Welcome to edition 03 of /n Procurement.

We just passed 1,000 subscribers before this send. A big thank you for reading 🎉

Last month I discussed how your SAP Business Network account has to be migrated to SAP’s Business Technology Platform (BTP) before you can do anything in Next-Gen Ariba.

This month, I want to show you how SAP is currently pricing AI functionality. Understanding how this works is how you answer the question:

“Is it worth building this AI agent in the SAP ecosystem? Is there going to be a return on investment?”

Every procurement team running SAP right now…

Three important takeaways:

  • You can start using LLM-supported AI agents with any SAP product today: Legacy Ariba, Next-Gen Ariba, Fieldglass, Concur, SAP ECC, S/4HANA, etc. No need to wait for Next-Gen Ariba for this.

  • Regardless of the SAP product, AI pricing works the same way (very happy about this… Would have been much more difficult to manage otherwise).

  • Some AI functionalities are free and included in SAP’s “Base AI” tier. Often, it’s just a question of turning them on in your instance/install (in collaboration with your internal solution administrator/owner).

As for the premium features? The devil is in the details… And I’m covering those details today.

Let's get into it.

📢 What you'll find in this edition:

  • 🔗 Worth Bookmarking

  • 🔁 SAP AI Pricing Explained: AI Units, Base vs. Premium

  • 🔵 Best LinkedIn Post of the Month

  • 💬 Worth Repeating…

  • 🌯 Wrap Up

Note: Some of the content listed above is only available in the email version of this newsletter. Don’t miss out! Sign up for free to get the next edition.

The Monthly Update

SAP Business AI Pricing Explained: AI Units, Base vs. Premium

SAP now prices Business AI in a virtual currency called AI Units. Here's how the model works.

For most organizations that have seen the LLM demos and run the proofs of concept, it's not the vision that's missing. The power of LLM-supported automation (or AI agents) is easy to see. However, what's not so easy to see and understand is the business case. LLMs are fundamentally usage-based creatures. When you use them at the organization level instead of the user/seat level, there are new costs associated with each use (or API call).

The “fixed-cost-per-seat” model is slowly going extinct… And not just in the SAP world but across the board with LLMs.

That means that understanding LLM/AI cost models and making sense of them in the context of your agent-by-agent business cases is a core skill going forward... I want to give you the foundation you need to do this in the SAP ecosystem. That's the topic today.

And since SAP pricing models are known to be complex in general (AI or not!), it's worth taking the time to understand! Let’s get to it.

Objectives of This Article

  • Explain SAP's commercial model for Business AI and the incoming updates.

  • Provide clarity on Base and Premium AI offerings, including pricing and packaging.

  • Share the tools available to support commercial planning and decision-making.

  • Learn how to transition to the new commercial agreement (SKU) from Q3 2026.

  • Recommended next steps for adoption and onboarding.

SAP Business AI: How to Approach the Journey from Capability to Value

SAP provides pre-built Base and Premium AI capabilities that customers can adopt out of the box. At the same time, it offers the tools and platform services needed to build custom AI solutions, or to extend SAP's standard AI use cases to meet specific business requirements.

The 3 types of SAP Business AI Functionalities

1) Base AI

Base AI is available at no additional cost and comes in two forms: either as AI capabilities already embedded within your SAP cloud applications, or through features included as part of the Joule Base license.

📌 Joule Base: Joule Base is SAP's co-pilot-style foundational AI experience, embedded across SAP cloud applications and accessible through natural language interactions.

Users can ask questions, navigate directly to the appropriate screens, retrieve information, receive contextual guidance, and complete everyday tasks, all within their existing workflow.

Customers need to activate the zero-dollar license to experience Joule Base features within the relevant cloud products.

Example: To get a sense of what Joule Base can deliver, SAP currently provides 61 Joule use cases across the SAP Ariba portfolio.

📌 Embedded AI: Embedded AI refers to AI capabilities built directly into specific SAP applications.

These features provide contextual insights, recommendations, and intelligent automation within existing business processes, allowing users to benefit from AI without needing a separate interaction.

Availability varies by SAP solution and product area.

