The software industry is currently facing a significant period of disruption, often described as a “SaaSpocalypse,” driven primarily by the rapid advances in generative AI tools. Since October, the sector has lost roughly 30% in market value with some firms, such as Workday (-41%), ServiceNow (-46%) and Salesforce (-27%). The sector is being repriced not merely on cyclical concerns but on a structural reassessment of its long-term competitive advantage.

“The double-digit decline in valuation in the software industry is observable across the world,” said Guy Stear, head of developed markets research at Amundi Investment Institute on 13 February 2026 in an interview.

From tariff shield to AI shock

This downturn marks a sharp reversal from early 2025, when software initially outperformed hardware because it was perceived as more resistant to potential trade tariffs. However, by the second half of 2025, the market logic shifted. What had been viewed as tariff-resilient intellectual capital was suddenly perceived as technologically replicable.”

Marc Andreessen, partner of Silicon Valley venture capital firm Andreessen Horowitz, is reported to have said in 2011 that “software is eating the world.” By devaluing the very code these companies produce, investors realised last year that “AI was effectively eating software,” said Stear, recoining slightly the quip.

The erosion of the software “moat”

While the emergence of DeepSeek in January 2025 may have acted as an immediate catalyst, a central driver of this volatility is the emergence of AI models like Anthropic’s Claude 3.5 (rolled out in early 2024), the plugin ecosystem for Claude Code, and ChatGPT. Stear noted that these new tools have demonstrated an astonishing ability to write code quickly and efficiently.

Stear argued that the ability to automate code generation significantly reduces the intellectual property value and the competitive “moat” that traditionally protected software firms.

What was once a vast protective barrier for these companies has, in the eyes of the market, shrunk to a “puddle,” noted Stear. Consequently, the long-term financial visibility that investors previously enjoyed—often spanning 10 to 15 years of predictable recurring revenue—has “vanished almost overnight,” argued Christopher Dembik, senior investment advisor at Pictet Asset Management, in an interview on 13 February 2026.

According to Dembik, positioning data suggests hedge funds remain net short. Stear added, "We are already seeing downgrades to 2026 earnings expectations, and I expect those revisions to accelerate.”

Dembik stressed that most companies remain sceptical about investing in AI tools, as the short-term return on investment is not yet clearly visible. Widespread adoption therefore seems optimistic in the near term, despite their strong long-term transformative potential.

Winners, losers, and strategic survival

Two characteristics appear to provide relative resilience: scale distribution and data entrenchment.

• Consumer and small business providers: Companies such as Microsoft that sell off-the-shelf products to individuals or small firms are expected to be more robust, as their clients lack the resources to build their own bespoke AI-driven software.

• Legacy data holders: Firms that have managed a company’s legacy data for long periods are more “impervious” to threats because the cost and complexity of switching to a new system are prohibitively high. For Stear, it remains unclear whether adapting legacy systems will prove more difficult than rebuilding workflows using AI-generated tools.

While some major players like Salesforce or SAP may successfully integrate AI due to their access to critical corporate data, Dembik argued that there is profound uncertainty regarding their future revenue streams. Even fundamentally strong companies with attractive valuations (price-to-earnings ratio of 12), such as Adobe, have seen their stock prices hit irrationally hard. Investors reassessed whether their historical pricing power remains defensible in an AI-native environment.

Conversely, Dembik suggested that companies that provide interface-layer functionality without control over core enterprise data – and are not operationally indispensable – such as Zoom will likely suffer more. More generally, he added, "We are a long-term investor in the sector. Yet we remain prudent, and we are not adding to our current positions.” `

The shift to the “physical economy”

If AI reduces the scarcity value of code, the next frontier may lie in embedding intelligence into physical systems. The next wave of innovation will likely move from the digital realm into the physical economy, involving drones, sensors, and mechanical engineering in sectors with weak productivity like construction and building maintenance.

Drones equipped with sensors could collect real-world data and transmit it to centralised systems operating like large language models, creating a feedback loop that continuously improves real-world performance.

This shift suggests a geographic diversification of opportunity, favouring regions with strong traditions in mechanical and electrical engineering, such as Northern Europe (including Northern Italy), Northeast Asia, and China.

Investment implications and market structure

For investors, he suggested applying a “barbell strategy,” focusing on small, innovative SMEs for software prototype development and large, established firms capable of scaling production through superior operations and M&A activity. Dembik stressed, however, that small players face significant hurdles, including high capex requirements and the difficulty of gaining visibility outside of major ETFs.

In summary, while the software sector remains under pressure from speculative shorting and structural AI disruption, the broader technological landscape may be shifting toward the integration of intelligence into physical infrastructure. Whether the current repricing proves excessive or prescient will ultimately depend on how quickly firms can rebuild defensibility in an AI-accelerated world.