Data platforms, side by side.

Data + AI Platform (lakehouse)

Cloud data warehouse / AI Data Cloud

Serverless warehouse; Lakehouse (was BigLake)

SaaS suite over lakehouse + warehouse

Apache Spark + Photon

Proprietary vectorized SQL engine

Dremel (proprietary engine)

Spark for engineering; T-SQL warehouse engine

Decoupled storage and compute

Fully decoupled storage and compute

Fully decoupled (Colossus + Dremel)

Yes, compute separate from OneLake

Delta Lake (and managed Iceberg)

Proprietary micro-partitions (FDN)

Proprietary (Capacitor columnar)

Delta Lake (Parquet) in OneLake

Native managed Iceberg, read+write GA

Native Iceberg, read+write

Managed Iceberg tables, read+write

Read+write via metadata virtualization (V2)

Native Delta read/write

Read via external/Iceberg, no native write

Read-only external tables for Delta Lake

Native read/write, default format

Unity Catalog Iceberg REST catalog

Horizon Catalog Iceberg REST (Apache Polaris)

Iceberg REST catalog (Lakehouse runtime catalog)

OneLake catalog; Iceberg REST read APIs only

Spark, Delta, Unity Catalog open source

Engine closed; Polaris/Arctic open

Engine proprietary; Iceberg format open

Spark/Delta open; warehouse engine proprietary

AWS, Azure, GCP

AWS, Azure, GCP

GCP native; AWS/Azure via BigQuery Omni

Azure-only SaaS; shortcuts to AWS/GCP

Runs in your cloud account

SaaS-only managed service

SaaS-only (Google Cloud managed)

SaaS-only (Azure-hosted)

Per-DBU

Per-credit (Platform Credits + AI Credits)

On-demand per-TB or slot capacity

Capacity Units (F-SKUs)

Per-second

Per-second, 60-second minimum

Slots per-sec; 1-min minimum unless fluid scaling

Per-second billing, one-minute minimum

Serverless SQL/compute, auto-suspend

Auto-suspend warehouses + serverless compute

Serverless; editions autoscale to zero

Pause/resume capacity; on-demand Spark billing

Classic: separate VM bill; serverless bundled

Bundled; no separate cloud VM bill

Compute/storage billed by BigQuery directly

Capacity billed apart from OneLake storage

Cloud bills own storage; default storage in DSUs

~$23/TB-month on-demand (AWS US East)

~$0.02/GB-mo active; $0.01 long-term

OneLake $0.023/GB-mo hot; cool+cold tiers

Free Edition + 14-day trial

30-day free trial with credits

1 TB queries + 10 GB storage/month

60-day free trial capacity

Committed-use contracts

Pre-paid capacity discounts

1yr/3yr capacity commitments discounted

Reserved capacity ~41% off PAYG

System tables, usage dashboards, budgets

Cost views, budgets, resource monitors

Cost dashboards, slot/admin monitoring

Capacity Metrics app; cost management

List DBU prices published

List credit/storage prices published

List prices published publicly

List prices published per SKU

ANSI SQL incl. window, recursive CTE

Broad ANSI SQL, window + recursive CTE

GoogleSQL; window + recursive CTE

T-SQL warehouse; CTEs, window functions

Native VARIANT and JSON support

Native VARIANT/JSON/Avro/Parquet

Native JSON data type

JSON as varchar in T-SQL; VARIANT in Spark

Spatial SQL GA, GEOMETRY/GEOGRAPHY, H3

GEOGRAPHY + GEOMETRY types and functions

GEOGRAPHY type + GIS functions

ArcGIS GeoAnalytics partner; KQL geo

SQL, Python, Scala, Java UDFs

SQL, Python, Java, Scala UDFs

SQL, JavaScript, Python UDFs

T-SQL, Python, Spark UDFs; user data functions

Native materialized views

Native materialized views

Native materialized views

Materialized Lake Views (Delta)

Query result caching

Persisted query result cache

Automatic query result cache

Result-set cache disabled; disk/memory caching

Lakehouse Federation

External tables; limited live federation

Cloud SQL, Spanner, external sources

Shortcuts + mirroring virtualize sources

Lakeflow pipelines, Lakeflow Jobs

Snowpark, Streams, Tasks, Openflow

Dataform, Dataflow, Data Transfer Service

Data Factory pipelines + dataflows

Structured Streaming, Real-Time Mode

Snowpipe Streaming, up to 10 GB/s

Storage Write API streaming ingest

Real-Time Intelligence eventstreams

CDC via AUTO CDC APIs, Lakeflow Connect

Streams; Openflow CDC connectors

Native CDC ingest; Datastream

Copy Job CDC (preview); open mirroring GA

Auto Loader

Snowpipe auto-ingest

DTS event-driven Cloud Storage transfers

Copy Job; eventstream file ingest

Lakeflow Jobs

Native Tasks scheduler / DAGs

Scheduled queries, Dataform workflows

Data Factory pipeline scheduler

First-class dbt adapter and task

dbt Projects on Snowflake (GA); dbt Core + Fusion

First-class dbt adapter

dbt-fabric/dbt-fabricspark; dbt job (preview)

Lakeflow pipelines (Spark Declarative Pipelines)

Dynamic Tables declarative pipelines

Dataform SQL declarative pipelines

Materialized Lake Views declarative ETL

Native training on Spark/GPU clusters

Snowflake ML native training (CPU/GPU)

