Snowflake Enters New Era: Adaptive Compute and Cortex Code Generally Available
Snowflake Enters New Era: Adaptive Compute and Cortex Code Generally Available
Summary
Snowflake has announced major platform updates: the workload-aware resource “Adaptive Compute” is now generally available (GA), alongside the integrated AI coding assistant “Cortex Code”. Additionally, the native real-time ingestion service “Snowflake Datastream” has entered Private Preview. Together, these enhancements significantly reduce the complexity of manual performance tuning and accelerate the creation of modern data pipelines for AI-ready workloads.
What happened?
- Adaptive Compute GA: Snowflake has launched the general availability of Adaptive Compute on Amazon Web Services (AWS). Users can deploy new Adaptive Warehouses or seamlessly transition existing standard warehouses without experiencing downtime.
- Cortex Code GA: The native AI coding agent Cortex Code is now generally available. It helps data and analytics teams write SQL, Python, and dbt pipelines directly within Snowsight or via a CLI.
- Snowflake Datastream (Preview): A native, Kafka-compatible streaming ingestion service is now available in Private Preview, allowing continuous streaming data to ingest directly into Snowflake without external middleware.
Why it matters
Historically, managing data architectures was constrained by two main factors: the burden of manual warehouse tuning (sizing, multi-clustering, Query Acceleration parameters) and the lack of context-aware AI tools for data engineering. Adaptive Compute removes the manual configuration process entirely for unpredictable workloads. Cortex Code distinguishes itself from generic LLMs by grounding its generation in the specific schemas, metadata, roles, and privileges of the user’s Snowflake account.
Evidence
These product updates are documented in recent announcements and releases:
- Snowflake’s release notes confirm the GA status of Adaptive Compute on AWS as of June 16, 2026.
- The agenda and presentation details for the “Snowflake Connect” event on June 17, 2026, highlight the integration of Cortex Code with dbt, Python, and SQL pipelines.
Analysis
The launch of Adaptive Compute signals a shift from classic serverless models to autonomous compute optimization. By requiring only two configuration parameters (MAX_QUERY_PERFORMANCE_LEVEL and QUERY_THROUGHPUT_MULTIPLIER), Snowflake shifts scaling decisions to its platform engine. This prevents high TCO caused by overprovisioned or misconfigured multi-cluster warehouses. Concurrently, Datastream addresses a key infrastructure pain point by consolidating real-time streaming pipelines, eliminating the overhead of external Kafka Connect deployments.
Practical Takeaways
- Deploy Adaptive Warehouses: Data infrastructure teams should migrate variable, spiky workloads to Adaptive Warehouses to reduce performance tuning overhead and lower costs.
- Leverage Cortex Code: Developers and engineers should adopt Cortex Code within Snowsight or their terminals to accelerate dbt and SQL development while staying aligned with account security privileges.
- Explore Snowflake Datastream: Organizations running Apache Kafka architectures should apply for the Snowflake Datastream Private Preview to simplify real-time ingestion stacks.
Open Questions
- What is the timeline for Adaptive Compute general availability on Microsoft Azure and Google Cloud Platform (GCP)?
- How will the automated TCO of Adaptive Compute compare to highly optimized, manually tuned standard warehouses for consistent, predictable workloads?