The recent software selloff has been swift and material. Almost $1-trillion in market value has been erased from global software and services stocks in weeks, with the S&P 500 software sub-index down about 27% from its late-2025 highs. In my discussions with investors the question is consistent: has artificial intelligence (AI) fundamentally undermined the software business model, and how would wealth managers be affected?
These are reasonable concerns. Systems developed by firms such as OpenAI and Anthropic can draft documents, generate code and analyse data at levels that seemed improbable only a few years ago. When general-purpose AI providers introduced more industry-customisable “agent” tools, markets reacted sharply, particularly in specialist software and wealth industries. Investors questioned whether vertical, domain-specific tools would remain necessary.
I believe distinction matters and there will be a more nuanced outcome. Across industries, companies are using AI agents to automate repetitive analytical and reporting tasks that previously consumed hours of skilled professional time. Routine forecasting, drafting and modelling work can be completed in minutes. That improves productivity, but it does not eliminate the broader professional function.
Yet, firms operating in regulated sectors continue to argue that domain expertise, structured workflows, audit trails and proprietary data remain critical. General models can accelerate tasks, but complex environments still require integration and oversight.
Furthermore, enterprise software companies are not simply selling isolated features. They operate platforms built on embedded workflows, compliance frameworks, proprietary datasets and distribution networks. AI is increasingly being integrated into those platforms rather than replacing them outright. Microsoft and Salesforce are examples of incumbents embedding AI to strengthen existing ecosystems.
That does not imply immunity. Certain functions will commoditise and pricing power may weaken in some segments. Profit pools could shift toward infrastructure providers or firms controlling scarce data and, yes, some companies will lose market share. The market is correct to reassess expectations.
What appears less justified is the indiscriminate nature of the repricing. The likely outcome is differentiation, not destruction. Over time, it is very likely markets will replace the current “shoot first, ask later” reactions with more thoughtful evaluation of which business models can use AI to enhance their economics.
The same logic applies to wealth management. The initial reaction for listed wealth managers has been similarly abrupt. The assumption is AI-driven investment tools will disintermediate advisers and compress fees. This conclusion overstates the case.
Advisers function as financial concierges. They package products, manage tax complexity, structure estates, ensure regulatory compliance and guide clients behaviourally through uncertainty. AI can improve their service including modelling, reporting and compliance efficiency. This will reduce administrative cost and error rates, but not replace judgment or trust.
Where pressure is more likely is in product packaging and underlying cost layers. Investment fees have been declining for years — the growth of low-cost providers illustrates this trend. Compliance, legal processing and certain insurance functions may also face cost deflation as AI reduces manual work. If savings are passed through, investors benefit.
That does not mean advisory economics collapse. It means differentiation increases. There will be winners and losers among South African wealth managers. Firms that integrate efficient AI tools into their service model will improve margins and client outcomes. Firms that resist change or are hamstrung by bureaucratic hurdles may face pressure. New entrants leveraging lower-cost technology will emerge gradually, but agile incumbents can adapt.
For investors, the implication is disciplined positioning. Abandoning global technology exposure entirely risks stepping away from a sector central to long-term productivity growth. Attempting to identify a single dominant AI winner in a volatile environment is speculative. Technological transitions typically produce multiple evolving beneficiaries rather than one permanent champion.
Software companies and wealth managers will be affected. Some margins will compress and some models weaken. But the broad selloff reflects uncertainty more than inevitability. As clarity improves, markets are likely to distinguish between those companies that can harness AI effectively and those that cannot.
• Dr Marais, a former president and CEO of Wells Fargo Asset Management and adviser to Allspring Global Investments, chairs Wealth Associates.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.