AI takes the leap: why investors need to stay on the ball
Software has been the driving force behind digital progress for years, but Artificial Intelligence (AI) is now confidently taking over this role, causing considerable turmoil in the financial markets: what impact will the AI revolution have on traditional software companies? How will it change the nature of knowledge-based work? And, most importantly, what does all this mean for you as an investor?
Joris Franck, Technology Expert and Portfolio Manager at KBC Asset Management, explains what’s going on and what this means for you as an investor.
Every new technology creates winners and losers, but when it comes to AI the losers have yet to be identified. The uncertainties and the pace of AI development spark concern among investors, but also offer interesting opportunities for long-term investors seeking to enter this market.
Joris Franck, Technology Expert and Portfolio Manager at KBC Asset Management
Why software shares are under pressure
Traditional software shares were hit hard in the past few months – not because those companies were necessarily performing poorly, but because no one knows exactly how AI will reshape the sector’s competitive landscape.
The software sector is shrouded in uncertainty. No one knows what the software sector’s earnings model will look like tomorrow, and the rapid progress made with AI in recent months has given rise to a sense of panic.
Joris Franck, Technology Expert and Portfolio Manager at KBC Asset Management
What exactly is going on?
- Artificial General Intelligence (AGI) – systems that equal or surpass human intelligence – threaten to replicate or enhance existing software products in the future at very low cost. This is putting pressure on the ‘value’ of software.
- New AI-native startups are challenging established software companies and are building solutions with full AI integration from the outset, almost fully eliminating the need for human intervention. As a result, those solutions are the cheaper option. They mostly appeal to young tech-driven companies which, therefore, may never become customers of traditional software companies. This is how growth slowly shifts away from established players to new ecosystems.
- AI promises an increase in the productivity of white-collar workers, which may lead companies to cut staff in the future. Many software companies work with per-user licences. Fewer white-collar workers automatically translates into fewer software licences, reducing the turnover of existing software companies.
- Where traditional software companies offer AI solutions, it makes sense that they have customers pay for every AI calculation, search, generated image or action performed by an AI agent. As per-user licences no longer cover the costs incurred, customers pay by use. One major drawback is that this diminishes the predictability of the income generated by software companies, which is a heavy blow for a sector formerly known for its stable recurring income.
- Traditional software companies boast extremely high gross margins, which are generally well above 80%. Once the software has been built, serving additional users or customers entails minimal extra costs. AI, by contrast, requires computing power, and computing power costs money. The creators of AI models (Anthropic, OpenAI, etc.) also charge a fee for the use of their models, putting pressure on the software companies’ high margins.
The spread of AI from software to the entire knowledge economy
New developments in early 2026 were a wake-up call for the market, as AI is evolving faster than anticipated and is now starting to take over a wide range of cognitive tasks.
What has changed?
- Sequoia Capital, one of Silicon Valley’s most influential venture capital companies, believes that the start of 2026 marked the era of functional AGI, in which AI is able to ‘figure things out’ on its own and perform complex tasks with minimum human instruction – not with human-like consciousness, but with digital intelligence. This is ‘functional AI’, which can preserve context and make plans over an extended time horizon, and which can correct itself and work towards a greater purpose.
- Whether this constitutes AGI or not, the progress AI is making is clearly accelerating. New model generations are launched at shorter intervals and each generation demonstrates a noticeable leap in capabilities. The releases of GPT-5.3 Codex and Claude Opus 4.6 represent the most substantial progress in the practical automation of software engineering to date, albeit in different ways.
- Most people underestimate what modern – paid – AI models are already capable of, partly because the capabilities of free versions lag behind. Professionals report a rapid increase in capabilities. Although hallucinations are still possible, their frequency has declined significantly.
Even today, AI is automating a large part of software developers’ work, and that is just the beginning. AI has the potential to automate nearly every white-collar job to a greater or lesser extent and will grow to become a full-fledged digital colleague in office-based professions in areas such as law, consultancy, analysis, marketing and accounting.
Joris Franck, Technology Expert and Portfolio Manager at KBC Asset Management
- Enter Claude Cowork, Anthropic’s agentic desktop tool that can read files, perform tasks and automate workflows directly on your own computer. Cowork has been explicitly designed to automate a wide range of office tasks.
- The consequence is that sectors that rely heavily on cognitive work, such as consultancy, legal services, IT services, professional information providers, financial services, marketing, and so on, feel threatened by AI. The market responded with substantial volatility – not because these companies had suddenly become unprofitable, but because their roles and earnings models need to be redefined in the future.
But what does this mean for you as an investor?
The financial markets reacted strongly, with software companies as well as numerous other companies in sectors such as IT services and professional services/information providers suffering from sharp price drops. These developments give rise to many questions and great uncertainty, but they also present opportunities for anyone willing to look to the longer term.
Opportunities and points to note for investors:
AI infrastructure remains the biggest winner in the AI saga for the time being, as more money flows to AI infrastructure investments every quarter. The AI infrastructure consists of several layers, and in each layer some companies benefit from this growth. The first layer consists of the physical data centres and the energy supplied to these data centres. The second layer is the hardware, specifically the AI chips, memory chips, networking chips, the actual servers, the cooling systems, etc.The third layer is made up of the cloud services provided by the hyperscalers (Google, Microsoft, Amazon, etc.). The fourth layer consists of AI models (Anthropic, OpenAI, etc.). The fifth layer is the data layer, which contains the data needed for AI applications. Many software companies are found in this layer. And the sixth and final layer consists of the actual AI applications, such as sector-specific solutions or (AI-)agentic systems. Every layer has its own winners, its own risks and its own pace of consolidation.
Although AI solutions and AI infrastructure may provide interesting opportunities in the long term, a real risk of a possible AI bubble remains. It is unclear how long current valuations will remain valid and whether and when any correction will occur. Investors should be aware of the possibility that companies engaged in AI may experience substantial fluctuations in value.
For software companies, it is currently debated whether software will become ‘smart’ enough in the future to actively provide input and perform tasks together with AI or that it will merely play a supporting role. In other words, to what extent can software companies also operate in the sixth layer? In the upside scenario, software will become some kind of AI-driven digital colleague who helps us make decisions, plan and act. In the downside scenario, all intelligence is found in the AI layer (i.e. not with the traditional software companies) and software remains a mostly passive archiving and record-keeping system that offers no real added value. In that case, software is trapped in the fifth layer. That would be unfortunate for those companies, as most of the added value will be generated in the top (sixth) layer in the future. The most important thing is to keep a close eye on which companies manage to adapt and which ones don’t.
The uncertainty surrounding software won’t disappear overnight but will gradually fade. At this stage it is still too early to identify winners and losers, although opportunities will undoubtedly arise in the next few months for those who stay on the ball.
Joris Franck, Technology Expert and Portfolio Manager at KBC Asset Management
Want to learn more about thematic investing?
The information contained in this publication is for information purposes only and should not be considered as investment advice.
