Unfiltered: SAP Public Cloud · Newsletter

Spend the euro once: AI credits or the Public Cloud move?

Edition 7 - The Good, The Bad & The Ugly of buying SAP Public Cloud: why the sequence of capital matters more than the attractiveness of AI or the Public C

The Good, The Bad & The Ugly of spending the SAP euro once. Episode 7

Why the smartest investment is not always the newest technology, but the decision that makes the next investment worthwhile.

As a German, I always wondered about the English expression, “You can’t have your cake and eat it too.”

But, what else am I supposed to do with a cake?

But after enough conversations about IT budgets, I think I finally understand.

Everyone wants AI. Everyone wants a modern ERP. And preferably, everyone wants both without paying for either twice.

Unfortunately, the CFO only has one euro to spend.

So where does it go? Into AI credits or the move to SAP Public Cloud?

And here is where the cake analogy becomes surprisingly useful. The problem is not having the cake. It is deciding what you can afford to consume now, what you need to preserve for later, and whether the second serving will be worth anything.

In technology investment, that is called sequencing.

The euro is not the problem. The sequence is.

Let’s get one misconception out of the way.

AI does not require Public Cloud.

A well-defined AI use case can deliver value on the ERP estate you already operate, including ECC. If the data is trustworthy, the business owner is accountable and the return is measurable, there is no compelling reason to wait for a multi-year transformation.

You can build governed business context around existing systems through integration, semantic models, APIs and controlled workflows.

But there is a difference between making one AI use case work and making AI work across an enterprise.

The first might need a good dataset, a capable team and a clear problem.

The second needs something harder to buy: consistent business meaning, governed decisions and processes capable of executing what AI recommends.

We explored that distinction in AI agents do not need more data, they need business context.

Fragmented processes and inconsistent data might not stop the first use case. They can make the tenth prohibitively expensive.

You end up building the same integrations, resolving the same data inconsistencies and debating the same decision rights all over again.

And that is when the apparently inexpensive AI pilot becomes an expensive enterprise programme.

Before approving another blanket purchase of AI credits, I would ask one question:

If we removed the words “artificial intelligence” from this proposal, would the business case still justify the investment?

If the answer is no, perhaps we are buying enthusiasm rather than an outcome.

“Just go Public” is not a purchase either.

Now for the other side of the budget.

Moving to SAP Public Cloud sounds wonderfully straightforward in a board presentation.

Standard processes. Lower technical complexity. Continuous innovation. Less custom code to maintain.

What’s not to like?

Well, ask the organisation that has spent twenty years modifying ECC to accommodate every local requirement, acquired business, exception and special approval process.

There is no magic button that turns that estate into a standardised cloud operation.

SAP S/4HANA Cloud Public Edition is designed around fit-to-standard processes, a clean core and controlled extensibility. SAP S/4HANA Cloud Private Edition allows more flexibility to accommodate existing complexity.

GROW with SAP and RISE with SAP are commercial transformation offerings, not technical deployment architectures. SAP’s positioning has also evolved, with RISE supporting transformation journeys to both public and private cloud ERP.

The practical architectural choice remains how much standardisation and flexibility the business needs.

For a deeper comparison, see RISE vs GROW and SAP Public Cloud vs Private Cloud.

Twenty years of ECC customisation do not automatically disqualify an organisation from Public Edition.

But they do create a bill that someone must eventually confront.

Which modifications still provide competitive advantage?

Which exist because nobody challenged the requirement when it was introduced?

Which can be replaced by standard capabilities?

And which genuinely differentiated requirements should be redesigned using clean-core extensions?

Those are business decisions disguised as technical questions.

You do not make them by selecting a cloud subscription.

You make them by deciding how the organisation should operate.

That is the part of SAP Clean Core that matters more than any technical checklist.

The real question is not whether your ECC estate can move to Public Cloud. It is how much of your historical process differentiation is still worth paying to preserve.

The one-euro investment test

So perhaps the CFO’s question needs rewriting.

Instead of asking whether the next euro goes to AI or ERP, ask:

What is stopping the next euro from creating value?

That leads to three legitimate investment decisions.

  1. AI first: when the value is already within reach

You have a contained use case, reliable data, clear decision rights and an accountable business owner.

The expected value is measurable.

Perhaps AI can reduce manual invoice handling, improve exception management or accelerate a specific decision.

There is no reason to hold that investment hostage to an ERP migration.

Fund it. Measure the result. Learn from it.

But do not confuse a successful pilot with permission to purchase an enterprise-wide credit balance.

The question after the first win is whether the second and third use cases can reuse what you have built.

  1. Core first: when yesterday’s complexity is blocking tomorrow’s value

Your processes differ from country to country. Custom code protects exceptions nobody remembers authorising. Data definitions change between systems.

