Wall Street closed sharply lower on Tuesday, February 3, 2026, as investors kept repricing what AI means for software vendors. The S&P 500 fell 0.84% to 6,917.81 and the Nasdaq Composite dropped 1.43% to 23,255.19; the Dow had touched a record intraday high of 49,653.13 early in the session before fading 166.67 points (-0.34%) to 49,240.99. The real damage, though, was concentrated in software — and it was not a one-day shock but a sector-wide re-rating still in progress.
The Session in Numbers
CNBC’s live coverage made the breadth of the selloff easy to see:
- ServiceNow and Salesforce each closed down close to 7%; Intuit dropped nearly 11%
- The iShares Software ETF (IGV) fell about 5% on the day, roughly 17% lower year-to-date and about 28% below its recent high
- The WisdomTree Cloud Computing Fund (WCLD) lost 3.2%, its sixth straight down session
- HubSpot, ServiceTitan, ServiceNow, Atlassian, Klaviyo, DocuSign, Asana, PagerDuty, Salesforce, Workday, and Adobe all printed 52-week lows
- Microsoft and Meta lost more than 2% and Nvidia slumped nearly 3%; Palantir was the exception, surging almost 7% after beating estimates with $1.41 billion in Q4 revenue and 25 cents in adjusted EPS
The Trigger: From Claude Code to Anthropic’s Legal Tools
The narrative behind the selling is that AI now lets companies build their own software. CNBC’s coverage singled out tools like Anthropic’s Claude Code, which let enterprises replace functionality they used to buy from SaaS vendors with in-house builds — a direct threat to subscription revenue, pricing power, and margins. Jim Cramer, on his show that day, pointed more specifically at new legal tools Anthropic shipped for its Cowork product as the latest catalyst. Bill Northey of U.S. Bank compressed the fear into one phrase: the “potential disintermediation that can occur from artificial intelligence.” Worth noting: the fear is about future competition, not current results. Most of the names hitting 52-week lows had not posted failed quarters — what changed is the story about who builds enterprise software three years from now.
Multiple Compression, Not Broken Earnings
Cramer’s framing is worth keeping: this is multiple compression, not an earnings collapse. Year-to-date, ServiceNow is down about 28%, Salesforce nearly 26%, and Intuit more than 34%, yet earnings estimates for these companies have not fallen apart — investors are simply paying less for the same profits. His warning is blunt: when the multiple keeps contracting, “you don’t know how low it can go.” He argues for heavy selectivity, bought CrowdStrike for his Investing Club portfolio on the theory that cybersecurity is harder to replicate with AI tools than generic SaaS, and noted that Procter & Gamble, FedEx, and Union Pacific — companies that use software rather than sell it — were Tuesday’s winners. In other words, what is falling is the valuation multiple, not the cash flow — which is also why the selloff can drag on: there is no earnings downgrade to force a floor under it.
Contagion: Private Credit and Rotation
The fear has spilled beyond tech. Private credit names with software exposure got hit hard: Blue Owl, TPG, Ares, and KKR fell double digits, Apollo lost 7%, and BlackRock 5%. Ritholtz’s Josh Brown described popular trades getting “absolutely nuked.” Two background signals sharpen the picture: Citadel’s Ken Griffin suggested some firms may invoke AI to mask the correction of past over-hiring, and Bitcoin slid below $74,000 to its lowest level since November 2024 — evidence that risk aversion was not contained to software.
What It Means for Developers and Product Teams
Three takeaways. First, the market is now pricing in AI agents replacing per-seat SaaS licenses: if agentic tools like Claude Code and Cowork really let enterprises do the customization themselves, usage-based and seat-based models get repriced first. Second, the contrast between the losers and the exception is informative — Palantir rallied on strong AI-driven results, which says the market is not punishing AI as a theme; it is punishing legacy business models without an AI story. Third, treat this as a roadmap signal: once investors assume “companies will build their own tools with agents,” every SaaS product has to answer why an agent cannot write it out of existence. In pricing terms, this move turns software’s pricing power into a new risk variable — when the marginal cost of custom software approaches zero, the premium gets hard to defend. For the broader arc of AI shifting from a model race toward agent deployment since the start of 2026, see our opening outlook.
Sources
- Wall Street ends sharply down as traders fret about AI disruption — Reuters
- Stock market news for Feb. 3, 2026 — CNBC
- AI disruption fears rock software stocks again. How Jim Cramer is navigating the sell-off — CNBC
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.