GILD verdict
Choose Copilot in Power BI on Power BI and Microsoft 365, and Gemini in Looker when metrics live in LookML.
Looker's metrics can sit on BigQuery or several warehouses. Copilot works on the models you already have, paid through Fabric capacity you can watch in the Capacity Metrics app. Looker keeps metric logic as reviewed code and can still serve Power BI users through its connector.
The Microsoft side of this comparison is Copilot in Power BI, the Power BI part of Copilot in Fabric. Microsoft licenses it by capacity: it needs paid Fabric capacity (F2 or higher) or Power BI Premium (P1 or higher) and bills that capacity in capacity units, apart from the per-user Microsoft 365 Copilot license. Copilot's AI cost is easier to forecast today. Microsoft prices a request of 2,000 input and 500 output tokens at 400 CU seconds, so an F64 capacity runs over 13,824 such requests a day. Looker's overage meter waits on 90 days' notice from Google, and the one published per-question figure is a pilot admin's 250,000 input tokens.
Choosing between Looker and Microsoft Copilot
Choose Looker if
- Your warehouse is BigQuery, or spans several SQL databases, and you want one LookML model across them queried live.
- Metric logic should be code in Git, reviewed in pull requests and tested in CI before dashboards and agents use it.
- You want AI answers inside Looker and in Gemini Enterprise, and Power BI or Excel users reading the same LookML through connectors.
Choose Microsoft Copilot if
- Your company already runs Power BI on Fabric or Premium capacity, and its people live in Excel, Teams, and Microsoft 365.
- Report authors want Copilot to draft report pages, DAX queries, and measure descriptions inside Power BI Desktop and the service.
- Agents in Microsoft 365 Copilot and Copilot Studio should answer from Power BI semantic models with their row-level security.
Side by side
| Dimension | Looker | Microsoft Copilot |
|---|---|---|
| Semantic model | LookML models and Explores; Gemini selects fields and Looker composes the SQL | Power BI semantic models with measures in DAX; when a question is not about the model's data, Copilot answers from the language model's general knowledge |
| Preparing the model for AI | Field labels, descriptions, synonyms, and hidden fields in LookML, plus agent instructions and verified queries | Prep data for AI: an AI data schema, AI instructions, and verified answers, with models marked Approved for Copilot |
| What it requires | A Looker license with Gemini in Looker enabled per instance and the gemini_in_looker permission on each model; Looker-hosted instances only | Paid Fabric capacity (F2 or higher) or Power BI Premium (P1 or higher); a Pro or Premium Per User license alone does not qualify, and trial capacity is excluded. The per-user Microsoft 365 Copilot license is licensed apart from it, as an add-on or inside Microsoft 365 E7 |
| How AI use is metered | Conversational Analytics data tokens are pooled per instance and expire monthly without rollover. Monthly allowances: 60M input and 1.2M output on Standard; 300M and 6M on Enterprise; 1.2B and 24M on Embed. Overage of US$3.00 per million input and US$20.00 per million output tokens applies only after the promotional period ends, with at least 90 days' notice | Fabric capacity units: 100 CU seconds per 1,000 input tokens, 10 for cached input, and 400 per 1,000 output tokens, smoothed as background jobs over 24 hours |
| PricingList prices in US dollars as published by each vendor. | Annual platform edition through sales (Standard for fewer than 50 users, Enterprise, or Embed). Each includes 10 Standard and 2 Developer users, plus Developer, Standard, and Viewer licenses. Google publishes no list price | Power BI Pro US$14 and Premium Per User US$24 per user per month, paid yearly; Fabric capacity priced separately, by reservation or pay-as-you-go |
| Data handling | No training on prompts or responses; data at rest in the instance's region, model processing on a global service, EU localized processing on Looker 26.10 and later | Azure OpenAI by default, with no training on customer data and no prompt storage for abuse monitoring. Outside the US and EU data boundary, Copilot stays off unless an admin allows cross-geo processing |
| Reach beyond the BI tool | Agents published to Gemini Enterprise (pre-GA), a managed MCP server (pre-GA), and connectors for Sheets, Excel, Power BI, and Tableau | Fabric IQ for Microsoft 365 Copilot, Copilot Studio, and custom agents, plus a Power BI authoring MCP server |
| Status | Conversational Analytics GA since November 2025; dashboard agents, agentic workflows, and the MCP server in pre-GA release | The report Copilot pane is GA; the standalone Power BI agent and app-scoped Copilot are in pre-GA release |
What practitioners report, side by side
Each product's figures come from its own practitioners, for the group you pick.
