TBO

L/S Equity. One man show. Ex-MF PE.

VIC member (seb1991)

[TBO @tbo1789]

$SAP Sharing my SAP VIC write-up (long post).

1/ SAP is undervalued. Intrinsic value is ~€240+ / ~40% gap to today's price. Intrinsic value to compound by ~20% p.a.: ~15% earnings growth + ~5% fcf yield (no multiple expansion)

2/ Mr Market is concerned about AI replacement risk but AI cannot replace ERP systems. LLM models are probabilistic while ERP are deterministic, i.e. requires 100% accuracy. Company cannot make mistakes in financial/tax/legal reportings.

3/ SAP updated its API policy restricting external AI sitting above ERP. Many investors missed this update and misunderstand the implications: SAP anointed itself as orchestration layer king. Any SAP data now needs to flow through specific, sanctioned SAP routes e.g. BDC/A2A Joule [this was not the case before]

4/ SAP found a way converting an open data flow into a monetizable toll road. In other words, it can monetize AI [could not do before]

5/ SAP moat (switching cost, operational embeddedness) is not diluted by AI. Customers cant/wont leave SAP so customer churn is naturally very low

6/ SAP offers its own AI Joule solutions. It has a structural advantage as it sits inside SAP perimeter. Joule is powered by Claude, it does not compete vs other AI models

7/ Cloud migration (key earnings driver) is still ongoing. Still $20-30bn incremental revenue upside. Margin to converge to SaaS levels given cloud larger % of rev

8/ Seat-pricing accounts for less than <35% of cloud rev. Most rev is tiered to business activities (€ GMV, # of docs generated etc,) and consumptions

9/ DCF value model assumes zero AI monetization. So "true" intrinsic value is probably higher than EUR240.

10/ SAP moat is intact. AI is not a terminal value threat as priced by Mr Market

[TBO @tbo1789]

$NVO Fasting concerns for wegovy pill were overblown - oh really??🤣🤣

[TBO @tbo1789]

It is always a red flag when sell-side offers pre and post earnings "catch-up" calls.

[TBO @tbo1789]

@investseekers Danish people work an average of 37 hours per week. I bet $LLY employees work more than that. A lot of $NVO scientists believe they are working at a university rather than at a profit-driven company. So Mike has a point...

[TBO @tbo1789]

Reality $SAP

[In Practise @_inpractise]

Former CTO at Procter &amp; Gamble on who is allowed to run agents inside a $125m $SAP estate

[TBO @tbo1789]

@moats_multiples @F_Compounders hotel comment is interesting; though all of them are not cheap. Accor makes probably most sense.

[TBO @tbo1789]

@moats_multiples @F_Compounders he had a few 360 turns. I would not be surprised if he comes back to $SAP . Sooner or later he might need reducing its $GE exposure and use that proceeds.

[TBO @tbo1789]

Not at all. It does not change the underlying fact that LLM are probabilistic. Today's LLM can now delegate certain executions to a deterministic tool doing a specific workflow. The LLM itself doesn't do the task but merely delegates the work (the LLM is the agent). LLM still stays probabilistic even with this tool delegation.

ERP systems are effectively system of records. You need 100% accuracy data points. So LLM cannot replace ERP system.

The fair question is to what extent can LLM extract values from those ERP systems utilizing $SAP data. However, this route is now blocked/restricted by SAP.
SAP blocks incoming API calls, SAP block data extraction to non-SAP systems.

[TBO @tbo1789]

$SAP Sharing my SAP VIC write-up (long post).

1/ SAP is undervalued. Intrinsic value is ~€240+ / ~40% gap to today's price. Intrinsic value to compound by ~20% p.a.: ~15% earnings growth + ~5% fcf yield (no multiple expansion)

2/ Mr Market is concerned about AI replacement risk but AI cannot replace ERP systems. LLM models are probabilistic while ERP are deterministic, i.e. requires 100% accuracy. Company cannot make mistakes in financial/tax/legal reportings.

3/ SAP updated its API policy restricting external AI sitting above ERP. Many investors missed this update and misunderstand the implications: SAP anointed itself as orchestration layer king. Any SAP data now needs to flow through specific, sanctioned SAP routes e.g. BDC/A2A Joule [this was not the case before]

4/ SAP found a way converting an open data flow into a monetizable toll road. In other words, it can monetize AI [could not do before]

5/ SAP moat (switching cost, operational embeddedness) is not diluted by AI. Customers cant/wont leave SAP so customer churn is naturally very low

6/ SAP offers its own AI Joule solutions. It has a structural advantage as it sits inside SAP perimeter. Joule is powered by Claude, it does not compete vs other AI models

7/ Cloud migration (key earnings driver) is still ongoing. Still $20-30bn incremental revenue upside. Margin to converge to SaaS levels given cloud larger % of rev

8/ Seat-pricing accounts for less than <35% of cloud rev. Most rev is tiered to business activities (€ GMV, # of docs generated etc,) and consumptions

9/ DCF value model assumes zero AI monetization. So "true" intrinsic value is probably higher than EUR240.

