Case Study: Cost Optimization – Transitioning from Power BI Premium Per User to MS Fabric Introduction

Most restaurant chains start with Power BI Premium Per User (PPU) because it offers a quick and affordable entry point. However, once you scale hundreds of managers or franchisees, the licensing invoice quickly starts to “eat” into your margins.

Migrating to Microsoft Fabric F64 represents a shift from a “per-head” model to dedicated compute capacity. In this article, we examine how this change regains control over the IT budget and why, at the right scale, it is simply the most logical business move.

In the low-margin restaurant industry, optimizing data analytics spending is a key operational strategy. While initial deployment using Power BI Premium Per User (PPU) licenses is justified at a smaller scale, this model generates high, progressive costs as the network grows and the number of report viewers increases.

Financial assumptions and reservation mechanism

With MS Fabric F64, we assume a capacity reservation from the start to optimize costs. This involves committing to a specific compute capacity for a set period—typically one or three years. This Azure financial mechanism provides significant cost reductions in exchange for a long-term commitment.

Assumptions made (rates in 2026):

24 USD/month* per user.
A fixed cost based on head count.

Fixed cost up to 5400 USD/month**.
No per-user cost.
For 2026, Fabric Reserved is currently 5280 USD/month**.
Report creators still require a Pro license (14 USD/month)
Averaging 10% of the workforce.

* Calculations are for illustrative and indicative purposes only. They do not necessarily reflect actual prices and do not constitute an offer. They are based on the Microsoft price list: https://www.microsoft.com/en-us/power-platform/products/power-bi/pricing.
** Calculations are for illustrative and indicative purposes only. They do not necessarily reflect actual prices and do not constitute an offer. They are based on the Azure Calculator https://azure.microsoft.com/en-us/pricing/calculator/ for the Central Sweden region.

Key Features of the Reservation:

• • High Discount:
    Compared to the standard Pay-As-You-Go model, a reservation can reduce Fabric F64 costs by approximately 36–40% (with a one-year commitment).
  • Budget Predictability:
    The price is “frozen” for the entire duration, protecting the organization from list price increases or currency fluctuations.
  • Payment Flexibility:
    Microsoft allows for a full upfront payment or fixed monthly installments. The discount rate remains identical for both options.
  • Technical Impact:
    A reservation is strictly a billing change “under the hood” in the Azure portal; it does not affect performance. If demand grows, you can add capacity via hourly billing or an additional reservation.

Therefore, assuming a booking discount, the price can drop to around 5400 USD per month for the Ms Fabric F64 (*, **).

What do we gain besides savings?

On the one hand, PPU enables a lot, but MS Fabric has additional mechanisms that are not present in PPU at all, which means that money is not the only argument for moving to MS Fabric.

Comparing the computing power of Power BI Premium Per User (PPU) and Microsoft Fabric F64 requires a distinction between shared and dedicated resources.

1. Computing power architecture

• • Power BI PPU:
    The user gets access to Premium features, but the computing power is allocated dynamically from Microsoft’s shared pool. Performance is guaranteed for a single person, but it is not measured in constant units of power
  • Fabric F64:
    Provides 64 Capacity Units (CU). This is dedicated, constant computing power assigned to the organization (equivalent to the former Premium P1 instance).

2. Workload management (Smoothing and Bursting)

In the Fabric F64 model, the system operates with mechanisms that do not occur in PPU:

• • Smoothing:
    Allows for the cushioning of temporary load peaks by spreading some operations over time according to the capacity schedule – Capacity may not run 24/7.
  • Bursting:
    Enables temporary power consumption exceeding 64 CU to speed up task execution, as long as the daily average remains within the norm.
  • PPU:
    In the event of exceeding operational limits, there is an immediate performance limitation (throttling) for a given user.

3. Technical comparison

• • Power units:
    In the case of PPU, resource management is dynamic and hidden from the user. Microsoft Fabric F64 offers 64 CU, which is the equivalent of Power BI Premium P1 performance. However, it should be noted that in Fabric, this pool is shared by all platform services (not just Power BI). In practice, however, when migrating solutions in a 1:1 model, the user receives the same computing power, while gaining access to the full Fabric ecosystem.

