AI Scare Trade and SaaSpocalypse

SAP share down 50 percent since the all-time high

The fact that many IT experts are right with this almost prophetic forecast was demonstrated at the end of January when SAP presented its balance sheet for the 2025 financial year - actually with very positive figures, except for the one percentage point shortfall in the annual target for short-term cloud revenues, the so-called current cloud backlog: This was only 25% instead of the expected 26% - and caused a medium tremor in SAP shares on the stock market.

Are SAP's best days behind it? AI is eating software - a general skepticism on stock markets worldwide that is currently dragging down heavyweights such as SAP and Microsoft, but also smaller software providers. Management consultancy BearingPoint sees the current volatility of shares as an echo of a fundamental upheaval in the IT market: those who lived off the number of licenses are heading into a monetization trap.

My first ERP system was SAP R/3 on an IBM RS/6000 server with AIX as the operating system and Oracle as the database. I was a newcomer to the scene and was amazed that IBM was barely making ends meet with software licenses and SAP was getting richer and richer with R/3 licenses! There were exactly two user licenses on the IBM RS/6000 server: an administrator and SAP R/3 itself. Each SAP user naturally needed their own R/3 license, but not an AIX operating system license. It was a question of the IT system architecture from which SAP benefited.

This IT scenario is repeating itself with AI agents: „AI software is eating the world“, experts predict. Why not replace hundreds of S/4 licenses with a single AI assistant that works around the clock and is supported, monitored and controlled by hundreds of former SAP users? A fully-fledged S/4 system could be controlled by five AI assistants (ERP, SCM, HCM, PLM and CRM) and their AI agents (Agentic AI).

We can only guess what exactly is going on

The panic selling of IT and AI shares destroyed billions in market value at times and caused the share price of global ERP market leader SAP to plummet by a dramatic 50 percent in one year, ruthlessly exposing the massive loss of confidence and fear of a disruptive turning point caused by AI pioneers such as Anthropic or OpenAI.

Leading financial analysts warn that in this scenario, SAP could be degraded to a simple, intelligence-less data store - a pure system of records - if the actual value creation and process logic migrates to the AI agents of the competition and the Walldorf-based company's margins implode.

Code Red at SAP

SAP CEO Christian Klein defends himself against these existential doomsday scenarios, arguing that the idea of agents displacing legacy software is simply illogical. His strategic line of defense is based on the well-founded assumption that artificial intelligence can only deliver reliable, hallucination-free results if it is deeply integrated into a company's proprietary business data and highly complex process logic.

As a result, SAP is currently focusing intensively on the concept of Agentic AI and positioning its AI assistant Joule as a central co-pilot that will orchestrate entire fleets of specialized AI agents across various business areas such as finance, purchasing or human resources in the future. Instead of marketing hundreds of isolated use cases, SAP is bundling them into role-based assistants that relieve end users of specific, complex work processes.

Exit strategy: SAP-RPT-1

The Walldorf-based ERP world market leader presented SAP-RPT-1, its own basic model specialized in relational business data and tables, which is intended to guarantee reliable business forecasts beyond the classic, often unpredictable language models.

However, a critical look at the SAP community reveals that SAP's loud AI marketing is miles ahead of operational reality and cannot yet be considered a real success. According to a recent survey by the German-speaking SAP User Group (DSAG), an alarming 77% of companies that are already using AI use cases productively rely on non-SAP solutions from hyperscalers or start-ups, while only a disastrous 3% use SAP's original AI tools.

Even major customers such as Volkswagen are testing the AI co-pilot Joule, but according to insider reports, they complain that the solution is still immature and simply lacks a tangible savings effect in terms of money or resources. SAP's existing customers definitely do not expect their ERP supplier to provide an inflationary, superficial AI sugar coating, but rather practical, secure and, above all, transparent solutions that organize the daily data chaos and create real added business value.

Christian Klein's tough strategic guidelines, according to which essential AI innovations such as Joule or the Green Ledger are reserved exclusively for those customers who can be forced into expensive cloud contracts via Rise with SAP or Grow with SAP, are also causing massive resentment. Loyal on-prem customers feel blackmailed and cut off from the future of ERP, which has permanently damaged trust in the global market leader.

SAP as a driven and hesitant latecomer

In the global AI market, SAP is therefore not acting as a visionary pioneer, but rather as a driven and hesitant latecomer, hastily following the rapid developments of Nvidia, OpenAI, Microsoft and Google.

As SAP does not have the gigantic computing capacities and the double-digit billion budgets for basic research into its own universal large language models (LLMs), the company is taking refuge in a confusing network of partnerships. The Generative AI Hub of the Business Technology Platform (SAP BTP) gives customers access to the hyperscalers' language models, while SAP provides the business-critical ERP data and the business context. The real opportunity and the ultimate vision of SAP therefore lie not in the construction of generic AI models, but in the exclusive refinement of this technology through the unique industry and process knowledge acquired over decades.

If Christian Klein succeeds in building a composable ERP in which the Business Data Cloud (SAP BDC) organizes the fragmented data chaos of customers and error-free AI agents control the complex structural and process organization with confidence, SAP can successfully defend its global market leadership. According to the user association DSAG, SAP BDC runs under the name Business Data Complexity.

However, if SAP fails to translate this gigantic proprietary treasure trove of data into clear, irreplaceable added value, the former software giant will inevitably be relegated to the status of an arbitrarily interchangeable infrastructure supplier in the era of agent-based AI.

SAP-RPT-1, Joule, Abap-1 and Agentic AI

Moltbook takes the principle of Agentic AI to the open Internet. Instead of a handful of AI agents, millions operate there, without a clear framework and without central control. „Moltbook is the first example of an agent ecology that combines scale with the messiness of the real world,“ writes Jack Clark, co-founder of AI company Anthropic, in a blog post. You can „definitely see the future here“, he concludes. (Source: handelsblatt.com) 

The quote „I have seen the future and it works“ is often attributed to the journalist Lincoln Steffens (1866 to 1936) and refers to the early Soviet Union. Steffens visited Soviet Russia from 1919 to 1921 and wrote enthusiastically: „I have seen the future and it works“. With the rise of Stalinism, however, he became increasingly disappointed with the Russian Revolution. „AI - and it works“ is now also a question of survival for SAP boss Christian Klein.