Software Stocks Hit as AI Jolts Pricing Power

The software selloff this week should not be a mystery. It is markets doing what they always do when a structural shift becomes impossible to ignore: repricing, fast and without sentiment.

What investors are reacting to is not fear of AI. It is a reassessment of what software businesses can realistically charge in an AI-first world.

The trigger may have been the release of a new AI automation capability from Anthropic, but the response was already building. The market was primed. The spark simply exposed how fragile many software valuations have become.

For years, software enjoyed a privileged status. Subscription models were treated as inherently resilient. Recurring revenues were assumed to be sticky. Complexity was mistaken for defensibility.

AI dismantles those assumptions.

When intelligent agents can perform legal review, data analysis, research and compliance instantly, the justification for high-priced licences weakens. When outputs become faster, cheaper and widely accessible, pricing power erodes. Markets understand this, even if many boardrooms still resist it.

The selloff reflects investors questioning whether decades-old assumptions around recurring revenues still hold. It is a margin debate, not an innovation debate. Innovation is abundant. Scarcity is not.

Software valuations have long rested on the idea that once a company embedded itself in a workflow, it became indispensable. AI breaks that lock-in. Switching costs fall. Interfaces become optional. Outcomes matter more than platforms.

This is why the speed of the repricing matters. Investors are not waiting for earnings downgrades or guidance cuts. They are moving ahead of them. AI accelerates disruption faster than quarterly results can capture.

There is also a broader point here about bargaining power.

AI shifts leverage away from software vendors and toward users. When a task can be performed by an agent rather than a product suite, buyers gain choice. Choice compresses margins. That dynamic is now being reflected in valuations.

Importantly, this is not a blanket rejection of technology stocks. It is differentiation at work. Markets are drawing a clear line between companies that own AI economics and those that merely integrate AI to defend existing models.

Owning AI economics means controlling the model, the infrastructure, the monetisation layer. Integrating AI often means passing efficiency gains directly to customers. Investors understand the difference, and they are pricing it in.

Another pressure point is revenue durability. Many software firms grew by monetising information asymmetry, process friction or labour intensity. AI strips those away. What remains must justify its price on output alone.

That’s a far harsher test.

The repricing also exposes how incumbency can become a liability. Scale built around legacy platforms, high headcount or complex workflows does not automatically translate into resilience. In some cases, it magnifies exposure.

None of this suggests the end of software. It signals the end of complacency.

Software can still be highly profitable. It can still scale. But its value will increasingly depend on defensibility, not familiarity. On economics, not narratives.

Markets are adjusting to that reality in real time.

AI removes the insulation that once protected software margins. What looked stable now looks exposed. Investors are acting accordingly, because waiting carries more risk than moving early.

This selloff marks a turning point. Valuation theory is catching up with technological reality. And once that adjustment begins, it rarely reverses.