Azure Synapse Analytics and Microsoft Fabric cover overlapping analytical workloads with fundamentally different commercial models. Synapse bills per-resource — Dedicated SQL Pools at $1.51/DWU100/hour, Spark Pools at $0.072–$0.116/vCore-hour, Data Pipelines at $0.25/DIU-hour plus per-run charges. Fabric bills via a shared CU capacity pool where all workloads — SQL, Spark, pipelines, real-time analytics — draw from the same monthly commitment. For most organisations running mixed analytical workloads, Fabric's consolidated model delivers 25–40% lower total cost vs equivalent Synapse spend. But the migration complexity is real and varies sharply by workload type.
Independent Advisory. Zero Vendor Bias.
500+ Microsoft EA engagements. $2.1B in managed spend. 32% average cost reduction. We negotiate on your behalf — never Microsoft's.
View Advisory Services →The Platform Landscape: What Each Service Covers
Azure Synapse Analytics launched in 2020 as a unified analytics platform combining data warehousing (Dedicated SQL Pool), big data processing (Spark Pools), data integration (Synapse Pipelines), and exploratory analytics (Serverless SQL Pool). It solved real problems for organisations running multiple Azure data services and remains a capable platform.
Microsoft Fabric, launched in 2023, absorbed and extended Synapse's functionality while replacing the billing model. Fabric's components map directly to Synapse's but with architectural improvements and a unified OneLake storage layer that eliminates data duplication across services.
| Synapse Component | Fabric Equivalent | Migration Complexity |
|---|---|---|
| Synapse Dedicated SQL Pool (DW) | Fabric Warehouse | High — T-SQL differences, concurrency model change |
| Synapse Serverless SQL Pool | Fabric SQL Analytics Endpoint | Low — largely equivalent |
| Synapse Spark Pools | Fabric Spark (Data Engineering) | Moderate — runtime and library compatibility |
| Synapse Pipelines | Fabric Data Factory | Low to Moderate — most activities supported |
| Synapse Analytics Workspace | Fabric Workspace | Conceptual change — no direct migration tool |
| Azure Data Lake Storage Gen2 | OneLake (with ADLS shortcuts) | Low — shortcut preserves existing ADLS access |
| Power BI (in Synapse Studio) | Power BI (native in Fabric) | None — same service, deeper integration in Fabric |
| Synapse Real-Time Analytics (KQL) | Real-Time Intelligence (Eventhouse) | Moderate — KQL compatible but architecture differs |
Billing Model Comparison
The structural difference between Synapse and Fabric billing is the difference between per-resource metering and capacity pool metering. Understanding this distinction is essential before comparing costs.
Synapse Billing Model
Every Synapse resource has independent billing: Dedicated SQL Pools pause automatically when idle (saving DWU cost) but still incur storage charges ($23/TB/month for Dedicated DW storage). Spark Pools charge per vCore-hour only during job execution. Pipelines charge per activity run plus DIU-hours. This model rewards single-workload optimisation but creates unpredictable costs for multi-workload analytics platforms.
| Synapse Resource | Billing Unit | Rate (East US) | Typical Monthly Cost |
|---|---|---|---|
| Dedicated SQL Pool (DW100c) | DWU100/hour | $1.51 | $1,094 (24/7) |
| Dedicated SQL Pool (DW500c) | DWU100/hour | $1.51 × 5 | $5,472 (24/7) |
| Spark Pool (Small, 4 vCores) | vCore-hour | $0.072/vCore | $207 (8h/day weekdays) |
| Spark Pool (Large, 16 vCores) | vCore-hour | $0.072/vCore | $828 (8h/day weekdays) |
| Data Pipelines (activity runs) | Per run + DIU-hour | $0.001/run + $0.25/DIU-hr | $200–$2,000 (varies) |
| Serverless SQL Pool | Per TB scanned | $5/TB | $150–$1,500 (varies) |
Fabric Billing Model
Fabric workloads — Warehouse, Spark, Data Factory, Real-Time Intelligence, Power BI — all consume from a shared capacity pool measured in Capacity Units (CUs). The F SKU tier determines the total CU pool available per month. Workloads compete for CUs from this shared pool, and the smoothing model allows 10-minute burst windows before throttling activates.
