AI fears wipe nearly US$2 trillion off global software stocks

Throughout market history, dominant industries have rarely been displaced overnight. Instead, they have been gradually, then suddenly, reshaped by technologies that fundamentally alter how value is created, delivered and captured.

The pattern may be familiar to investors: early signs of disruption are often dismissed as incremental innovation before deeper shifts in behaviour, cost structures and distribution models force a reassessment of entire sectors.

We have seen this before. Traditional media lost structural ground as audiences and advertising migrated online. Video rental collapsed as streaming transformed distribution economics. Film photography was overtaken by digital capture, while brick-and-mortar retail has spent more than a decade adjusting to the rise of ecommerce and platform-based consumption.

In each case, the disruption was not merely technological but economic. Revenues slowed, margins compressed and markets eventually revalued what these businesses were worth in a structurally different world.

Today, investors are increasingly asking whether software is entering a similar period of reassessment, with artificial intelligence emerging as the central catalyst.

What began as artificial intelligence-enhancing productivity is now evolving into something more consequential. Large language models and emerging “agentic AI” systems are no longer limited to assisting users with search or content generation. Increasingly, they can interpret instructions, execute multi-step tasks and operate across workflows that historically required structured software environments and specialist users.

This matters because it changes the nature of interaction with software itself.

For decades, enterprise software has relied on structured interfaces, predefined workflows and trained users navigating increasingly complex systems. These systems remain deeply embedded within corporate infrastructure and continue to play a critical role in business operations. However, AI is beginning to abstract much of that complexity. Instead of navigating multiple applications, users can increasingly describe an outcome in natural language, while AI coordinates execution across systems in the background.

For investors, this raises a more fundamental question: if AI can replicate or automate meaningful portions of traditional software functionality, what happens to the economics that have defined the sector for decades?

One of the clearest concerns relates to pricing models. Enterprise software has historically benefited from “seat-based” pricing, where revenue scales with the number of users. But if AI materially improves productivity per employee, or reduces the need for human interaction altogether, that relationship may weaken over time.

At the same time, lower barriers to software creation may accelerate competition, shorten product cycles and compress differentiation across parts of the industry. The implication is not simply that software becomes cheaper to build, but that long-standing assumptions around durability, pricing power and market dominance may need to be revisited.

This evolution has driven a sharp reassessment across global software equities over the past several months.

Software companies have materially underperformed broader equity indices year to date as investors attempt to reprice long-duration growth assets in a world where the future economics of software feel less certain than they once did.

Even some of the world’s largest and most profitable software businesses have not escaped the sell-off. Collectively, leading global software companies had, at the peak of the drawdown, lost close to US$2 trillion in market capitalisation since late 2025, reflecting the extent to which the market is discounting potential structural disruption.

What makes this particularly notable is that the de-rating has occurred despite relatively stable near-term operating performance across much of the sector. Revenues, margins and cash generation in many cases remain robust. The debate, therefore, is not primarily about what these companies earn today, but what investors believe they can sustainably earn into perpetuity, in essence, a terminal value debate.

This distinction matters because most equity value is ultimately derived from cash flows far into the future rather than the next, say, three years. Even modest changes in assumptions around future growth, competitive advantage or pricing power can materially affect valuations.

Yet, while the disruption narrative is compelling, there are also reasons to believe the market may be extrapolating the pace and breadth of change too aggressively.

Enterprise software is deeply embedded within organisational workflows. Replacing core systems is often expensive, operationally risky and time-consuming. The process of replacing them could be compared to “open-heart surgery” for large organisations. Furthermore, while AI outputs are probabilistic in nature, many business operations require deterministic answers and outputs – this is where traditional software excels.

A more plausible near-term outcome may therefore be integration rather than outright replacement. Companies that can successfully embed AI into their offerings may ultimately strengthen their competitive positions instead of losing them. AI agents may well operate on top of pre-existing software packages too – in some cases, enhancing the use and value of those platforms.

In that sense therefore, the current environment may prove less about the collapse of software economics and more about a separation between durable platforms and vulnerable ones.

The investment implications extend beyond strategy and technology. Falling share prices have also shifted the dynamics of capital allocation; lower valuations increase the relative impact of stock-based compensation while simultaneously making buybacks more accretive for companies with strong free cash flow generation.

None of this suggests the risks posed by AI are overstated or irrelevant. Technological disruption has repeatedly demonstrated the market’s ability to underestimate how quickly industries can change once economic incentives align with new capabilities.

But history also shows that disruption is rarely uniform. Some incumbents fail to adapt, while others evolve and strengthen their positions precisely because they possess scale, infrastructure, customers and distribution advantages that newer entrants struggle to replicate.

For investors, the challenge is therefore unlikely to identify whether disruption exists, but rather in determining how much disruption is already reflected in valuations, and whether markets are appropriately distinguishing between structurally exposed businesses and those capable of adapting as well as demonstrating adaptation.

What is unfolding in software is unlikely to be a binary outcome. Rather, it is shaping up to be a period of separation between businesses that successfully integrate AI into their ecosystems and those that fail to evolve alongside it. Within that distinction lies both significant risk and potential opportunity.