Every new AI use case requires another round of reconciliation.

Here, additional AI credits may simply finance more workarounds.

The better investment may be process standardisation, data governance or an ERP transformation.

Not because modern ERP is automatically more valuable than AI, but because the underlying constraints are preventing AI from delivering value at scale.

And if Public Cloud is the destination, the biggest investment decision may not be the subscription.

It may be the business change required to operate within the standard.

  1. Both, selectively: when the roadmap supports it

Not every transformation needs to be an all-or-nothing programme.

A staged ERP modernisation can proceed alongside targeted AI investments.

Some AI use cases might deliver immediate value on the existing estate. Others should wait until the target processes and data foundations are available.

The condition is simple: one business roadmap, transparent dependencies and separate investment gates.

ERP must demonstrate its own business value.

AI must demonstrate its own business value.

Neither should be allowed to hide behind the other’s promises.

Three paths. One principle: invest first where removing a constraint unlocks the greatest subsequent value.

The hidden cost of getting the order wrong

There is another reason this matters.

Technology budgets rarely account properly for the cost of sequencing mistakes.

Buy AI capability too early and you may spend the next two years rebuilding integrations, fixing data quality and negotiating process ownership.

Modernise ERP without challenging existing complexity and you may recreate yesterday’s operating model on a more expensive platform.

Choose Public Cloud without accepting the organisational changes it requires and the project risks becoming an extended argument about why the standard should make an exception for you.

These are not necessarily failures of the technology.

They are failures to make the business decisions that the technology depends on.

And they are expensive because the organisation pays twice: first for the investment, then for the constraints it failed to address.

My take: spend on the constraint, not the demonstration

If I had only one euro to allocate, I would not automatically spend it on AI.

But I would not automatically spend it on Public Cloud either.

I would look for the constraint that prevents the next euro from producing a return.

Sometimes it is data quality. Sometimes it is technical debt. Sometimes it is a decision-rights problem that no new software will solve.

And sometimes it is an uncomfortable conversation about twenty years of ECC customisation.

We should also be clear about the cloud decision:

You cannot customise your way into Public Cloud. You have to standardise your way there.

That does not mean abandoning genuine differentiation. It means preserving it through supported extensibility while challenging everything that exists simply because somebody once said,

“This is how we do it.”

The next ERP investment should make future innovation easier, not preserve the cost of past decisions indefinitely.

And the next AI investment should buy a business outcome, not simply permission to experiment.

Which brings me back to the cake.

I still think the English got that expression backwards.

But perhaps the CFO version makes sense after all.

You can have the cake. You can eat it.

Just make sure the first bite does not consume the money you needed to make the rest worthwhile.

So where is the real constraint in your organisation today?

And is your next euro going towards removing it, or towards the technology that looks best in the next board presentation?

Further perspectives

Two sapperment contributors approach this investment dilemma from different directions.

In The AI Investment Test, Nina challenges the business case behind AI spending. Her question is not whether AI can work, but whether the investment deserves the capital.

In The Capacity Dividend, she takes the argument further. AI may free capacity, but that capacity creates no business value until leadership decides what to do with it.

Explore more of Nina’s thinking in The Growth Mandate.

Luk Smeyers | The economics of delivering the transformation

In The SAP partner ecosystem has a revenue quality problem, Luk examines a different investment constraint: the commercial model behind enterprise transformation.

When delivery partners earn primarily from days billed rather than outcomes achieved, the incentives deserve scrutiny.

Before committing the euro to a core transformation, understanding how your delivery partner creates value is part of the investment decision.

More perspectives from Luk Smeyers.

sapperment is independent and not affiliated with SAP SE. This edition provides a commercial perspective, not legal advice.

Further reading:

AI agents do not need more data, they need business context → https://www.sapperment.com/sap-executive-clarity/ai-agents-business-context/

RISE vs GROW → https://www.sapperment.com/sap-executive-clarity/rise-vs-grow/

SAP Public Cloud vs Private Cloud → https://www.sapperment.com/sap-executive-clarity/public-vs-private-cloud/

SAP Clean Core → https://www.sapperment.com/sap-executive-clarity/clean-core/

The AI Investment Test → https://www.sapperment.com/insights/ai-investment-test/

The Capacity Dividend → https://www.sapperment.com/insights/capacity-dividend/

The Growth Mandate (Nina’s author page) → https://www.sapperment.com/authors/nina-simosko/

The SAP partner ecosystem has a revenue quality problem → https://www.sapperment.com/insights/sap-partner-revenue-quality/

More from Luk (guest page) → https://www.sapperment.com/guests/luk-smeyers/

Sources checked on 04 November 2026


First published on LinkedIn. Subscribe to Unfiltered: SAP Public Cloud to get new editions as they appear — or browse the sapperment library.