| Fabric SKU | CU Pool | PAYG (/month) | 3-yr Reserved (/month) | Power BI Equivalent |
|---|---|---|---|---|
| F2 | 2 CU | $263 | $170 | Dev/test only |
| F4 | 4 CU | $527 | $340 | Dev/test only |
| F8 | 8 CU | $1,054 | $681 | Limited |
| F16 | 16 CU | $2,108 | $1,361 | None (below P1) |
| F32 | 32 CU | $4,217 | $2,722 | None (below P1) |
| F64 | 64 CU | $8,434 | $5,980 | P1 equivalent |
| F128 | 128 CU | $16,869 | $11,961 | P2 equivalent |
Key insight: A DW500c Synapse Dedicated Pool running 24/7 costs $5,472/month. An F64 Fabric capacity ($5,980/month reserved) provides the equivalent data warehouse compute plus Spark processing, Data Factory pipelines, Real-Time Intelligence, and Power BI Premium. For any organisation using more than one Synapse service, Fabric's consolidated billing almost always wins on cost.
Total Cost Comparison: Three Organisations
Organisation A: Data Warehouse Dominant
1,200-seat financial services firm. Architecture: DW500c Synapse SQL Pool (24/7) + 2× Large Spark Pools (6h/day weekdays) + Synapse Pipelines (active workload, ~$400/month) + Serverless SQL ($300/month scans). Monthly Synapse total: $5,472 + $1,656 + $400 + $300 = $7,828/month. Equivalent Fabric F128 3-year reserved: $11,961/month. Verdict: Synapse cheaper at this workload profile — but only because DW500c is severely underutilising Spark. Adding Spark scale-out or Real-Time Intelligence shifts the calculation.
Organisation B: Mixed Analytical Platform
2,500-seat retail firm. Architecture: DW300c SQL Pool + 4× Medium Spark Pools (8h/day) + Synapse Pipelines ($800/month) + Serverless SQL ($800/month) + separate Power BI Premium P1 ($4,995/month). Monthly total: $3,283 + $2,764 + $800 + $800 + $4,995 = $12,642/month. Equivalent Fabric F128 3-year reserved: $11,961/month. Saving: $681/month ($8,172/year) — plus all Fabric capabilities included.
Organisation C: Real-Time + BI Platform
4,000-seat manufacturer. Architecture: Synapse Pipelines ($1,200/month) + Azure Stream Analytics ($2,800/month) + Power BI Premium P2 ($9,990/month) + Serverless SQL ($600/month). Monthly total: $14,590/month. Equivalent Fabric F128 with Real-Time Intelligence: $11,961/month. Saving: $2,629/month ($31,548/year) — and Stream Analytics replaced by Fabric Real-Time Intelligence at no incremental cost.
Migration Complexity: The Real Cost
Cost savings calculations are only meaningful if migration is achievable. Here is an honest assessment of migration effort by workload:
Synapse Pipelines → Fabric Data Factory
Microsoft provides a migration wizard that converts most Synapse Pipeline definitions to Fabric Data Factory equivalents. Coverage is approximately 85% of common activity types. Custom connectors, .NET activities, and some on-premises integration runtime configurations require manual rework. Expected migration timeline for 50–200 pipelines: 4–8 weeks with testing. Low-risk migration.
Synapse Spark → Fabric Spark
Fabric Spark uses an updated runtime (currently Spark 3.4/3.5) vs Synapse Spark (3.1–3.3). PySpark, Scala, and SparkSQL notebooks generally migrate with minimal changes. Key watch-outs: Delta Lake version alignment (Synapse uses older Delta protocol by default), library versions (Synapse packages may not be available in Fabric's MSSparkUtils environment), and native Synapse-specific APIs (SynapseLink, MSSparkUtils Synapse context). Expected migration timeline for 20–50 notebooks: 3–6 weeks. Moderate complexity.
Synapse Dedicated SQL Pool → Fabric Warehouse
This is the highest-risk migration path. Fabric Warehouse is not a direct lift-and-shift of Synapse Dedicated SQL Pool. Key differences: Fabric Warehouse uses a standard SQL engine (not MPP like Dedicated SQL Pool), concurrency is managed differently (CU-based vs DWU-based), and some T-SQL syntax specific to Synapse DW (like CTAS with distribution hints, materialised views syntax) requires modification. For workloads exceeding 1 TB of data and high concurrency requirements, conduct a formal POC before committing to migration. Expected timeline for complex EDW: 3–9 months.
Get an Independent Second Opinion
Before committing to a Synapse-to-Fabric migration, speak with an adviser who has assessed both platforms across 500+ enterprise environments.
