10 Things You Need to Know About AWS's New Graviton-Powered Redshift RG Instances

Introduction

AWS has launched new Graviton-powered Redshift RG instances to help enterprises lower analytics costs and simplify lakehouse operations. By integrating the data lake query engine directly into Redshift, these instances address the complexity and unpredictable pricing of the earlier two-engine setup. Here are 10 key insights you need to understand about this release.

1. What Are Redshift RG Instances?

Redshift RG instances are the latest hardware option for AWS's data warehouse service, powered by the company's own Graviton processors. These instances are designed to run analytics workloads more efficiently, especially those involving both warehouse data and data stored in Amazon S3. The key improvement is the inclusion of an integrated data lake query engine, which eliminates the need to coordinate between separate Redshift and Spectrum engines. This unified approach aims to reduce operational overhead, improve query performance, and lower overall analytics costs—making it a compelling upgrade for enterprises dealing with large-scale, mixed-data scenarios.

2. The End of the Two-Engine Headache

Earlier Redshift systems, like the RA3 instances, used two separate engines: one for warehouse data and another (Spectrum) for querying S3 data lakes. When a query needed data from both, AWS had to stitch results together, adding complexity and slowing performance. Worse, Spectrum's per-scan pricing made costs unpredictable. The RG instances combine these into a single engine running inside Redshift itself. Now, data in Iceberg, Parquet, and other open formats can be queried natively alongside warehouse data with less data movement and better optimization. Overhead drops, and the need to manage separate scan charges disappears.

3. Lower Costs by Eliminating Spectrum Charges

A major pain point for enterprises was Spectrum's separate scan-based pricing. As AI workloads drove higher query volumes and more machine-generated analytics, the cost of scanning S3 data could spike without warning. The RG instances remove this unpredictability entirely. By integrating the lake query engine into Redshift, AWS collapses two pricing models into one. This not only simplifies billing but also makes it easier for finance teams to forecast analytics spend. For organizations with heavy S3 lake usage, the cost savings can be significant—especially as data volumes continue to grow.

4. Performance Gains from Integration

The unified engine in RG instances allows AWS to optimize queries across both warehouse and lake data simultaneously. This means faster query execution for mixed workloads, as data doesn't need to be moved between separate systems. Performance improvements come from tighter coupling of the compute and storage layers, as well as better caching and query planning. While AWS hasn't published specific latency benchmarks, early user feedback suggests noticeable speedups for complex analytical queries that previously required cross-engine coordination. Enterprises running BI dashboards or real-time analytics should see reduced wait times.

5. Designed for Modern Lakehouse Architectures

The RG instances are clearly a response to the growing popularity of lakehouse architectures that blend data lakes with warehouse capabilities. Rivals like Databricks and Snowflake have already pushed unified platforms, and AWS needed to catch up. By baking the lake query engine into Redshift, AWS offers a more seamless path for organizations that already store large amounts of data on S3. This is particularly relevant for companies using open table formats like Iceberg and Parquet. The new instances reduce the friction of managing separate systems, aligning with the industry trend toward simplification.

6. A Defensive Move Against Competitors

Industry analysts view the RG instance launch as a defensive move rather than a breakthrough. Competitors such as Databricks (AI/data science), Snowflake (multi-cloud simplicity), Google Cloud (AI-native analytics via BigLake), and Microsoft (Fabric + Copilot integration) are all vying for enterprise lakehouse workloads. AWS's bet is on the sheer scale of Amazon S3 and tighter Redshift optimization to keep workloads within its ecosystem. While the RG instances strengthen Redshift's position, they don't fundamentally change the competitive landscape—they simply close the gap in areas where AWS was lagging.

7. CIOs Should Focus on the 'Painful Overlap'

According to Greyhound Research Chief Analyst Sanchit Vir Gogia, the best fit for RG instances is not every workload but the 'painful overlap' where several factors converge: Redshift, S3, open formats, BI, recurring analytics, cost pressure, and AI-assisted querying. In these scenarios, RG can materially reduce friction. CIOs should inventory exactly where that overlap exists in their organization—typically in departments that run frequent cross-data queries with cost sensitivity. By targeting those use cases, enterprises can maximize the value of the new instances without trying to force all workloads onto them.

8. How It Differs from Competitor Offerings

Each major cloud vendor approaches lakehouse unification differently. Databricks leans on AI and data science; Snowflake emphasizes multi-cloud simplicity; Google Cloud offers AI-native analytics through BigLake; and Microsoft integrates tightly with Fabric, Power BI, and Copilot. AWS's approach with RG instances is to double down on its core strengths: the massive scale of S3 and deep optimization within Redshift. This means that while competitors may offer broader AI features or easier cross-cloud mobility, AWS offers potentially lower cost and simpler management for organizations already heavily invested in the AWS ecosystem.

9. Who Benefits Most from RG Instances?

The primary beneficiaries are enterprises that (a) run significant analytics workloads on both Redshift and S3 data lakes, (b) are frustrated by unpredictable Spectrum scan costs, (c) use open table formats like Iceberg, and (d) need faster performance for mixed queries. Also, organizations with growing AI/ML workloads that generate high query volumes will find the predictable pricing and integrated engine advantageous. Smaller companies or those with simple warehouse-only workloads may not see as much benefit. The RG instances are built for scale and complexity, making them ideal for data-intensive enterprises.

10. Getting Started and Migration Considerations

Migrating to RG instances involves moving from existing RA3 or other instance types. AWS provides tools for resizing clusters, but careful planning is needed to avoid downtime. Enterprises should test the new instances on a subset of workloads first, especially those that heavily use Spectrum. The switch eliminates Spectrum charges, so the cost savings should be estimated based on historical scan volumes. Additionally, organizations should evaluate whether their data lake formats are fully compatible—Iceberg and Parquet are well supported. With proper planning, the migration can yield significant long-term savings and performance gains.

Conclusion

The Graviton-powered Redshift RG instances represent a significant step forward for AWS in the lakehouse war. By integrating the data lake query engine into Redshift, AWS addresses key pain points around cost unpredictability and operational complexity. While it is a defensive move against rivals, it offers real benefits for enterprises with the right mix of workloads. CIOs should carefully assess where the 'painful overlap' exists in their organization and consider migrating those specific use cases to RG instances to lower costs and boost performance.