AI-driven growth will reshape enterprise data estates in both size and complexity. Frontier models and the applications built around them will depend on a growing network of databases, tables, semantic views, Cortex Search services and agents, each holding data that will be essential to recovery during a service disruption. As these workloads move into production, teams face a widening recovery footprint without more time to meet their recovery objectives, putting pressure on replication to keep pace with both rising data volumes and multiplying objects.
Recovery also carries a security requirement: applications must resume with appropriate access controls in place. As AI workloads reach more enterprise data, recovery planning needs to account for who can access that data and how those permissions carry into the recovery environment.
With faster replication and contractually backed recovery commitments for eligible workloads, Snowflake is raising the bar for enterprise readiness in AI data platforms.
Snowflake's next generation of account replication — Optimized Refresh and RPO Assurance, both now generally available — can help customers modernize their most critical workloads while supporting the recovery requirements their businesses depend on.
A new engine for faster replication
A brand-new, high-performance replication engine powers both Optimized Refresh and RPO Assurance. Select customer workloads in preview have seen improvements ranging from 1.6x to more than 20x compared with the previous generation. Results varied by workload configuration. Faster refreshes are designed to keep recovery environments current and reduce the volume of data that needs to be caught up after failover, helping enterprises support their recovery point objective (RPO) and recovery time objective (RTO) requirements during a service disruption.
That performance builds on Snowflake’s account replication capabilities across regions and clouds. Customers can replicate databases alongside users, roles, supported privilege grants, network policies and supported security integrations. Including these controls and their dependencies in the recovery configuration helps teams preserve governed access through failover and reduces the need to reconstruct permissions during a disruption.
Optimized Refresh for broad disaster recovery coverage
Optimized Refresh is designed to meet the disaster recovery needs of most Snowflake workloads, from production analytics to data pipelines and AI applications. It is designed to be cost efficient and to deliver predictably low RPOs on a best-effort basis, with refresh schedules you control.
"Snowflake’s Optimized Refresh has transformed our replication operations. By making replication far more predictable and substantially reducing runtime, it has materially improved our RPO and RTO. The result is a stronger recovery posture and greater confidence in our ability to support critical business operations."
You can enable it with one setting on the primary failover group:
RPO Assurance for your most critical workloads
For highly sensitive and mission-critical workloads, RPO Assurance delivers these additional benefits:
- A contractual, financially backed service-level agreement (SLA): Once enabled for eligible failover groups, replicas stay within a 30-minute RPO 99.9% of the time, subject to the service terms and operating conditions.
- Highest priority: Snowflake prioritizes these workloads over other replication workloads across regions and clouds, giving them first access to replication compute and networking during contention.
- Continuous managed refreshes: Snowflake manages the refresh cycle without a customer-configured schedule.
- Intelligent throughput: Snowflake automatically allocates additional compute when beneficial, delivering up to twice the throughput of Optimized Refresh.
- RPO visibility: Replication lag history helps platform, security and risk teams assess recovery readiness against their organization’s data-loss tolerance.
"State Street Alpha is leveraging RPO Assurance at scale alongside robust automation, governance and operational excellence to deliver a resilient data platform that helps safeguard critical workloads and strengthen confidence in recovery when it matters most."
After onboarding, enable RPO Assurance on an eligible primary failover group with no existing replication schedule:
Simplified pricing
Both options use volume-based pricing, making costs easier to estimate from the data you expect to replicate to the target account:
| Replication option | Credits per terabyte replicated as of October 2026 |
|---|---|
| Optimized Refresh | 5 |
| RPO Assurance | 10 |
For both options, metadata charges apply after the first 25 million changed objects per account per month. Standard storage and data transfer charges remain separate and unchanged from the prior model.
Get started
Learn more about Optimized Refresh and get started protecting your critical workloads today.
To onboard eligible critical workloads to RPO Assurance, contact Snowflake Support or your Snowflake account team.
