Why Microsoft Fabric for Financial Services Now
Microsoft Fabric (GA late 2024) is the most consequential data platform release for financial services since the cloud data warehouse. For risk + finance + surveillance + reg reporting workloads, Fabric replaces the fragmented Teradata + Oracle + Hadoop + Snowflake + Databricks + Synapse stacks that most banks + asset managers run today. The convergence is driven by: OneLake unified storage eliminating data movement between BI / ML / warehouse workloads, Direct Lake on Power BI removing import-refresh delays critical for intraday risk dashboards, native Microsoft Purview integration providing catalog + lineage + classification at the data-platform layer, Sentinel integration for audit + compliance, and Copilot for Fabric bringing natural-language queries to risk + finance analysts.
EPC Group's Fabric FSI practice is built on Errin O'Connor's Lead Architect tenure at the Federal Reserve Bank of New York + 4× Microsoft Press Power BI / data platform authorship. The firm has migrated Fortune 500 banks + asset managers from Snowflake + Teradata to Fabric.
Reference Architecture
Ingestion. Trade + order + market data via Eventstream (real-time). Position + balance + ledger data via Data Factory (batch). Customer + counterparty + sanctions data via Data Factory. Communication data (email + chat + voice transcription) via Microsoft Purview connectors for surveillance workloads. Reference data (instruments, counterparties, market data hierarchies) via custom ingestion or third-party reference data providers.
Storage. OneLake with bronze (raw) / silver (cleansed + validated) / gold (analytics-ready) medallion architecture. Delta Lake format throughout. Separate workspaces per regulatory domain (risk, finance, surveillance, AML, treasury) with Microsoft Information Barriers preventing cross-domain access where ethical walls apply.
Compute. Fabric Lakehouse + Warehouse + Real-Time Analytics + Notebooks + Data Science workloads. Notebooks in Python + Spark for risk modeling (VaR, expected shortfall, PD/LGD/EAD, FRTB SBM/IMA, stress testing). Warehouse for regulatory reporting (FRTB, OATS/CAT, MiFID II). Real-Time Analytics for surveillance + intraday risk. Data Science for AML ML models, fraud detection, churn prediction.
Serving. Power BI semantic models via Direct Lake. Row-level security + object-level security for ethical-wall enforcement. Copilot for Power BI for natural-language risk + finance queries (with appropriate Restricted Search + IB configuration for MNPI workspaces).
Governance. Microsoft Purview catalog + lineage + classification across the entire Fabric estate. Sensitivity labels for MNPI + customer PII + trade secret content. Audit logs to Sentinel.
High-Value Use Cases
Portfolio risk aggregation. Position-level + counterparty-level + market-data-level aggregation across asset classes (equities, fixed income, derivatives, FX, commodities, structured products). Real-Time Analytics for intraday risk. Lakehouse for end-of-day risk + stress testing + scenario analysis. Power BI semantic models on top for risk committee reporting + regulatory submissions (Basel III RWA, CCAR / DFAST stress testing, ICAAP).
Trade surveillance. Trade + order + market data + communication ingestion into Real-Time Analytics + Lakehouse. Custom KQL + Spark analytics for spoofing, layering, front-running, wash trading, insider trading patterns. Integration with NICE Actimize, NASDAQ SMARTS, or custom rule sets. Documented as the surveillance system of record for the firm's WSPs.
Regulatory reporting. Consolidated regulatory reporting platform for FINRA OATS / CAT, MiFID II / MiFIR, SFTR, EMIR, FRTB capital reporting, LIBOR transition, CCAR / DFAST. Fabric Warehouse as the regulatory-grade store of record with audit-quality lineage via Microsoft Purview.
Counterparty exposure + credit risk. Counterparty-level exposure aggregation across products + entities + jurisdictions. Margining + collateral management. Pre-trade + post-trade credit risk. Default risk modeling (PD + LGD + EAD) via Fabric Notebooks + Azure ML.
AML transaction monitoring. Customer + transaction + counterparty + sanctions data ingestion. Rule sets + ML models for suspicious activity detection. Case management integration with operational AML platforms. Documented as the AML data layer for the firm's BSA / AML program. SAR drafting via Copilot Studio agent.
Engagement Investment
Foundation ($250K-$500K, 16-24 weeks): Single-workload Fabric implementation — risk OR surveillance OR regulatory reporting OR AML. F64 or F128 capacity. ~50 user pilot.
