$NVDA (+1,67%) and other pure AI companies are currently being punished because many market participants fear that AI could be a bubble. At the same time, classic software vendors such as $SAP (-0,48%) or $IBM (+1,07%) are also coming under pressure - ironically on the grounds that AI can now develop better software. And energy companies are also being eyed with suspicion.
All of this cannot be true at the same time and does not fit together logically.
In my opinion, there is currently a largely unfounded fear of any kind of change in the market. This fear leads to a strange reticence and at the same time feeds the doomsday apologists.
On closer inspection, the thesis that "AI is replacing traditional software" is highly questionable - especially when it comes to business-critical applications:
Imagine every user simply building their own accounting software with AI. Sounds tempting at first. But in practice, an accounting system must function reliably and deterministically function reliably and deterministically. But that's exactly what modern large language models and neural networks can't do.
Neural networks are generally not deterministic (or only under very narrow conditions). More importantly, they cannot be proven mathematically correct - and the problem of formally verifying large neural networks is NP-hard (or in many cases even NP-complete).
In concrete terms, this means that if an AI-based SAP system has to process exactly the same calculation three times in succession, it is not that the result will always be exactly the same. And even less so for another, but actually equivalent, calculation.
The situation is even more critical in security-relevant areas (cybersecurity, medicine, aviation, autonomous systems, etc.). Here it is essential that threats are reliably and reproducibly recognized.
In short: Determinism, formal verifiability, governance and reliable reproducibility are still extremely important values for business software. This is why the hype that "AI will soon replace all traditional software" should be viewed much more calmly and critically.
