Copilot in Power BI is a chat-based assistant that answers questions about your reports and data, summarises reports and writes DAX for report creators. As of October 2026 it needs paid Fabric capacity at F2 or higher, or Power BI Premium at P1 or higher, so a Power BI Pro or Premium Per User licence on its own won't switch it on.
That capacity rule is what stops most people. You can have a Pro licence, open a report and find no Copilot at all, because Microsoft ties it to your organisation's capacity rather than to your seat. Here's everything Copilot needs, the three places it appears, which of those are still in preview, and the limits worth knowing, all from Microsoft's own documentation as of October 2026.
Copilot in Power BI requirements at a glance
| Requirement | Detail |
|---|---|
| Capacity | Fabric F2 or higher, or Power BI Premium P1 or higher (no trial capacity, no free SKU) |
| Admin setting | Users can use Copilot and other features powered by Azure OpenAI: on |
| Region | Capacity in a supported region |
| Sovereign clouds | Not supported |
The admin setting is on by default, but your admin may have turned it off, so if everything else is in place and Copilot still isn't there, that's the next thing to check.
Why a Pro licence isn't enough
Microsoft is direct about this: a Power BI Pro or Premium Per User (PPU) licence alone isn't sufficient, because Copilot requires organisational capacity. Trial capacities and free SKUs aren't supported either, so you can't test Copilot on a trial and then decide.
There's a second trap for the standalone experience. Even if Copilot is enabled at capacity level, where that's been delegated, it isn't enough for standalone Copilot. That needs the tenant-level settings, including one called Users can access a standalone, cross-item Power BI Copilot experience (preview).
The region rule outside the US and EU
If your tenant or capacity sits outside the US or EU data boundary, Copilot is disabled by default. Your Fabric tenant admin has to enable a setting called Data sent to Azure OpenAI can be processed outside your tenant's geographic region, compliance boundary, or national cloud instance. That's a data decision as much as a technical one, so it's worth a conversation with whoever owns compliance before anyone flips it.
The standalone and apps experiences also aren't available yet in a list of regions: India West, Indonesia Central, Korea South, Malaysia West, New Zealand North, Qatar Central, Taiwan North and North West, UAE Central, France South, Germany North and Norway West. If your capacity is in one of those, the report pane is what you'll have.
The three places Copilot appears
| Surface | Status | What it does |
|---|---|---|
| Report Copilot pane (on the right of a report) | Generally available | Answers questions about the open report or summarises it; summaries can go into report subscriptions |
| Standalone Copilot (full screen) | Preview | Finds and answers questions about any report or semantic model you can access |
| Copilot in apps | Preview | Scoped to one app's content; supports verified answers prepared by the author |
The difference between the first two is reach. The report pane only knows the report you've got open. Standalone Copilot can go looking across every report and semantic model you're allowed to see, which is far more useful for a business user who doesn't know which report holds the answer. Copilot in apps sits in between, and the verified answers feature lets an author decide in advance how the common questions get answered.
What report and model authors can do with it
For people who build reports, Copilot can create and edit report pages, summarise the semantic model, create a summary visual, write DAX queries and add descriptions to measures.
For model authors working in web modeling, which is in preview, Copilot can spot unclear structures and inconsistent naming, give best-practice recommendations, and refine DAX measures, tables, columns and relationships. If you're newer to the tool, my beginner's guide to Power BI covers the basics this builds on.
How to get Copilot working
- Confirm your workspace runs on paid Fabric F2 or higher, or Power BI Premium P1 or higher.
- Ask your admin to check that Users can use Copilot and other features powered by Azure OpenAI is on at tenant level.
- For standalone Copilot, ask them to switch on the standalone, cross-item Power BI Copilot experience (preview) setting as well.
- If your capacity is outside the US or EU data boundary, agree whether to enable processing outside your region, and have the Fabric tenant admin set it.
- If you've just bought or scaled capacity, allow up to 24 hours for Copilot to recognise the change.
- Prepare your semantic model before you lean on the answers.
That last step matters more than it looks. Microsoft warns that without preparation Copilot can misread the data and return generic or inaccurate results. Clear table and measure names go a long way.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Limits worth knowing
- Each prompt can be up to 10,000 characters, across every Copilot surface: standalone, report pane, apps, Power BI Mobile and Power BI Embedded.
- If you send the exact same prompt against an unchanged semantic model (same schema, data and custom instructions) within a rolling 24 hours, Copilot answers from cache, so you'll see the same response.
- The clear chat button wipes the conversation context but doesn't affect that cache.
The cache point is useful when you're testing. If you're tweaking a model and want a fresh answer, change the model or the prompt, because clearing the chat won't do it.
Is Copilot in Power BI worth it?
If your organisation already runs on Fabric F2 or Premium P1 capacity, I'd switch it on and start with the report pane, which is generally available, while the standalone and apps experiences are in preview. If you're on Pro or PPU only, Copilot isn't a setting you're missing, it's a capacity purchase, and that's a bigger decision than turning on a feature.
If you're still choosing a BI tool, my Tableau vs Power BI comparison is a better place to start than Copilot.
Getting a team to trust and use AI answers in their reporting takes more than a licence change. I run AI training and workshops for teams, and help businesses with AI enablement, AI automation and GEO, which is about getting your business seen and cited in AI answers. Most of that work is with SaaS marketing teams, telecom companies and small businesses. Here's how we can work together.
Frequently asked questions
Can I use Copilot in Power BI with a Pro licence?
Not on its own. As of October 2026 Copilot needs paid Fabric capacity F2 or higher, or Power BI Premium P1 or higher. A Pro or Premium Per User licence alone isn't sufficient.
Does Copilot in Power BI work on a trial?
No. Microsoft says trial capacities and free SKUs aren't supported.
Why can't I see Copilot after buying capacity?
It can take up to 24 hours for Copilot to recognise new or scaled capacity. Also check the tenant setting for Azure OpenAI features and your region.
How long can a Copilot prompt in Power BI be?
Up to 10,000 characters per prompt, in every Copilot surface including the report pane, standalone Copilot, apps, Power BI Mobile and Power BI Embedded.