Example: A few SAP Ariba embedded AI examples include item recommendations in Guided Buying, supplier and content recommendations, and event summaries in SAP Ariba Sourcing. These capabilities are integrated directly into the product user experience.

Base AI One-pager

2) Premium AI

Premium AI is where SAP's AI capabilities move beyond assisting users to executing business processes. It is centered around Joule Agents, which can automate workflows, orchestrate tasks, and perform complex activities on behalf of the organization.

Premium AI also includes high-value AI features and SAP Joule for Consultants (the only offering that continues to be licensed on a per-user basis).

To make sense of SAP's Premium AI offerings, it helps to start with the consumption-metric terminology.

AI Unit and Premium AI consumption-metric terminology

AI Units are SAP's consumption-based credit system: a virtual currency used to access and run Premium AI capabilities across SAP solutions.

Customers purchase AI Units once and use them flexibly across multiple SAP AI capabilities, drawing from a central balance available across your SAP ecosystem. AI Units remain valid for 12 months from purchase.

AI Units are consumed based on defined usage metrics, such as actions executed, requests processed, documents processed, or records extracted.

There’s a lot of novel terms that act as the raw material of SAP’s AI pricing model, so let’s define them…

📌 Action: An "Action" is a distinct operation performed by an AI agent that is invoked and processed as part of completing a task within an AI agent run.

An AI agent run may be triggered by a human query or a system event, and it ends when a task is completed, tools or services are invoked, a workflow is completed, a response is received, or the interaction is terminated.

An AI agent run may consist of one or more Actions.

📌 Request: A "Request" is a response or action that uses AI to retrieve information or perform tasks. The number of requests consumed is a measure of the time and effort per use of the Active AI Service.

📌 User: A "User" is an individual authorized to use the Active AI Service.

📌 Record: A "Record" is an individual data object of a supported file type, ingested by the Active AI Service or available for retrieval.

AI Units are consumed in 2 ways

Consumption-Based Usage of AI Units

  • Joule Agents: Agents are the core of Premium AI. Each time an agent performs a business process on behalf of the business, it consumes AI Units based on the actions executed. SAP-delivered enterprise agents use a standard rate of 0.02 AI Units per action.

  • High-value AI features: Some Premium AI capabilities are priced based on consumption, typically per request or per record processed.

    For example, Document Grounding (indexing) costs 0.005 AI Units per record to prepare content for retrieval. Once indexed, users can retrieve answers from that content through Joule Base at no additional AI Unit cost.

Seat-Based Usage of AI Units

There is still a possibility of getting user-based pricing through Joule for Consultants

  • Joule for Consultants: Meant for SAP IT experts or key users, SAP Joule for Consultants is SAP's per-user, per-month premium offering, designed for professional-services scenarios where AI value is tied directly to individual productivity.

    Note: While named Joule for Consultants, this offering is not exclusively for consultants… It’s for power users. Fear not.

    SAP charges each assigned user 35 AI Units per month, which includes a pooled fair-use request allowance of 22,900 requests per user per month. You can add or remove SAP Joule for Consultants users monthly as your needs change.

To illustrate, here's an example calculation for a large enterprise. Imagine a global organization has purchased 7,000 AI Units for the year.

This could be broken down according to the following randomized usage snapshot over a single month:

When you imagine managing this at scale with many licenses active at the same time, you start seeing the implications…

That’s why the SAP for Me portal lets customer IT administrators activate, manage, and monitor SAP Business AI consumption from a single place. Administrators can track AI Unit usage and proactively disable agents or features to control consumption.

If usage exceeds the purchased AI Unit allocation, the additional consumption is billed transparently based on SAP's applicable excess-use rates.

3) Custom-Built AI

For organizations that want to go beyond packaged AI and build solutions tailored to their specific needs, SAP provides a complete platform for developing and running custom agents, applications, and workflows.

Customers can build custom AI solutions for free during the design phase and pay only for production runtime. This is why some of the tooling below is provided free of charge (for development/testing) and some is paid (for running in production, where the real value comes from).

Here are the core technology components that enable custom-built AI in the SAP context:

📌 Joule Work: Joule Work is the access layer for custom-built SAP AI. It is the unified workspace that brings together data, applications, and AI-driven workflows in a single experience. It sits on SAP’s Business Technology Platform (BTP) and is also free of cost.

