Adaptive Compute Delivers High Performance That Evolves with Your Workloads
The fast-changing data and AI landscape is reshaping how organizations evaluate their compute for data analytics and engineering. With workloads becoming more diverse and less predictable, it’s even more critical to balance performance, ease of use and scalability to deliver outcomes that move the business forward.
Snowflake Adaptive Compute (generally available soon) directly addresses these core operational challenges by providing high performance without operational complexity. Warehouses created using Adaptive Compute, known as Adaptive Warehouses, eliminate the manual effort and technical complexity required to configure, tune and manage compute resources at scale, enabling you to enhance throughput, accelerate time to insight and free your engineering teams to focus on innovation rather than maintaining compute resources.
What makes Adaptive Compute different
Adaptive Compute is workload-aware and dynamically adjusts to evolving and unpredictable demand without requiring manual sizing, cluster management or capacity planning. It is the “tip of the spear” for performance, hardware and software innovation within the Snowflake compute portfolio, which also includes:
Gen2 Warehouses: Provide predictable, high-performance execution for steady-state analytics and production workloads with familiar sizing and multi-cluster controls.
Interactive Warehouses: Built for sub-second, high-concurrency real-time analytical use cases like real-time dashboards and data-backed APIs.
Snowpark-Optimized Warehouses: Provide memory-intensive compute for ML training, large-scale transformations and data science workloads.
Migrating a classic warehouse to an Adaptive Warehouse is a zero-downtime process. An Adaptive Warehouse provides you the same logical grouping of queries, but with fewer parameters. System defaults help you quickly get started without needing to do much tweaking; operational tools remain the same and continue to work as expected, making for a smooth transition.
Adaptive Compute is designed for exceptional performance and ease of use. Simply create a warehouse and point your workloads at it; Snowflake handles the resource allocation and scaling and query routing against a shared pool of compute in your account, and continuously assesses performance to enhance query speed and throughput. The Adaptive Warehouse can adapt to your queries in real time, determining and allocating the compute and software resources each query needs on the fly.
The result is a unified, fully managed experience for teams that want better performance and throughput with reduced operational overhead, compared to other compute options:
Hyperscaler-native warehouses offer ecosystem breadth, but organizations often find that mixed workloads require additional services and more involved configuration to effectively use and manage compute.
Custom-built lakehouse stacks offer flexibility but can demand significant engineering investment, ongoing tuning and operational maintenance.
Specialized engines deliver strong performance for individual workloads (for example, ML or real-time analytics) but may introduce fragmented architectures, data movement and governance overhead.
High performance without guesswork
Adaptive Compute incorporates the latest hardware and performance enhancements, demonstrating meaningful performance gains (based on TPC-DS and internal benchmarks) over standard Snowflake compute, both Gen1 and Gen2, across workloads:
- Up to 1.6x faster for analytical workloads, such as exploratory analytics, data science and ad hoc analytics
- Up to 2.2x higher throughput (queries/hour) for highly concurrent operational analytics workloads
- Up to 3.5x faster execution for DML-heavy workloads such as data transformations, ingestion and data pipelines
Adaptive Compute replaces a fixed compute engine with one that dynamically responds to the performance level your workloads actually require. You still have management parameters that provide guardrails to the system to meet your workload, performance and pricing requirements.
This intelligent scaling is especially critical when dealing with mixed environments with variable workloads. It enables Adaptive Compute to shorten the path to action for both technical and business teams:
Supporting rapid innovation and exploration of new use cases by reducing compute constraints and performance tradeoffs
Facilitating monetization of data and AI initiatives with performance that scales to match business demand, without adding operational risk
Combined with a query-based billing model and continuous delivery of the latest hardware and software enhancements, Adaptive Warehouses can run significantly more queries at a similar cost to Gen2.
Exceptional ease of use
Every manual compute configuration decision carries risk. What is the optimal warehouse size for my project? Will the multi-cluster policies that worked last quarter fit this quarter's workload mix? How much monitoring will my query acceleration settings need to stay effective?
