Union

Name: Union
Brand: Union

AI Native

The enterprise Flyte platform for AI development and orchestration.

Est 2021|Valuation $40M|Raised $53.1M|15 Ref Customers

LinkedIn 9K|Web traffic 9K|FTEs 43|HQ Bellevue, WA

MarketsWorkflow Orchestration

01Competitive Landscape

Narrative analysis

- Union versus managed Airflow (Astronomer): Vendor comparisons frame Airflow as less suited to dynamic AI workloads and emphasize differences in caching, secrets, and RBAC. Buyers should validate whether their workflows require Flyte-style dynamic execution or whether an Airflow-compatible approach meets their needs. - Union versus Dagster/Prefect: Dagster and Prefect compete on developer experience and integration ecosystems. Union emphasizes ML-specific scaling, Kubernetes-native execution, and infrastructure-aware orchestration. Public third-party review evidence for Union is smaller compared with these competitors in the sources reviewed.

Market share and adoption trends

No independent market share statistics were found in the reviewed sources. Press coverage confirms fundraising and product launches but does not quantify market share.

	Union.ai	Astronomer (Astro)	Dagster+
Starting Price	Team: monthly subscription plus usage (entry-level monthly subscription in the low-to-mid range)	Developer deployments billed on an hourly basis	Small monthly tiers available for individuals and small teams; higher tiers for teams
Primary Use Cases	ML/AI orchestration, Kubernetes-native workflows, scaling training/inference pipelines	Teams standardized on Airflow needing managed operations and enterprise controls	Data teams adopting asset-based orchestration with developer-oriented ergonomics
Key Strengths	Flyte-based durability, ML-oriented workflow patterns, support for serverless and BYOC execution models	Mature Airflow ecosystem, managed hosting, transparent usage pricing	Lower entry cost for small teams; asset abstractions for lineage and testing
Key Weaknesses	Limited independent review footprint; relatively high minimum entry cost; public performance and benchmark claims lack published methodology	Airflow may be less tailored to ML-native dynamic workflows	Conceptual model differences may require adaptation; credits/pricing can be non-intuitive at scale

01Competitive Landscape

Narrative analysis

Union versus managed Airflow (Astronomer): Vendor comparisons frame Airflow as less suited to dynamic AI workloads and emphasize differences in caching, secrets, and RBAC. Buyers should validate whether their workflows require Flyte-style dynamic execution or whether an Airflow-compatible approach meets their needs.
Union versus Dagster/Prefect: Dagster and Prefect compete on developer experience and integration ecosystems. Union emphasizes ML-specific scaling, Kubernetes-native execution, and infrastructure-aware orchestration. Public third-party review evidence for Union is smaller compared with these competitors in the sources reviewed.

Market share and adoption trends

No independent market share statistics were found in the reviewed sources. Press coverage confirms fundraising and product launches but does not quantify market share.

	Union.ai	Astronomer (Astro)	Dagster+
Starting Price	Team: monthly subscription plus usage (entry-level monthly subscription in the low-to-mid range)	Developer deployments billed on an hourly basis	Small monthly tiers available for individuals and small teams; higher tiers for teams
Primary Use Cases	ML/AI orchestration, Kubernetes-native workflows, scaling training/inference pipelines	Teams standardized on Airflow needing managed operations and enterprise controls	Data teams adopting asset-based orchestration with developer-oriented ergonomics
Key Strengths	Flyte-based durability, ML-oriented workflow patterns, support for serverless and BYOC execution models	Mature Airflow ecosystem, managed hosting, transparent usage pricing	Lower entry cost for small teams; asset abstractions for lineage and testing
Key Weaknesses	Limited independent review footprint; relatively high minimum entry cost; public performance and benchmark claims lack published methodology	Airflow may be less tailored to ML-native dynamic workflows	Conceptual model differences may require adaptation; credits/pricing can be non-intuitive at scale