Did You Know Microsoft Fabric Is Getting a Zero-Cost Entry Point?

Microsoft Fabric

13 Oct

Until now, starting with Microsoft Fabric meant committing to a capacity SKU upfront. The smallest entry point was F2. It might not be expensive in enterprise terms, but it is a real barrier for teams that wanted to evaluate Fabric properly before committing. And honestly, an F2 did not really give you that much. For organisations with a need for running workloads that are too unpredictable to size couldn’t really get the full benefit of a reserved capacity either.

This changes with F0. (Ok, not really free, but you get the gist.)

Announced at FabCon Europe Barcelona, F0 is a zero-provisioned capacity SKU.
There is no longer an upfront compute charge. Supported workload use on-demand billing by default, meaning you pay for what you use (measured in CU-hours).

Microsoft positions F0 for evaluation, development, and early production. For workloads where capacity spikes are a real demand, it can be used as a buffer.
Everything which most finance teams experience at the start of a Fabric journey.

If you have an F2 or larger capacity, on-demand billing and scaling can be turned on for individual billing categories, moving specific workloads outside the shared capacity CU and limiting them separately. A finance team running steady semantic model refreshes alongside occasional burst-heavy data warehouse queries can keep the steady work on provisioned capacity and let the spikes scale independently.

How do I keep it under control?

No worries, there are built-in spent control and limitations where administrators set a limit in CU-hours over a rolling 24-hour window. If this threshold gets breached, new operations pause, already running operations finish, and capacity resumes when the window resets.

A few things worth knowing before you get excited:

F0 and on-demand billing across Fabric are currently in preview, listed as "coming in the coming weeks" from the Barcelona announcement. It is NOT GA today.
Prices have not been announced. On-demand is priced at a multiplier of the pay-as-you-go rate, but the multiplier is not yet disclosed.
On-demand consumption has no smoothing, unlike provisioned capacity. You are metered as you go. (One example of how real-time analytics can be used)

For finance teams who consider Fabric but are held back by the upfront commitment, F0 is worth watching closely. The ability to actually measure the real usage and find the right capacity to reserve is a real win. It allows you to stop guessing what you need to commit to before you have an actual production-ready workload.