Request a Consultation →Decision Framework: When to Migrate vs Stay
| Scenario | Recommendation | Rationale |
|---|---|---|
| Using Synapse Pipelines + Power BI Premium | Migrate to Fabric now | Fabric consolidation saves P SKU + pipeline costs immediately |
| Synapse Dedicated DW only (<500 DWUs, complex T-SQL) | Pilot Fabric Warehouse first | Validate T-SQL compatibility before full migration commitment |
| Synapse DW + real-time analytics requirement | Migrate — Fabric is the only Azure native option | Real-Time Intelligence is Fabric-only; staying on Synapse requires Stream Analytics at additional cost |
| Large Synapse Dedicated DW (1,000+ DWUs, high concurrency) | Architecture review required | Fabric Warehouse may not match Synapse DW performance at scale — validate POC |
| Synapse + Databricks hybrid | Evaluate Fabric vs Databricks migration | Databricks has competitive features and often better Spark pricing at enterprise scale |
| Greenfield analytics platform | Build on Fabric | No migration cost, full Fabric capabilities, OneLake from day 1 |
EA and MACC Considerations
Synapse Dedicated SQL Pool and most Synapse services count toward Azure MACC commitment. Microsoft Fabric F SKU also counts toward MACC. When migrating, the total Azure spend footprint should remain MACC-eligible — but the mix of services changes, which can affect which MACC tier you qualify for and when.
A migration from $12,000/month Synapse to $11,961/month Fabric technically reduces your Azure MACC contribution by $39/month — trivial, but worth noting if your MACC pacing is tight. If you are adding new Fabric capabilities (eliminating Stream Analytics, consolidating separate Power BI Premium), the MACC contribution will likely increase net of migration.
For large EA customers, the right play is to negotiate the Synapse-to-Fabric migration as a MACC expansion event. "We are migrating $84,000/year of Synapse workloads to Fabric and will add $24,000/year of new Fabric workloads — what MACC acceleration pricing can you provide?" Microsoft's Azure commercial team has pricing discretion for MACC expansions and has applied 15–25% discounts in competitive situations.
📄 Free Guide: Microsoft Fabric Licensing Guide 2026
Complete guide to Fabric F SKU model, Synapse migration, capacity sizing, OneLake costs, and EA negotiation strategy.
Download Free Guide →Frequently Asked Questions
Is Azure Synapse Analytics being discontinued?
Microsoft has not announced end-of-life for Azure Synapse, but active feature development has largely shifted to Microsoft Fabric. Synapse Dedicated SQL Pool, Spark Pools, and Data Pipelines remain supported. Microsoft's migration documentation and field team incentives strongly point toward Fabric as the strategic direction for new workloads.
What is the Fabric equivalent of Synapse Dedicated SQL Pool?
Microsoft Fabric Warehouse is the closest equivalent, using the same T-SQL surface area. Key differences: Fabric Warehouse uses CU-based billing on Fabric capacity (no separate DWU billing), supports Delta/Parquet natively, and does not require pre-allocation of compute. For large existing Synapse DW workloads, validate Fabric Warehouse carefully — some high-concurrency workloads require significantly larger F SKU than their Synapse DWU equivalent.
Can I use Azure Synapse and Microsoft Fabric simultaneously?
Yes. Many organisations run Synapse for legacy EDW workloads while migrating new analytical workloads to Fabric. Both can connect to the same Azure Data Lake Storage Gen2 via Fabric OneLake shortcuts. This hybrid approach avoids a disruptive cutover but creates duplicate licensing cost — plan a migration timeline with clear workload handoff milestones.
How does Synapse billing compare to Fabric billing?
Synapse bills per-resource independently (DWU-hours, vCore-hours, activity runs, TB scanned). Fabric bills via a shared Capacity Unit pool — all workloads consume from the same F SKU monthly commitment. Fabric's model rewards diverse mixed workloads; Synapse's model rewards dedicated high-throughput single workloads.
What is the migration effort from Synapse to Fabric?
Pipelines: low effort (85% automated conversion). Spark notebooks: moderate (runtime version differences, library compatibility). Dedicated SQL Pool → Fabric Warehouse: high effort for complex workloads (T-SQL compatibility testing, concurrency model differences, data volume validation required). Plan 3–9 months for complex EDW migrations.
Related Microsoft Fabric & Analytics Guides
- Microsoft Fabric Licensing: Complete Enterprise Guide
- Microsoft Fabric F SKU vs P SKU: Complete Comparison
- Microsoft Fabric Capacity Planning: CU Sizing & Monitoring
- OneLake Licensing & Storage Cost Optimisation
- Azure Cost Optimisation: Complete Enterprise Guide
- Azure Reserved Instances: Enterprise Optimisation Guide
- Azure MACC Negotiating Leverage: Complete Guide