Turbocharge Ad Hoc Query Performance with Query Acceleration Service, Now Enabled by Default
In the world of data analytics, we've all been there: Dozens of small queries are humming along perfectly until a huge ad hoc query is run on the warehouse and causes a traffic jam. Snowflake's answer to this is the Query Acceleration Service (QAS).
QAS allocates temporary, extra capacity to accelerate long-running scans and filter operations. When you run a query on a QAS-enabled warehouse, QAS automatically assesses if there's an opportunity to meaningfully accelerate your query by further parallelizing work on the pool of QAS compute nodes. If QAS doesn't identify eligible queries, then there is no additional cost. In other words, the feature is completely "serverless" and you typically only incur additional cost when QAS is applied to improve query performance.
You can use QAS to improve performance on all Snowflake managed data, regardless of whether the data is Snowflake native tables or Apache Iceberg™ tables.
Today, Snowflake is excited to announce several new improvements to QAS.
QAS now enabled by default on Gen2 and multi-cluster warehouses
When you create a new multi-cluster and/or Gen2 warehouse, QAS is now enabled by default. In scenarios such as multi-cluster and Gen2 warehouses where customers need market-leading performance, QAS provides an out-of-the-box boost. Based on aggregated internal data from calendar year 2025, QAS improved query performance by an average of 42% across a range of eligible workloads. As a result of enabling QAS by default, customers can benefit from immediate, automatic performance improvements for eligible queries.
Multi-cluster warehouses are a powerful feature for horizontal scaling. Combined with QAS, you can now automatically scale horizontally and temporarily scale vertically for QAS-eligible queries. Similarly, Gen2 already improves Snowflake performance at baseline. Since QAS is now the default on Gen2 warehouses, you now automatically get the best of both worlds: a baseline engine that is significantly faster out of the box, paired with a serverless "burst" capacity that kicks in the moment a query hits an expensive scan.
You typically only pay for QAS if it actually helps query performance. Snowflake is designed to apply QAS when it is expected to provide performance benefits. Best of all, in some cases, customers may see similar total compute costs while achieving improved performance, depending on workload characteristics.
New support for DML, Dynamic Tables and data unloads
Over time, Snowflake has expanded QAS to more and more eligible queries. As of April 2026, we're excited to share these recent improvements to QAS coverage:
| Improvement | Description | Impact |
|---|---|---|
| Expanded support for DML statements | Allow QAS to accelerate Merge, Update and Delete statements | Approximately 2x increase in the number of DML statements eligible for acceleration compared to prior coverage |
| Support for Dynamic Table refresh | Allow QAS to speed up Dynamic Table refresh queries | In testing, Dynamic Table refreshes completed in approximately 70% of prior execution time for supported workloads. |
| Support for data unloads | Allow QAS to speed up data unloads | In testing, data unloads finished in approx. 40% of the prior execution time for supported workloads. |
Some customers saw dramatic performance improvements. In one example, a financial services company reported that an expensive DML statement that they ran every day suddenly sped up from 35 minutes down to three minutes! The warehouse that ran this ETL workload reported consuming fewer than one-third of prior credits for that specific workload than it did before Snowflake rolled out QAS improvements. Additionally, a U.S.-based automotive company saw Dynamic Table refreshes drop from an average of 12 minutes down to 2.5 minutes in its environment — all without requiring additional configuration. (Results may vary across customers.)
Next steps
Try out the Query Acceleration Service today by creating a Gen2 or multi-cluster warehouse. If you already are using the Query Acceleration Service, your warehouses automatically benefit from all future QAS eligibility improvements.
Authors
