Sep 25, 2026/1 min readCore Platform
What's New in Snowflake Interactive Analytics: Summer 2026
Since our Spring 2026 update which introduced fallback warehouses, AI-assisted clustering, CoCo skills and enterprise replication, we've continued investing in making Snowflake Interactive Analytics fit seamlessly into the workflows and tooling that enterprise teams already use. Today, we're sharing three new capabilities focused on infrastructure-as-code, data pipeline integration and query performance:
- Terraform support for interactive warehouses (public preview)
- dbt support for interactive tables (public preview)
- Query engine optimizations
These features address a common theme in customer feedback: I love interactive analytics, but I need it to work with my existing deployment stack and deliver consistently fast performance for my queries.
Terraform support for interactive warehouses
For teams managing Snowflake infrastructure through Terraform as many of our largest Interactive Analytics customers do, there was a gap. Interactive warehouses had no dedicated Terraform resource. Customers were forced to use snowflake_execute as a workaround, losing the plan/apply semantics, drift detection and state management that make Terraform valuable in the first place. That gap is now closed. The new snowflake_warehouse_interactive resource in the Snowflake Terraform provider gives you full management of interactive warehouses:
resource "snowflake_warehouse_interactive" "analytics" {
name = "prod_interactive_wh"
warehouse_size = "MEDIUM"
max_cluster_count = 3
min_cluster_count = 1
auto_suspend = 86400
auto_resume = "true"
tables = [
"\"ANALYTICS\".\"PUBLIC\".\"ORDERS\"",
"\"ANALYTICS\".\"PUBLIC\".\"LINE_ITEMS\"",
]
fallback_warehouse = snowflake_warehouse.standard_wh.fully_qualified_name
comment = "Production interactive warehouse for retail dashboards"
}This single block will:
- Create the interactive warehouse via CREATE INTERACTIVE WAREHOUSE
- Associate tables and trigger cache warming
- Configure the fallback warehouse for queries exceeding the 5-second timeout
- Track all state with full drift detection and plan/apply semantics
This brings your existing infrastructure under Terraform management without recreation. The snowflake_warehouse_interactive resource is available now in the Snowflake Terraform provider as a preview feature. Enable it via the preview_features_enabled field in your provider configuration. The resource will be promoted to general availability (GA) in a future release.
dbt support for interactive tables
Starting in dbt-fusion v2.0.0, the Snowflake adapter supports interactive_table as a native materialization type. Adding an interactive table to your dbt project is as simple as adding a config block:
{{ config(
materialized='interactive_table',
cluster_by=['customer_id']
) }}
select
customer_id,
customer_name,
region
from {{ ref('stg_customers') }}dbt compiles this to a CREATE INTERACTIVE TABLE statement and manages the full deployment, creation, refresh and schema changes, just like any other dbt model. This is available now in dbt-fusion v2.0.0 as a beta feature. The cluster_by parameter is required. Full configuration reference: dbt Snowflake docs.
Why this matters: Many customers already use dbt to define their data models. When they adopt interactive tables today, they must step outside dbt to create and manage those tables, fragmenting the unified project structure that dbt provides for the transformation layer. Native dbt support means interactive tables can participate in the same dbt build, dbt run and deployment workflows as everything else in the project.
Query engine optimizations
Not all queries can benefit from Snowflake's standard micro-partition pruning. Predicates such as substring LIKE filters or equality on a nonleading clustering column cause the engine to scan every candidate micro-partition on every execution, even when most contain no matching rows. For recurring queries, that repeated wasted scan adds up.
We've introduced engine-level optimization in Interactive Warehouses that eliminates this repeated work. On the first execution, the engine identifies micro-partitions where the WHERE clause matched zero rows. On subsequent executions of the same query, the engine automatically excludes those partitions before dispatching any work. They aren't scanned again.
The result: Recurring queries that previously couldn't benefit from partition pruning now scan significantly fewer micro-partitions. Early production data shows an average of 66% reduction in scan IO for queries benefiting from this improvement, with further optimizations underway. This operates transparently on Interactive Warehouses with no configuration required.
Conclusion
Terraform, dbt and query engine optimizations represent three different layers of the Interactive Analytics stack: infrastructure provisioning, transformation pipelines and query execution. Together, they address the full deployment of what it takes to run interactive analytics in production: how you deploy it, how you build on top of it and how it performs for your most important recurring workloads.
As always, our commitment is to continuously improve based on the feedback our customers give us. These features were directly requested by customers running Interactive Analytics in production, and we're delivering them as part of the ongoing effort to make interactive a natural extension of your existing Snowflake stack.
Stay tuned. There's more to come.
This article contains forward-looking statements, including about our future product offerings, and are not commitments to deliver any product offerings. Actual results and offerings may differ and are subject to known and unknown risk and uncertainties. See our latest 10-Q for more information.
