Apache Airflow remains centered on coordinating workflows: defining tasks, deciding when they run, tracking their state, and responding when they fail. ETL and ELT are still central uses, while Airflow 3 announcements point toward more asset-aware scheduling, broader task-authoring options, and stronger support for multi-team operations. Those announcements show the project’s direction—not proof that Airflow is the best choice for every pipeline or workload.
What Airflow does—and what it does not
Airflow lets teams author, schedule, and monitor workflows, commonly represented as directed acyclic graphs (DAGs). A DAG describes tasks and their dependencies: for example, extract weather data, transform it, then load it for a dashboard. Airflow coordinates that sequence and tracks its progress; the work of processing data generally belongs in the tasks and systems the workflow calls, rather than in the orchestration layer itself.
As an Amazon Associate I earn from qualifying purchases.
This distinction matters when evaluating Airflow. The orchestrator is responsible for sequencing work and handling workflow execution, but the underlying tasks still need appropriate compute, storage, and application logic. The Apache Airflow use-case material presents data extraction, transformation, and loading as a core pattern.
ETL and ELT remain central, but the figures need context
Airflow’s official ETL/ELT page reports that 90% of respondents to its 2023 survey used Airflow for ETL/ELT analytics. That is a result from the 2023 survey, not a measurement of Airflow users in 2026 or a count of all organizations using the software.
#1 Best Overall
The project’s January 2026 announcement of the 2025 survey reports more than 5,818 responses from 122 countries. Those figures describe participation in a community survey; they do not establish market share, and response volume alone cannot show that respondents represent all data engineers. The announcement is available on the Apache Airflow Survey 2025 page.
What Airflow 3 announcements say about its direction
Airflow 3.0 became generally available on April 22, 2025. In that announcement, the project described Airflow as expanding beyond ETL/ELT toward MLOps and generative AI workflows. Later Airflow 3 release announcements signal several related priorities. The project blog listing includes an Airflow 3.3.0 announcement dated July 6, 2026; the features below should be understood as announced capabilities, not as an independent assessment of their adoption or impact.
Rank #2
| Announced direction | What it could change for workflow teams | What the announcement establishes |
|---|---|---|
| Asset-aware orchestration | Workflow decisions can be organized around data assets and their partitions, rather than only around a fixed schedule or task sequence. This is relevant when downstream work should depend on which data is ready. | Airflow 3.2 announced asset partitioning; the 3.3 announcement described expanded asset partitioning and a first-class state store for tasks and assets. |
| More task-authoring options | Teams may have more flexibility when their task code is written in languages already used in their services or data systems. | The Airflow 3.3 announcement described a Language Task SDK for Java and Go. That does not by itself establish that every provider, integration, or existing DAG can be moved to those SDKs unchanged. |
| Multi-team operation | Deployment and ownership practices can matter more as separate teams share an orchestration environment. Team boundaries and operational responsibility still need to be designed. | The Airflow 3.2 announcement described support for multi-team deployments; it does not provide a comparative benchmark of team-scale operations. |
| Failure and deadline handling | Retry behavior and deadline alerts affect how teams detect and recover from delayed or failed work. | The 3.3 announcement described pluggable retry policies, while the 3.2 announcement described synchronous deadline alert callbacks. The announcements do not quantify reliability improvements. |
Airflow’s 2025 project announcement also reported more than 30 million monthly downloads and 80,000 organizations using Airflow, along with over 30% of Airflow users using it for MLOps and 10% for generative AI workflows. These are figures reported by the project in 2025; the announcement’s terms and collection methods differ from the survey response count and the 2023 ETL/ELT result. They should not be combined, treated as independently verified market estimates, or used to infer market share.
How to judge whether the new direction fits your pipelines
Feature announcements are most useful when mapped to a concrete workflow need. Evaluate the shape of your pipelines, who owns them, and how work is executed and recovered—not just the number of integrations or the breadth of a release announcement.
- Workflow triggers: Determine whether your work is primarily scheduled, triggered by data availability, or a mix. Asset-aware orchestration is most relevant when readiness or partitions of data should drive downstream work.
- Task execution: Identify where computation runs, which languages your teams use, and what isolation tasks require. A language SDK is an authoring option, not a substitute for checking execution boundaries and integration compatibility.
- Team ownership: Clarify who deploys DAGs, manages shared infrastructure, handles upgrades, and responds to incidents. A multi-team capability does not remove the need for operating rules and clear ownership.
- Recovery and observability: Check whether your workflows need configurable retry behavior, deadline alerts, useful state history, and a practical recovery path for partial failures.
- Migration cost: Inventory existing DAGs, providers, operational practices, and dependencies in the surrounding data platform before changing versions or execution patterns.
The available Airflow announcements establish that these are areas of active development, but they do not provide a like-for-like benchmark against other orchestrators. A claim that Airflow is the universal “go-to” choice would go beyond that evidence.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the evidence supports about Airflow’s current state
The clearest conclusion is that Airflow remains a workflow orchestration platform with ETL/ELT at its center, while its project is developing capabilities for data-aware triggering, broader task authoring, and team-scale operation. Survey and adoption figures offer context, but their dates, definitions, and methods differ. For a decision about a particular deployment, check the release notes for the version under consideration and test compatibility with the DAGs, providers, and operating model you actually use.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