Adaptive Compute removes these decisions from your engineering team. Users simply set two parameters (Maximum Query Performance Level and Query Throughput Multiplier) and Snowflake takes on the work of finding the best compute configuration for each query. Cost governance also works as you would expect, using similar budgets, resource monitors and showback/chargeback mechanisms as with standard warehouses. You retain full visibility and control over spend while Snowflake optimizes how that spend translates into performance and underlying compute.
“At Observe by Snowflake, we manage a large fleet of over a thousand Snowflake warehouses with various sizes to find the most price-performant way to serve our customer’s queries. We have built a very sophisticated scheduler to manage these warehouses to achieve high performance and efficiency, while keeping the interactive query’s latency down. Our tests have shown that with just a handful of Adaptive Warehouses, we can achieve significantly better performance — up to 30% reduction in query latency — with comparable cost. We are very happy to outsource our complex scheduling logic to Snowflake and are actively adopting Adaptive so that we can focus on building observability products for our customers.”
Gabriel Tavridis
Adaptive Compute delivers business outcomes: Four use cases
The one-two punch of performance and ease of use delivers benefits across the enterprise. Here are four examples of how Adaptive Warehouses can drive measurable results.
Mixed analytics workloads
Goal: Support concurrent BI dashboards, data exploration and analysis, and ad hoc queries with usage that fluctuates throughout the day as well as week to week.
The Adaptive Compute advantage: Maintains consistent performance across workloads by delivering the right amount of compute at the right time.
Data loading pipelines
Goal: Consistent ingestion speeds across data types and sources.
The Adaptive Compute advantage: Dynamic scheduling matches ingestion parallelism; automatic resource management supports consistent performance without manual tuning.
AI experimentation
Goal: Manage unpredictable training cycles, feature engineering and model iteration.
The Adaptive Compute advantage: Automatically scales during intensive compute bursts to support and accelerate model development and AI iteration while controlling costs and preventing infrastructure bottlenecks.
Mixed BI + ETL workloads
Goal: Handle diverse, complex, unpredictable workloads simultaneously without constant resizing or triage latency.
The Adaptive Compute advantage: Per-query scheduling handles bursty, varied query shapes better than static sizing.
Streaming analytics
Goal: Process workload spikes driven by real-time events and streaming data, such as fraud alerts and Internet of Things (IoT) signals.
The Adaptive Compute advantage: Dynamically adjusts compute to maintain low latency and high throughput for time-sensitive data.
Get started now: How to create an Adaptive Warehouse
You can create an Adaptive Warehouse using Snowsight interface (UI), SQL or Cortex Code. To create an Adaptive Warehouse using Snowsight, follow these five steps:
Sign in to Snowsight
In the navigation menu, select Compute » Warehouses
Select +Warehouse
In the Type dropdown, select Adaptive
Optionally, select Advanced and configure:
Maximum query performance level (default: XLarge)
Query throughput multiplier (default: 2)
The warehouse is now ready and can be used normally. Adaptive Warehouse is currently available in three AWS regions (US, EU and APAC) with rollout continuing as capacity is needed across regions.
Adaptive Compute: The next generation of compute
Whether you're scaling AI initiatives, consolidating enterprise analytics or building data-powered applications, Adaptive Compute provides two capabilities you need to move faster with confidence:
High performance that dynamically adapts to evolving workloads and scales seamlessly with business demands, enabling systems to deliver consistent analytics and insights for faster action and improved efficiency
Ease of use with a fully managed compute service that scales resources to avoid overspending and removes infrastructure decision-making and manual optimization to free up engineers for more strategic work
Learn more about how Adaptive Compute can benefit your organization by visiting the Adaptive Compute page and attending the Adaptive Compute session at Snowflake Summit 2026 on Wednesday, June 3, at 12:30 p.m. PDT.
This content 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.
Snowflake for AI: Putting Intelligence Where Work Happens
Snowflake CoWork: The Personal Work Agent for Every Knowledge Worker
Snowflake CoCo: The Coding Agent Built for Data Now Available Where You Work
