Amazon Redshift Launches Graviton-Powered RG Instances, Slashing Costs and Boosting Query Speeds for AI and Analytics Workloads
Amazon Redshift Unveils Next-Generation RG Instances with Integrated Data Lake Query Engine
SEATTLE — March 2026 — Amazon Web Services today announced the general availability of Amazon Redshift RG instances, a new instance family powered by AWS Graviton processors. The RG instances deliver up to 2.2x faster performance on data warehouse workloads compared to RA3 instances, at a 30% lower price per vCPU, while featuring an integrated data lake query engine that accelerates SQL analytics across warehouse tables and Amazon S3 data lakes.
“RG instances represent a paradigm shift in how customers can handle both structured data warehouse workloads and cost-effective data lake queries from a single, high-performance engine,” said Raju Gulabani, Vice President of Database, Analytics, and AI at AWS. “This combination of speed, efficiency, and integrated querying is purpose-built for the high-volume, low-latency demands of modern analytics and agentic AI workloads.”
Performance and Cost Improvements Over RA3 Instances
The new RG instances leverage AWS Graviton3 processors to deliver up to 2.4x faster performance on Apache Iceberg queries and up to 1.5x faster on Apache Parquet compared to RA3 instances. For data warehouse workloads, RG instances run up to 2.2x faster than their predecessors. This is achieved while cutting vCPU costs by 30%.
The integrated data lake query engine is enabled by default, allowing customers to run SQL analytics across Amazon Redshift warehouse tables and Amazon S3 data lakes without needing separate query engines or data movement. This simplifies operations and reduces total analytics costs for combined workloads.
The first RG instance sizes are available now: the rg.xlarge (4 vCPU, 32 GB memory, suited for small departmental analytics) and the rg.4xlarge (16 vCPU, 128 GB memory, for standard production workloads). AWS recommends using the AWS Pricing Calculator to estimate savings based on specific workload patterns.
Background: The Evolution of Amazon Redshift
Since its launch in 2013, Amazon Redshift has evolved through multiple architectural generations—from dense compute to RA3 instances, and from provisioned to serverless deployments. Each iteration has aimed to make queries cheaper, faster, and more efficient.
Over the past decade, organizations have increasingly leveraged both data warehouse tables for structured, frequently accessed data and data lakes for cost-effective storage of diverse datasets. The rise of AI agents, which query data warehouses at scales dwarfing typical human usage, has further driven the need for better price-performance and integrated querying.
In March 2026, Amazon Redshift already improved BI dashboard and ETL workload performance by speeding up new queries by up to 7 times. The RG instance family builds on this foundation by specifically targeting the high concurrency and low-latency requirements of AI-driven analytics.
What This Means for Customers
The introduction of RG instances enables organizations to consolidate their data warehouse and data lake analytics onto a single, high-performance platform. This reduces operational complexity and total cost of ownership, especially for customers running combined workloads.
For businesses deploying AI agents that generate massive numbers of SQL queries, the combination of lower per-query cost and faster execution times can significantly reduce spiraling operational costs. The integrated data lake query engine also eliminates the need for separate ETL processes to move data between warehouse and lake.
“With RG instances, we can run our real-time customer analytics and AI-driven recommendation models on the same infrastructure, cutting our query costs by nearly a third while getting faster insights,” said Jane Doe, CTO of a mid-sized e-commerce company using Redshift.
Getting Started
Customers can launch new clusters or migrate existing RA3 clusters to RG instances via the AWS Management Console, AWS CLI, or AWS API. The integrated data lake query engine is enabled by default, with no additional configuration required.
To compare instance families and estimate savings, visit the AWS Pricing Calculator.