  • Max model size:
    In the case of PPU (and the additional option with large semantic model format) it is 100 GB, whereas in MS Fabric F64 it is the memory limit of the given instance.

  • Concurrency:
    In the case of PPU, it is isolation at the user level, and in the case of Microsoft Fabric F64, it is a shared pool for all processes.

  • Direct Lake:
    It does not occur on PPU, whereas in MS Fabric F64 it is something that changes everything we are used to, because it enables lightning-fast access to data.

  • Data engineering:
    In the Power BI Premium Per User model, we rely mainly on Dataflows, which in more complex projects can be a challenge and forces external ETL support (e.g., Azure SQL). However, an experienced team can squeeze the maximum out of this configuration, finding additional pluses in tight cost and architecture control. On the other hand, Microsoft Fabric (from F64 upwards) is a completely different league – we receive a full engineering environment (Notebooks, Lakehouse, Spark), which eliminates technological barriers and allows for full freedom without having to go “outside” the ecosystem.

Cost considerations – when it’s worth it:

Let’s consider a case study covering five stages of a company’s lifecycle:

1. A company just starting out with Power BI – 50 active users per month
2. A company just starting out with Power BI – 50 active users per month plus 5 new users added per month
3. An experienced company with 150 monthly users and 5 new users added per month
4. An experienced company with 220 monthly users and 5 new users added per month
5. An experienced company with 270 monthly users and 3 new users added per month.

Let’s look at the next 12 months:

Comment on the above illustration:
The calculations are for illustrative and indicative purposes only. They do not necessarily reflect actual prices and do not constitute an offer. They are based on the Microsoft price list https://www.microsoft.com/en-us/power-platform/products/power-bi/pricing and the Azure Calculator https://azure.microsoft.com/en-us/pricing/calculator/ for the Central Sweden region.

It’s important to keep these Pro licenses in mind—they account for 5–15% of users and are intended for report creators, though not exclusively. If a company plans to have more than approximately 200 users and that number is growing, a tipping point will occur at around 220 users. Above that number, Microsoft Fabric becomes the more cost-effective option.

Pro licenses are particularly relevant for self-service reporting—any user connecting via Excel to an existing semantic model, in addition to having Build or at least Contributor permissions, must also have a Pro license; this has been factored into the assumption.

Summary and recommendation

Choosing between PPU and Fabric F64 is a strategic decision that must be based on a TCO analysis and an assessment of process maturity. While PPU works well on a smaller scale, complex ecosystems present risks associated with managing multiple processes simultaneously. Issues such as refresh collisions can negatively impact report availability. The capacity-based model (Fabric F64) addresses these challenges by replacing the dynamic and unpredictable limits of PPUs with a single, centralized, and controllable unit of power.

The main conclusion of the analysis is the identification of the cost equilibrium, i.e., the optimal number of users. At current rates (24 USD per PPU vs. ~5000 USD*** per F64 Reserved), the break-even point is approximately 220–230 users.

• • Below 200 users:
    PPU is unbeatable on a per-unit basis.
  • Above 230 users:
    Each additional user in the Fabric model costs the company 0 USD (Free license) or 14 USD (Pro creators), generating exponential savings compared to PPU.

*** Calculations are for illustrative and indicative purposes only. They do not necessarily reflect actual prices and do not constitute an offer. They are based on the Microsoft price list: https://www.microsoft.com/en-us/power-platform/products/power-bi/pricing.

Choosing Microsoft Fabric F64 is about more than just optimizing Power BI. For the same price, your organization gains access to a complete data ecosystem:

• • OneLake:
    A central data lake that eliminates data silos.
  • Data Factory & Synapse:
    Advanced ETL and data engineering tools with no additional licensing costs.
  • Scalability:
    The ability to share reports with an unlimited number of recipients without worrying about “per-head” costs.

The author’s recommendation is as follows:

1. Use PPU until you reach the threshold of 180–200 active users.
2. Prepare the infrastructure for Fabric (Medallion architecture in OneLake) three months in advance.
3. Migrate to F64 (Reserved) once you exceed 230 users to immediately “freeze” analytics costs at a fixed level.
4. If you already have more than 230 users at the time of reading this article, don’t wait. DataRiseLab provides comprehensive services in this area, and our team of experienced and certified specialists will find a comprehensive solution for you.