Why AI Is Shaking the Software Industry – and Who Could Still Win

The technology sector is undergoing one of its most turbulent periods in years. Nowhere is this more visible than in software, which long seen as the safe, high‑margin backbone of the digital economy. While semiconductor stocks continue to rise, software valuations have come under heavy pressure. The global semiconductor sub‑index (BBG World Semiconductors Large, Mid & Small Cap Price Return Index) has gained roughly 9% YTD (measured in USD), whereas the global software sub‑index (BBG World Software Large, Mid & Small Cap Price Return Index) has lost more than 18% over the same time.* 

For decades, Software‑as‑a‑Service (SaaS) thrived on predictable subscription revenue and steady price increases. This model outperformed the broader market for years until late 2022, when generative AI reset expectations. Since then, semiconductors have surged on AI‑hardware demand, while software multiples have compressed. U.S. software P/E ratios have fallen from ~50x in mid‑2025 to about 35x today.*

What’s Driving the Pullback

Several fears now dominate investor sentiment: AI writes code efficiently, raising questions about how much value traditional software delivers. Competition intensifies as AI lowers barriers to building new tools, potentially eroding margins. AI increases computing costs, pressuring profitability. Furthermore, seat‑based pricing weakens, as AI agents are not tied to user logins.

These concerns escalated after the launch of autonomous AI modules such as Anthropic’s new “Claude CoWork” plug‑ins, which triggered a sharp sell‑off across software names. This reaction has pushed institutional investor positioning to historic lows.

Is This a Structural Crisis or an Overreaction?

Despite the turmoil, AI is also fuelling one of the largest infrastructure booms in decades. Hyperscalers are deploying hundreds of billions of dollars annually into data centres and AI‑optimized chips, supported by record semiconductor sales and rapid growth in hardware demand. Unlike the dot‑com era, today’s leaders are profitable platforms with entrenched workflows and large enterprise footprints.

The central question for investors is not whether AI reshapes software, but who it reshapes.

Not all software is equally exposed. AI may automate interfaces and code, but it cannot easily replicate proprietary, protected datasets. Legal, financial, scientific, and regulatory‑grade databases, aggregated over decades, remain defensible assets. By contrast, software products that rely primarily on user interfaces or simple workflow logic face the greatest risk. A clearer framework emerges across three layers of digital value creation:

• Interfaces: Tools built around user experience or workflow logic are under the most pressure. Their seat‑based pricing models struggle when AI agents perform more work without adding human users.

• Data: Exclusive, verified, domain‑specific data is becoming the most valuable input in the AI era. Many data‑rich companies already charge on a usage basis, meaning AI can increase their revenue as query volume grows.

• Platforms: We believe these are positioned to be the biggest long‑term winners as they combine identity, workflow, billing, and data in one ecosystem. They can blend subscription fees with usage‑based AI charges, therefore scaling as automated workloads increases.

The Business Model Shift: From Users to Usage

AI is accelerating the move away from per‑seat software pricing toward consumption‑based monetization: compute cycles, data calls, transactions, and automated tasks. Companies with proprietary data, deep integration, or platform economics benefit most. Those offering easily replaceable interfaces face increasing headwinds.

What This Means for Investors

The current sell‑off reflects multiple compression, not collapsing earnings. While fear is driving indiscriminate repricing, we think platforms, hyperscalers, and data‑rich providers are positioned for long-term growth. The shift is painful, but it is restructuring the industry around defensible assets and scalable AI economics.

In short: AI is not killing software, just weak software models. The centre of gravity is shifting toward scale, data, and platform effects, and that is where we believe long‑term opportunity is emerging.