Software stocks suffered their worst single-day losses in years Tuesday, with Thomson Reuters down 15.83%, Adobe falling 7.31%, and Salesforce dropping 6.85%. The trigger: Anthropic's release of AI tools targeting legal workflows, which Wall Street read as proof that AI could replace entire software categories rather than just enhance them.
The panic spread beyond legal tech. Companies selling workflow software, from LegalZoom to Expedia, saw sharp declines. Private equity firms with software-heavy portfolios, including Ares and Apollo, also took losses as investors reassessed valuations across the sector.
History suggests caution here. Similar fears emerged when cloud threatened on-premise software in 2010, when mobile threatened desktop in 2008, when the internet threatened client-server in 1995. Each time, incumbents adapted and new categories emerged. The question isn't whether AI disrupts software, it's which vendors adapt fast enough.
The timing is notable: this follows months of enterprise AI hype where software vendors positioned themselves as beneficiaries. Now the market is pricing in the opposite scenario, where AI models become the interface and traditional software becomes infrastructure or disappears entirely.
For APAC CTOs watching this unfold, three implications matter. First, software procurement strategies written six months ago may need revision. Second, vendors claiming AI partnerships will now face harder questions about their actual defensibility. Third, the premium paid for enterprise AI features just got harder to justify when foundation models offer similar capabilities.
JPMorgan analysts called the sell-off premature, noting that "software stocks are now sentenced before trial." That may be right. But the speed of the market's repricing suggests institutional investors see something CTOs should consider: the economics of enterprise software might be fundamentally changing, faster than anyone budgeted for.
Worth noting: Anthropic's tools remain unproven at enterprise scale. The gap between demo and deployment is where many AI promises have stalled. We'll see if this time is different.