10/ SAP moat is intact. AI is not a terminal value threat as priced by Mr Market

[TBO @tbo1789]

@MedtechLS @mithridatistcap @lfg_cap what about aero engine? Just because you buy a GE engine doesn't mean you can MRO it by yourself. You still need to adhere to GE's rule; whether you like or not. Same for $SAP

[TBO @tbo1789]

@lost_analyst very fair point.

[TBO @tbo1789]

@lost_analyst I did not mean to attack you. But I disagree with your premise and from what i heard/read/talked from/with SAP customers the opposite is true. But happy to agree that we disagree.

[TBO @tbo1789]

@lost_analyst you are always changing your arguments after proving your facts😉

[TBO @tbo1789]

@lost_analyst DSAG is one of the largest independent SAP user associations in the world. Should have provided this as context.

[TBO @tbo1789]

@lost_analyst Not true. See DSAG survey asking this Q to SAP customers:

[TBO @tbo1789]

this is a wrong interpretation. SAP is not charging a new fee for accessing your SAP data, it is a fee (+restrictions) for extracting your SAP data; two very different things.

also, SAP is not the only vendor doing it, other SaaS vendors are also implementing such API restrictions.

[TBO @tbo1789]

@mithridatistcap @lfg_cap The definition of a strawman argument is something else; just fyi.

AI models evolves and will continue to do so, and we can come with new/fancy labelling. However this does not change the whole premise that if you want to access data, you need permission to do it.

[TBO @tbo1789]

@mithridatistcap @lfg_cap maybe because it is true, and Mgmt needs to tell investors how the real world actually works instead of living in a bubble and following the sirens singing that “AI will replace humanity”

[TBO @tbo1789]

@Vince85623723 under the premise that token cost increases: i am not concerned about it LT as SAP will pass through costs to end-consumers. Today, it is doing so with hosting fees at a 40-50% mark-up. That is the beauty of such biz when you have pricing power.

[TBO @tbo1789]

1/ Need to differentiate. IT budget wont cut ERP budgets. However other IT spending will surely be re-allocated away from hardware to somewhere else. I believe LT we will see a move back from best-of-breed to best-of-suite.
2/ Yes but this exactly what SAP has been doing and addressed.

[TBO @tbo1789]

Disagree a bit. SaaS apocalypse was caused by people proclaiming that SaaS is dead/AI eating SaaS's lunch. IT budget is not relevant for SAP as ERP is mission critical. You cant reduce spending bc SAP customer contracts prohibits downsizing. Also, i addressed the agentic layer above ERP point. SAP eliminated that risk.

Palantir is interesting. They developed an API data migration tool but this accelerates the migration from ECC to S4 HANA. This was co-engineered by SAP and Palantir from my understanding. Palantir does not offer an ERP solution to be clear

[TBO @tbo1789]

@lfg_cap @lfg_cap

[TBO @tbo1789]

Hemingway Capital – no offense, but you are a bit confused here and there. Happy to help you out. (also a bit odd to quote instead of replying to my post.. anyways)

You say: “Now the "toll road" argument: SAP already gets paid MILLIONS to host and organize its clients” data”.

-> You are confusing a few things here. ERP cloud subscription incl. hosting fee. The toll road is about charging for the value extracted from that SAP data. Two different things.

You say: “Fundamentally, nobody serious thinks LLMs replace the ERP in the next few years”

-> that was exactly what happened during “Saas apocalypse” as people proclaimed “SaaS is dead” incl. ERP.

You say: "SAP blocked external AI above ERP" and "Joule doesn't compete with other AI models" can't both be true.

-> No contradiction at all as two different layers (model layer vs orchestration layer). I would suggest to read through SAP API Policy update (Section 2.2.2, API Policy v4/2026) to learn more.

You say: moat outside of just "storing and organizing data
-> did you read my pitch? The moat: is the embedded business logic, not just data.

You say: “Start with the deterministic vs probabilistic argument - it shows a complete lack of understanding of how models have evolved in the last 18 months and what harnesses & agentic systems do”

-> I believe you are contradicting yourself as your previously states: “nobody serious thinks LLMs replace the ERP in the next few years”; you are effectively agreeing with me without noticing it. LLM cannot replace core ERP systems – this stands. To the 2ndpart: any LLM + harnessing tool -> that requires SAP API access approval as previously described.

You say: DCF modelling is nonsense.
-> I would recommend to read my pitch again: my main DCF driver is about the cloud migration.

You say: “SAP as 2% grower vs 8% grower," >75% of growth from expansion/pricing”
-> no idea where you got this numbers from.

You say: “Many of these bull points are bear points”
-> this is essentially called investing, investors are deriving to different conclusions. The key of investing is to focus on what matters and what not.