It’s still being debated whether AI will disrupt traditional software, but stock market investors don’t seem to be taking any chances.
Data tracked by a16z Growth between October 31, 2025 and February 27, 2026 tells a striking story of divergence. While the Nasdaq 100 Index (NDX) — a broad measure of overall technology performance — held roughly flat over that period, the iShares Expanded Tech-Software Sector ETF (IGV), which tracks software stocks, fell by nearly 30%. That’s not a minor blip. It’s a sustained, months-long selloff that reflects a growing unease among investors about where traditional software companies stand in the age of AI.
The Widening Gap
The chart from a16z Growth makes the divergence visually stark. The Nasdaq 100 barely dipped below zero for most of the period, oscillating in a narrow band and finishing close to flat. The Software Index, meanwhile, began sliding in November, briefly recovered in late December, and then accelerated sharply downward through January and February 2026, hitting a trough of roughly -33% before a partial recovery to around -30%.
The gap between the two lines grew wider with each passing month — suggesting that whatever pressure software stocks are under, it isn’t simply a reflection of broader market stress. Technology as a sector is doing fine. Software, specifically, is where the concern is concentrated.
Why Software?
The anxiety has a name: AI-driven disruption. The fear is that large language models and AI coding tools are compressing the value that traditional software companies once provided — and doing so faster than those companies can adapt.
Replit CEO Amjad Masad has argued that the cost of building software is rapidly falling toward zero, driven by tools that allow anyone to create functional applications without writing a line of code. If building software becomes nearly free, that puts enormous pressure on companies that have historically charged for packaged software as a service. Anthropic CEO Dario Amodei has gone further, predicting that AI could be writing 90% of all code within months. NVIDIA’s Jensen Huang has similarly suggested that his own engineers may no longer need to write code at all.
Those are bold predictions, and not everyone agrees with them. But investors often act on what they fear may happen, not what they’re certain will happen — and right now, a lot of money is being moved out of software stocks on the basis of that fear.
AI Winners and Software Losers
What makes the selloff particularly interesting is its selectivity. Overall tech — as represented by the Nasdaq 100 — hasn’t suffered. The companies driving the AI wave, from semiconductor makers to hyperscalers building AI infrastructure, have more than held their own. It’s the traditional software layer — the SaaS companies, the productivity suites, the enterprise vendors — that is being re-rated downward.
This reflects a broader narrative shift in how the market thinks about the technology stack. AI is being seen less as a feature that software companies can bolt on to their existing products, and more as a potential replacement for the kind of workflow automation that software has historically provided. If an AI agent can handle tasks that once required a dedicated SaaS subscription, the revenue model underlying much of the software industry looks a lot less secure.
Claude Code is now reportedly written almost entirely by Claude itself, a development that underscores how quickly the boundary between human-written and AI-written software is dissolving. Meanwhile, Google Gemini’s traffic surged 643% year-over-year in February 2026, while ChatGPT’s grew 37% — figures that signal enormous and accelerating consumer adoption of AI tools that increasingly overlap with traditional software use cases.
Not a Consensus View — Yet
To be clear, the disruption thesis is not settled. Many software companies are themselves investing heavily in AI capabilities, and some argue that AI will expand the market for software rather than cannibalize it. Enterprise customers, in particular, are slow-moving — they don’t swap out vendors lightly, and the switching costs embedded in legacy software deployments are enormous.
There’s also the question of whether the AI coding and productivity tools currently in the market are genuinely capable of replacing enterprise software at scale, or whether they are, for now, better suited to augmenting individual productivity than to displacing institutional infrastructure.
But the stock market, as it often does, is pricing in a worst-case scenario before the evidence fully arrives. Software stocks are down nearly 30% since October while the broader tech index flatlines — and that gap is a measure of investor anxiety, not necessarily of what will ultimately happen to the industry.
The debate over whether AI will disrupt traditional software is likely to play out over years. The verdict from investors, for now, is that the risk is real enough to act on.