Enterprise software stocks are taking a beating. The WisdomTree Cloud Computing Fund is down 20% in 2026, with individual vendors faring worse: HubSpot down 39%, Figma down 40%, Atlassian down 35%. Even stalwarts like Salesforce and ServiceNow have lost a quarter of their value.
The trigger: generative AI tools that can build apps, websites, and workflows in seconds. Investors are betting enterprises will stop paying for software and just build it themselves with AI.
Vendor CEOs are calling this wrong. Box's Aaron Levie argues businesses would rather pay specialists than carry the liability of building critical systems in-house. Salesforce's Marc Benioff points to data moats: "We've got all the customers' data." ServiceNow's Bill McDermott positions his platform as "the semantic layer that makes AI ubiquitous in the enterprise."
They have a point. Global IT spending is forecast to exceed $6 trillion in 2026, driven by AI infrastructure. Over half of organizations are implementing foundational AI principles, but they're sequencing carefully: predictive AI for core operations, generative AI for R&D.
The real question is displacement speed. Anthropic's Claude Cowork just added legal and marketing features, the kind of specialized workflows that kept software companies valuable. If AI agents can handle these tasks without vendor platforms, the valuation compression makes sense.
What this means in practice: CTOs evaluating software renewals now have AI build-vs-buy calculations to run. For vendors, the "we have your data" argument only works if integration complexity stays high. ServiceNow's semantic layer bet is interesting because it assumes enterprises will need orchestration even if individual tools get commoditized.
Worth noting: Box CEO Levie calls this "the most exciting moment" in 20 years. The cognitive dissonance between vendor optimism and market panic is notable. One of them is mispricing risk.
The pattern matters here. Software stocks got hammered in the cloud transition too. The vendors that adapted survived. The question is whether AI is a platform shift or a replacement cycle. We'll have clearer answers when Q1 renewal data comes through.