Overall signal
Looker reads −8 and Microsoft Copilot −3, each from its own practitioners. Looker: 7 positive, 10 negative of 27. Microsoft Copilot: 12 positive, 13 negative of 40.
The score counts the authors from leaders and teams that take a side, one vote each with their latest view, and is how far positive votes lead negative ones, per 100 votes, so Looker's −8 rests on votes split 7 positive, 10 mixed and 9 negative and Microsoft Copilot's −3 on 12, 13 and 13.
A lead of a fifth of the votes reads positive and a lead of half strong positive, with the same lines below zero. The dots show each source, positive, mixed, negative. Run a pilot on questions your analysts already answer to measure how often its numbers match theirs.
Footprint on the AI Knowledge Work stack
They differ on six of 7 layers. Looker owns one that Microsoft Copilot does not, and Microsoft Copilot owns two that Looker does not. Coverage is each product's placement on this stack, as its product page shows it. Core is a job the product owns; Supports and Touches are jobs it helps with. Coverage says nothing about how well a product does them.
| Layer | Looker | Microsoft Copilot |
|---|---|---|
| Assist | ||
| AI assistant | Touches | Core |
| Find | ||
| Web research | Touches | Supports |
| Enterprise search | Touches | Core |
| Create | ||
| Documents | Touches | Supports |
| Meeting notes | Touches | Supports |
| Analyze | ||
| Data analysis | Core | Supports |
The 1 layer where they match
| Layer | Looker | Microsoft Copilot |
|---|---|---|
| Automate | ||
| Automation | Supports | Supports |
What practitioners praise
Among leaders and teams, Looker alone has three praise themes that reach 5 sources from 3 authors. Governed Metrics and Self-Serve Exploration lead, at 9 of 11 and 5 of 8. Microsoft Copilot alone has three praise themes that reach 5 sources from 3 authors. Meeting Recaps and Enterprise Data Terms lead, at 5 of 5 and 5 of 7.
| Praised for | Looker | Microsoft Copilot |
|---|---|---|
| Governed Metrics | 9 of 11 | Not a theme in its sources |
| Self-Serve Exploration | 5 of 8 | Not a theme in its sources |
| Gemini & AI Agents | 2 of 7 | Not a theme in its sources |
| Meeting Recaps | Not a theme in its sources | 5 of 5 |
| Enterprise Data Terms | Not a theme in its sources | 5 of 7 |
| Agents & Cowork | Not a theme in its sources | 4 of 10 |
Each count is the sources that praise a theme, or raise it as a concern, out of those that discuss it. A row pairs the two products when both carry a theme under the same or an equivalent name.
A count ending in "so far" has under 5 sources from 3 authors among leaders and teams. "Not a theme in its sources" means GILD found no theme of that name in that product's sources.
What practitioners raise as concerns
Among leaders and teams, Microsoft Copilot alone has four concern themes that reach 5 sources from 3 authors. Admin Controls and Oversharing lead, at 6 of 6 and 5 of 5. LookML Learning Curve and Cost reach 5 sources from 3 authors only for Looker, at 10 of 11 and 8 of 9. Microsoft Copilot needs a pilot on the same questions before you choose on either.
| Raised as a concern | Looker | Microsoft Copilot |
|---|---|---|
| LookML Learning Curve | 10 of 11 | Not a theme in its sources |
| Cost | 8 of 9 | Not a theme in its sources |
| Admin Controls | Not a theme in its sources | 6 of 6 |
| Oversharing | Not a theme in its sources | 5 of 5 |
| License Cost & Confusion | Not a theme in its sources | 5 of 6 |
| Low Adoption | Not a theme in its sources | 4 of 5 |
17 sources
- Google Developer forums: Conversational analytics / gemini token usage data (Sep 14, 2026)
- Looker pricing
- Looker docs: Gemini in Looker overview
- Looker docs: Conversational Analytics in Looker overview
- Looker docs: Best practices for configuring Conversational Analytics
- Looker docs: Looker, Power BI Connector
- Looker docs: Admin settings, BI Connectors
- Google Cloud docs: How Gemini products in Google Cloud use your data
- Microsoft Learn: Copilot for Power BI overview
- Microsoft Learn: License options for Microsoft Copilot
- Microsoft Learn: Privacy, security, and responsible AI use of Copilot in Fabric
- Microsoft Learn: Copilot in Fabric consumption
- Microsoft Learn: Prepare your data for AI to improve Copilot results
- Microsoft Learn: Power BI service features by license type
- Microsoft Learn: Overview of the Power BI MCP servers
- Microsoft Learn: What is Fabric IQ?
- Microsoft: Power BI pricing