Access all our platforms from a single account.
Software’s AI stress test: mapping threats and opportunities across US and Europe
Key takeaways
AI changes how software charges, shifting from “per user” to “per task” in many products.
Workflow and data owners tend to have the best defence, and often the best upside.
A basket approach helps, because the winners are clearer in hindsight than in headlines.
Software stocks look like they are being asked to justify their existence. That sounds harsh, but it is basically the current mood among investors.
The trigger is not “bad software”. It is a new fear: artificial intelligence (AI) makes people more productive, so companies may need fewer software licences. Think of AI as the new colleague who does not sleep. Great for output. Awkward for seat counts.
The real fear: seats shrink, not demand
Most enterprise software is sold as software as a service (SaaS), meaning a subscription, often priced “per user” (per seat). If AI lets one person do the work of two, the company might cut headcount, and then cut seats. Revenue falls even if the business runs fine.
That is why “AI disruption” can hurt even strong franchises. The market is not rejecting the products. It is questioning the pricing model.
This also explains why the sell-off feels generalised. Investors do not know which vendors will become the ‘AI upgrade’ customers pay extra for, and which become the ‘AI cost’ that eats margins.
The threat map: where AI can “unbundle” the product
The most exposed products are the ones that sell routine knowledge work by the seat.
Human resources and back-office suites can face this narrative risk because they sit close to headcount. Workday runs core staff, payroll, and finance workflows for large firms, often priced per employee. Sage does similar work for small and mid-sized businesses, where cost pressure shows up quickly. If AI helps teams do the same work with fewer people, investors worry seat counts grow more slowly, even if these systems remain painful to replace.
Creative tools face a different risk: “good enough for free”. Adobe powers design, documents, and media work for many professionals. AI makes creation faster, but it also makes basic output easier to copy with cheaper tools. If free assistants become good enough for simple jobs, Adobe must defend pricing with quality, workflow integration, and reliability, not just shiny new features.
The opportunity map: who gets paid when work gets automated
The potential winners tend to sit in three places: platforms, workflows, and data.
Platforms: Microsoft, Oracle, SAP. They sit close to where business data lives, and they control distribution. If AI becomes a default feature, platforms can attach it broadly, and charge for value delivered over time, not just seats.
Workflows: ServiceNow and, in a different way, Salesforce. They are not just databases. They are the systems where work gets routed, approved, logged, and audited. That is where AI agents can act, and where “per task” pricing makes more sense than “per person”.
Data and trusted content: Snowflake, RELX, Wolters Kluwer, Experian, Gartner. AI needs clean data and high-quality sources. These firms already sell decision support, risk checks, and specialist information. AI can make those products faster and more personalised, without making them optional.
Europe has an extra twist: regulation, language, and local market structure. Temenos sells into banks, where change is slow and audit trails matter. Dassault Systèmes and Nemetschek sit in design and engineering, where AI can speed up modelling and reduce mistakes, but the software still has to be precise.
Amadeus IT is a reminder that “software” also includes infrastructure for industries. Travel bookings are messy, multi-party workflows. AI can help, but reliability still pays the bills.
Risks to watch
First, pricing pressure can arrive before new AI revenue shows up. If customers negotiate harder today while AI products ramp slowly, margins can take a hit even with stable demand.
Second, the competitive set can change fast. Some AI tools come from the big model providers, not traditional software firms. If those tools sit on top of existing apps, they can weaken the app’s pricing power.
Third, regulation and data rights matter more. If rules limit how data is used for AI, some product roadmaps slow down. Watch for delayed launches, cautious language, or rising legal costs.
Investor playbook
If a company talks more about “users” than “usage”, track whether it shifts to per-task pricing and whether customers accept it.
If a firm owns a workflow, watch for metrics that show automation adoption, like higher activity, higher attach rates, or better retention.
If a firm sells data or trusted content, look for product upgrades that increase speed and accuracy, not just marketing slogans.
If you build a basket, balance platforms, workflows, and data, because the timing of winners rarely matches the headlines.
The colleague who never sleeps
AI does not kill software. It changes what customers pay for.
In a seat-based world, growth is about adding users. In an AI-enabled world, growth is about outcomes: tasks completed, errors avoided, time saved, and risks reduced. Some vendors will struggle to prove that value, especially if they sell “routine work in a box”. Others will thrive because they sit where work happens, or where data and trust live.
That is why this sell-off is not a simple “good versus bad” story. It is a rewiring story. The new colleague who never sleeps is not taking every chair. But it is making everyone renegotiate the office seating plan.
This material is marketing content and should not be regarded as investment advice. Trading financial instruments carries risks and historic performance is not a guarantee of future results.
The instrument(s) referenced in this content may be issued by a partner, from whom Saxo receives promotional fees, payment or retrocessions. While Saxo may receive compensation from these partnerships, all content is created with the aim of providing clients with valuable information and options.