Rather than switching between systems or manually coordinating tasks, users simply describe their objective in Joule Work. Joule Assistants then provide relevant insights, automate routine activities, and orchestrate AI agents to perform the work.

Joule Work also enables free access to the SAP AI Agent Hub, development spaces, shared experiences, and BTP-hosted extensions during the custom AI build phase.

📌 Design-time tooling: A complete authoring environment to build, test, and iterate on agents, applications, and workflows, with AI-assisted development available within fair-use guidelines. Free of cost.

And, when you want to deploy to production…

📌 Joule Studio runtime: Productive runtime for custom-built agents, apps, workflows, and extensions, billed based on resource consumption across compute, storage, transactions, and AI tokens.

It is a paid offering, and fees are calculated according to the customer's SAP BTP Enterprise Agreement (BTPEA), Cloud Platform Enterprise Agreement (CPEA), or Pay-As-You-Go (PAYG) contract, depending on the commercial model in place.

Changes Expected Starting Q3 2026

Now that you’ve got a good base on how the model works, here’s what SAP is changing in the months ahead:

Today, most generative AI capabilities are part of Premium AI and consume AI Units.

Starting in Q3 2026, many of these capabilities will move to Base AI, significantly increasing the AI functionality included with standard SAP cloud subscriptions (but you still need to enable them in your own solution with administrators if you want to use them!)

Existing AI capabilities that customers have already activated will continue to operate without interruption, ensuring a seamless transition to the new model.

So, whether you are an existing customer already using SAP's AI features or completely new to SAP's AI portfolio, there’s a new reality coming this fall. The table below summarizes what the evolved commercial model means for you and whether any action is needed.

So What?

Now that you have a good understanding of SAP Business AI’s pricing model, what should you do next as a procurement team running on SAP products?

As you can see, there's a learning curve to climb before you can master LLM-supported AI agent usage at scale in the SAP ecosystem.

My advice is to start earlier than you think you're ready.

You should budget AI Units to fund R&D and analysis exercises, knowing they will probably not generate great bottom-line ROI just yet. The ROI you're looking for in the beginning is AI pricing literacy in YOUR SAP ecosystem context...

IT teams running SAP will be looking for partners in the business to experiment and learn with. It's the perfect time to raise your hand to experiment with them if you haven't already done so.

That way, you'll be in the game when there are "learning dollars" on the table and be ahead of the curve when comes the time to justify AI Agent investments.

How Should Customers Approach Their SAP Business AI Journey?

  • Start realizing value immediately with Base AI (ensure desired functionalities are turned on! You’re already paying for them)

  • Partner with IT and start experimenting with confidence having understood the underlying consumption and commercial model from pilot to enterprise-wide deployment.

  • Accelerate time to value with SAP-delivered AI capabilities available out of the box.

  • Extend and differentiate by building AI solutions tailored to your business needs, knowing they will probably be a bit pricier.

  • Develop the governance models to pay for usage, not access. It can be powerful to align costs with actual AI consumption and business outcomes. However, this requires you have the responsibilities, processes and tools in place to monitor value in real-time, at scale. This is the challenge of the age of AI.

Disclosure: This article represents the understanding and views of the author at the time of writing. /n Procurement is an independent publication and is not affiliated with, endorsed by, or sponsored by SAP. Readers should independently validate all information provided before making any major business decisions. Use at your own risk.

Worth Repeating…

Not everything that counts can be counted, and not everything that can be counted counts.

William Bruce Cameron

Wrap Up

  1. A Skeptic’s Take on AI in Procurement.
    Last year, Joël Collin-Demers took the stage at Zip Forward: LLM market overview, vendor motivations, and a framework to chart your own path… Without the hype.

  2. Want to cut through the AI noise?

    Most vendors say they “use AI.” Very few tell you which type, where it fits, or what it should actually do in procurement. This poster helps you sort signal from spin by breaking down the AI subdomains that matter, where they apply, and where they don’t.

The /n Procurement Team

P.S. Please rate today's newsletter.

Your feedback shapes the next edition.

Login or Subscribe to participate

Reply

Avatar

or to participate