Monday Was 'AI Kills Everything.' Tuesday Was 'Actually, Maybe Not.' Here's What Really Changed.

In the span of 24 hours, the AI narrative whipsawed. Monday: IBM fell 13%, software stocks hit multi-year lows, and a viral research paper imagined AI pushing unemployment to 10% by 2028. Tuesday: Thomson Reuters surged 11%, Salesforce recovered, and Wedbush told clients the AI threat to software was 'overblown.' What actually changed?

The framing shift

On Monday, the story was displacement. Anthropic's announcement that AI could automate COBOL consulting work, IBM's bread and butter, read as a direct threat to an entire business model. Investors extrapolated: if AI can replace IBM's consultants, what else can it replace?

On Tuesday, Anthropic's enterprise event reframed the story as integration. The announcement wasn't 'Claude replaces Salesforce', instead it was 'Claude works inside Salesforce.' The message: software companies that embrace AI may see it as a revenue driver, not an existential threat.

The harder question underneath

Both things can be true simultaneously. Companies that integrate AI into existing workflows may gain efficiencies and grow revenue. And some categories of work that were previously done by expensive humans will be automated.

The real test will be the next 6–12 months of enterprise software earnings, where we'll see whether companies are actually spending more on AI-integrated tools or using AI to spend less on software and people altogether.

Sources: CNBC, Wedbush, Federal Reserve, Bloomberg