Microsoft Fabric Capacity (CU) Estimator
Model a Microsoft Fabric workload with a few sliders and get an estimated monthly cost, CU utilization, headroom, and a recommended F-SKU. A quick sanity check before you provision — or right-size what you already run.
Runs entirely in your browser — nothing is uploadedEstimated CU utilization
Recommendation
How smoothing & throttling work
Fabric doesn't bill per-second spikes directly — it smooths capacity usage over time (interactive operations over a short window, background operations over up to 24 hours) and lets you burst above your SKU briefly by borrowing future capacity. That's great for occasional peaks, but if sustained demand keeps utilization high, the borrowed capacity adds up and Fabric applies throttling — interactive delays first, then rejection — until usage recovers. Keeping average utilization with real headroom (this tool targets under ~65%) leaves room for bursts without tipping into throttling.
About this Microsoft Fabric capacity estimator
Microsoft Fabric is billed on capacity units (CU): you provision an F-SKU (F2, F4, F8 … up to F2048) and every workload — Data Factory pipelines, Spark notebooks, warehouse queries, Power BI, real-time analytics — draws from that shared pool of CUs. The hard part of adopting Fabric is picking the right SKU: too small and you hit throttling, too big and you burn budget on idle capacity. This free Fabric capacity estimator gives you a fast, back-of-the-envelope answer. Choose an SKU, describe your workload with a few sliders (pipeline runs, interactive load, background load), and it estimates your average CU demand, utilization, headroom, monthly cost, and the smallest SKU that keeps you comfortably under a safe utilization target.
It's aimed at data engineers, architects, and FinOps folks doing Fabric capacity planning and cost estimation — the kind of quick modeling I did when tuning a Fabric warehouse from ~70% to ~30% CU usage and cutting infrastructure cost. Because it runs entirely in your browser, none of your workload assumptions are sent anywhere. Every figure is an approximation: real CU consumption depends heavily on your specific queries, data volumes, and region pricing, so treat the output as a starting point and confirm against Microsoft's metrics app and pricing page before committing.
FAQ
What is a Fabric capacity unit (CU)?
A CU is Fabric's unit of compute. Each F-SKU maps to a fixed number of CUs (F2 = 2 CUs, F64 = 64 CUs, and so on), and every operation across Data Engineering, Data Factory, Power BI, and the other workloads consumes CU-seconds from that pool. Pay-as-you-go capacity is billed per CU-hour.
How do I choose the right Fabric SKU?
Estimate your average CU demand, then pick the smallest SKU that keeps sustained utilization below roughly 65% so you retain headroom for bursts. This tool does that math for you and suggests a recommended SKU; validate it against the Fabric Capacity Metrics app once you're running real workloads.
Are these prices official?
No. The SKU prices here are approximate US pay-as-you-go estimates baked into a single rate constant for illustration. Fabric pricing varies by region and changes over time, and reserved-capacity discounts apply — always check Microsoft's official Fabric pricing page for current numbers.
Is this tool affiliated with Microsoft?
No. It's an independent community tool with no affiliation to or endorsement from Microsoft. It provides rough estimates only and stores nothing — all calculations happen in your browser.
Independent community tool — not affiliated with or endorsed by Microsoft. Estimates only; verify against official Microsoft Fabric pricing and documentation.