unified query engine for warehouse and data lake positions Redshift as a compelling choice for organizations struggling with spiraling compute costs from AI-driven workloads.”

Background: Evolving Demands on Warehousing

Since its 2013 launch, Amazon Redshift has progressed from dense compute to RA3 instances and serverless options—each generation making queries cheaper and faster. Today, enterprises increasingly rely on both structured data warehouses and cost-effective data lakes (e.g., Amazon S3) for diverse datasets.

The rise of AI agents that query warehouses at massive scale has accelerated the need for efficient, low-latency infrastructure. In March 2026, Redshift already boosted new query performance by up to 7x for BI and ETL workloads. RG instances build on that by combining hardware and software optimization.

RG Instances: Performance and Pricing Details

RG instances use AWS Graviton processors, custom-designed for cloud workloads. Compared to current RA3 instances:

• ra3.xlplus → rg.xlarge: 4 vCPU, 32 GB RAM, ideal for small departmental analytics.
• ra3.4xlarge → rg.4xlarge: 16 vCPU (1.33:1 ratio), 128 GB memory (1.33:1 ratio), suited for standard production workloads.

The integrated data lake query engine is enabled by default, allowing customers to run SQL across both warehouse tables and S3 data lakes from a single engine. This reduces total analytics costs and simplifies operations.

What This Means for Analytics and AI Workloads

The new instances directly address the high query volumes and low-latency requirements of modern analytics and agentic AI—autonomous agents that perform goal-seeking queries. Faster performance at lower cost means organizations can scale AI-driven analytics without budget surprises.

Customers already using RA3 instances are recommended to migrate using the AWS Pricing Calculator to estimate savings. Existing clusters can be migrated via Console, CLI, or API, and the data lake query engine activates automatically.

“We’re enabling customers to get more value from their data lakes without separate compute,” said Gupta. “It’s a unified approach that slashes both latency and cost.”

Getting Started

New RG instances are available immediately. Launch through the AWS Management Console, AWS CLI, or AWS API. For a full comparison, visit the background and implications sections above.