BigQuery ML trains models in SQL

Native notebooks + Spark training

Native feature store in Unity Catalog

Native Snowflake Feature Store

Agent Platform Feature Store on BigQuery

No dedicated managed feature store

Managed MLflow

Snowflake ML Experiments (GA Feb 2026)

Agent Platform Experiments (separate product)

Native MLflow experiment tracking

Databricks Model Serving (real-time, batch)

Model Serving on Container Services

Serving via Agent Platform endpoints

ML model endpoints (preview)

AutoML

Partial automation, no full AutoML product

AutoML models in BigQuery ML

AutoML via FLAML

Databricks AI Search (formerly Vector Search)

Native VECTOR type + Cortex Search

Native vector index + VECTOR_SEARCH

Native vector type in SQL DB; AI functions

Unity AI Gateway (multi-provider routing)

Cortex AI hosts multiple LLMs

Gemini via ML.GENERATE_TEXT/AI functions

Microsoft Foundry models; Copilot FM access

Genie Agents (formerly Genie Spaces)

Cortex Analyst NL-to-SQL

Gemini SQL generation in editor

Copilot NL-to-SQL in warehouse/notebooks

Agent Bricks, Agent Framework, MCP (preview)

Cortex Agents + managed MCP server

Data agents + BigQuery MCP server

Data agents GA; Fabric Core MCP server preview

GPU instances for ML

GPU compute pools (Container Services)

GPUs via Agent Platform, not native BigQuery

No native GPU Spark pools

AI/BI Dashboards

Streamlit apps; Snowsight dashboards retired

BI Engine; Looker Studio dashboards

Power BI dashboards built in

Unity Catalog Metric Views

Semantic views

Looker; BigQuery Graph measures (preview)

Power BI semantic models; Fabric IQ (preview)

Native notebooks

Native Snowflake Notebooks

Native notebooks in BigQuery Studio

Native Spark/Python notebooks

Genie One and Genie Agents natural language

Cortex Analyst ask-your-data

Data Canvas natural-language analytics

Copilot Q&A on Power BI reports

Tableau, Power BI, Looker connectors

Tableau, Power BI, Looker connectors

Tableau, Power BI, Looker connectors

Power BI native; Tableau via connector

Unity Catalog across data and AI

Horizon Catalog across data + AI

Knowledge Catalog (formerly Dataplex)

OneLake catalog + Purview governance

Fine-grained RBAC in Unity Catalog

Role-based access control

IAM roles, fine-grained permissions

Workspace roles + OneLake item RBAC

ABAC with governed tags, GA

Tag-based ABAC (masking GA; row access preview)

Policy tags / taxonomy-based access

Purview sensitivity labels; limited ABAC

Dynamic column masks

Dynamic data masking

Dynamic data masking via policy tags

Dynamic data masking; column-level security

Row filters

Row access policies

Row-level access policies

OneLake RLS across engines; warehouse RLS

Automatic lineage in Unity Catalog

Automatic data lineage (Horizon)

Automatic lineage via Knowledge Catalog

Automatic lineage view + Purview

Automated data classification GA

Automatic sensitive-data classification

Automatic PII discovery via Knowledge Catalog

Purview auto PII classification + labels

Audit logs / system tables

Access History / audit logs

Cloud Audit Logs for access

Purview/M365 audit logs

Customer-managed keys

Tri-Secret Secure customer-managed keys

CMEK via Cloud KMS

Customer-managed keys (workspace encryption)

PrivateLink, VNet/VPC injection

PrivateLink / Private Link networking

VPC Service Controls, Private Service Connect

Private Link + managed VNet

OpenSharing (formerly Delta Sharing), any cloud

Secure cross-cloud data sharing

BigQuery sharing / Analytics Hub

External data sharing across tenants

Clean Rooms GA

Native Data Clean Rooms

Native data clean rooms

No native Fabric clean rooms

Databricks Marketplace

Snowflake Marketplace (3,400+ listings)

Google Cloud Marketplace / data exchanges

Azure Marketplace; Fabric workload hub

Public Status API; email, webhook, Slack alerts

Statuspage with JSON status API

Google Cloud status dashboard published

Service Health in Fabric/M365 admin portals

99.9% control plane SLA; Lakebase SLA published

99.99% / 99.9% monthly availability SLA

99.99% uptime SLA published

99.9% uptime SLA

Cluster autoscaling

Multi-cluster elastic auto-scaling

Autoscaling slots in editions

Spark autoscale pools; on-demand Spark billing

Managed DR (gated preview) via Mission Critical

Cross-region replication and failover

Cross-region dataset replication, multi-region

Availability zones; multi-geo; BCDR

Separate warehouses/clusters per workload

Separate virtual warehouses per workload

Reservations isolate ETL vs BI

Separate capacities per workload

Lakeflow Connect 100+ sources

Openflow + broad partner connectors

Data Transfer Service + partner connectors

170+ Data Factory connectors

SOC 2, HIPAA, PCI-DSS, FedRAMP, ISO

SOC 2, HIPAA, PCI, FedRAMP, ISO

SOC 2, HIPAA, ISO, PCI, FedRAMP

SOC 2, HIPAA, FedRAMP, ISO, PCI

Dozens of regions across AWS/Azure/GCP

Global regions across AWS/Azure/GCP

40+ Google Cloud regions

Azure public regions worldwide

Tiered support plans

Snowflake Support (ex Premier); Priority Support

Standard, Enhanced, Premium support

Fabric subscription, Unified, Premier plans