Enterprise ($550K-$1.4M, 28-44 weeks): Multi-workload + full data platform + Center of Excellence + Managed Microsoft Support. F128 or F256. Mid-size bank or asset manager.
Platform ($1.4M-$4M, 44-72 weeks): Enterprise + Microsoft Cloud for Financial Services + multi-entity federation + DR + audit support across multiple jurisdictions. F256+. Large bank, GSE, large insurance carrier.
Related Pages
FAQ
How does Microsoft Fabric compare to Snowflake + Databricks for financial services?
Fabric is the modern Microsoft replacement for fragmented Teradata + Oracle + Hadoop + Snowflake + Databricks + Synapse stacks. For financial services specifically: (1) OneLake unified storage eliminates data copy between BI + ML + warehouse workloads; (2) Direct Lake on Power BI removes import-refresh delays critical for intraday risk; (3) Native Microsoft Purview integration provides catalog + lineage + classification at the data-platform layer rather than as a bolt-on; (4) Sentinel integration for audit + compliance reporting; (5) Copilot for Fabric brings natural-language queries to risk + finance analysts. EPC Group has migrated Fortune 500 banks + asset managers from Snowflake + Teradata to Fabric.
What is the right Fabric capacity for a mid-size bank?
For a mid-size bank ($10B-$50B assets) running risk + finance + surveillance workloads, EPC Group typically recommends F128 ($10,515/mo reserved) or F256 ($21,030/mo) Fabric capacity. F128 provides 128 capacity units — sufficient for ~50-200 concurrent power users + automated workloads. F256 is required when intraday risk + real-time surveillance + ML training run concurrently. Capacity is elastic — pause/resume + autoscale available. See /power-bi-cost-and-licensing-guide-for-desktop-pro-and-premium for the full F-SKU pricing matrix.
Can Fabric handle trade surveillance for spoofing + layering + front-running?
Yes. EPC Group ships Fabric-based trade surveillance with Real-Time Analytics + KQL + Spark for: spoofing pattern detection, layering, front-running, wash trading, marking-the-close, insider trading patterns. Integration with NICE Actimize, NASDAQ SMARTS, or custom rule sets. Trade + order + market data + communication data ingested via Eventstream (real-time) + Data Factory (batch). Surveillance officer Power BI dashboards on top with case management workflow integration.
Does Fabric support regulatory reporting (FRTB, OATS/CAT, MiFID II, SFTR, EMIR)?
Yes. EPC Group implements Fabric as the consolidated regulatory reporting platform with: FRTB (Fundamental Review of the Trading Book) — capital reporting under SBM + IMA; FINRA OATS / Consolidated Audit Trail (CAT); MiFID II / MiFIR transaction reporting; SFTR (EU + UK); EMIR (EU + UK); LIBOR transition tracking; ICAAP (European entities); CCAR + DFAST (>$100B asset banks). Fabric Warehouse as the regulatory-grade store of record with audit-quality lineage via Microsoft Purview.
How does Fabric handle MNPI in trading workloads?
MNPI controls in Fabric: (1) Microsoft Purview sensitivity labels applied to datasets + reports + Fabric items; (2) Customer Key + Double Key Encryption for the highest-sensitivity MNPI; (3) Workspace + lakehouse-level access controls aligned to Information Barriers; (4) Audit Premium captures every access event for SEC 17a-4; (5) Restricted Search prevents MNPI workspaces from being indexed by Power BI Copilot. EPC Group designs MNPI controls as part of every FSI Fabric deployment.
Can Fabric host AML transaction monitoring?
Yes. Fabric AML transaction monitoring pattern: customer + transaction + counterparty + sanctions data ingested via Data Factory + Eventstream. KQL + Spark rule sets for suspicious activity detection. ML models for false-positive reduction (typical reduction 40-60% vs rule-only AML). Case management integration with operational AML platforms (Actimize, NICE, Oracle Mantas, SAS AML). Documented as the AML data layer in the firm's BSA / AML program. SAR drafting via Copilot Studio agent.
Why EPC Group for FSI Fabric consulting?
4× Microsoft Press Power BI / data platform author. Federal Reserve Bank of New York Lead Architect pedigree (Errin O'Connor). Microsoft Solutions Partner with Data Platform designation. Migrated Fortune 500 banks + asset managers from Snowflake + Teradata to Fabric. Hundreds of financial services Microsoft engagements. See /industries/financial-services for broader FSI practice.
Schedule Fabric for FSI Discovery
FRBNY pedigree. Fortune 500 bank + asset manager migrations from Snowflake + Teradata to Fabric.