How is AI restructuring the software market?

From tools to outcomes, AI is redefining what software companies deliver and which survive.

A month ago, the release of Anthropic’s Cowork plugin triggered a sharp sell-off in public software stocks, crystallising years of built-up fears around AI. The shock erased roughly $1T in market cap and pushed software ETFs into bear market territory. Key SaaS performance indicators such as the Bessemer Emerging Cloud Index and iShares Extended Tech Software ETF fell as much as 30% and 40% since December 2025.

However, a month on, despite markets still trading 20–25% lower than December 2025, the picture is more nuanced.

The sell-off was not fully irrational. Seat-based pricing models face real structural pressure from AI. AI is eroding value in coding, analytics, and workflow automation. Startups operating purely on top of AI are increasingly difficult to defend, as AI plugins reduce switching costs and replace application-layer value. This is already visible in credit markets: the $235B software loan market is already showing signs of strain. Credit quality is concentrated in B- and below, spreads are at their highest average since the post-pandemic era, and demand for software loans is deteriorating rapidly.

However, markets arguably overreacted. The nuance lies in the composition of the software industry. While much software may become replaceable as AI improves, some companies retain defensibility through compliance constraints, legal accountability requirements, proprietary data, network effects, and as a result high switching costs. This is particularly evident where regulated data cannot easily be moved or where liability cannot be delegated to AI.

As a result, we are not seeing a full market shutdown, but a separation into distinct categories. Companies with clear data moats and the ability to integrate or acquire AI capabilities into their offering will pull through. This implies that the impact of AI on SaaS is narrower than the sell-off implied: only companies with little proprietary data and seat-based tools built on task automation are at clear existential risk.

The remaining software providers are in a more complex position and will take time to adapt. This helps explain why markets remain down since the start of the year. Over time, these companies may evolve from providing tools and insights to delivering outcomes directly; shifting from enabling teams to produce results faster, to becoming the entity that produces the result itself.

Hence, the story in hindsight is not the death of software companies, but a new era with more selective winners. This will still come with lasting impacts, including significant pressure on PE portfolios heavily concentrated in software. Investments into the sector hit record highs in 2025 at $203B, leaving many funds vulnerable to write-downs as AI displaces some companies.

The transition from service-based software to outcome-based solutions has begun. The key question for investors is not whether to hold software, but which software to hold.

Source: Pitchbook