Software stocks just suffered one of their harshest selloffs on record — but the fallout may be opening doors to opportunities. The S&P North American Technology Software Index briefly dipped below 20x forward earnings for the first time, with valuations now sitting near 23x versus its long-term average of 34x. Even so, strategists say the damage reflects forced de-risking, not weaker fundamentals — leaving high-quality software names mispriced after weeks of indiscriminate selling.

Hitting rock bottom: The selling has been so widespread that Jefferies says 42% of the 64 software companies it tracks are trading at or near record-low valuations. That shows investors are unloading the sector as a whole instead of picking winners. Goldman’s David Solomon called the selloff “too broad,” while Janney’s Mark Luschini said traders are acting like “software is on a straight line to zero.” Both suggest conditions may be ripe for a counter-trend bounce. Jefferies’ Michael Toomey put it bluntly: the setup could spark a vicious software rally.

• MicrosoftMSFT
  , ServiceNowNOW
  , and SalesforceCRM
  are being flagged as long-term compounders caught in the selloff.
• Even SnowflakeSNOW
  — down 27% in six sessions — still gets Jefferies’ vote as a top AI beneficiary after securing $200M+ multi-year deals with OpenAI and Anthropic.

The Consumption Gap

The market’s mistake was treating all software the same — even though they’re two very different kinds of businesses. Some software is built for humans to click — think dashboards and seats at Salesforce or Monday.comMNDY

, where revenue depends on people logging in. Other software is built for machines to call — infrastructure like CloudflareNET

, MongoDBMDB

, and DatadogDDOG

, which charge per API call, query, or event. That distinction matters because AI agents reduce the need for human seats but massively increase machine usage.

• One person might click a few times an hour, while a single AI agent can trigger hundreds of calls a minute — and usage-based platforms get paid more as those bots get busier.
• That same shift is putting pressure on IT services built on labor arbitrage, as firms like InfosysINFY
  , CapgeminiCAPMF
  , and CognizantCTSH
  face lower demand as AI replaces billable hours — setting up 12–24 months of estimate cuts while API-driven infrastructure outperforms.

Through the smoke: The market has priced in speculative AI disruption without distinguishing between companies positioned to benefit and those genuinely at risk. Infrastructure providers charging per API call are essentially tollbooth operators — and as AI agents multiply, traffic through those tollbooths will surge. The names trading at historic lows today may be tomorrow’s comeback stories — if investors can separate the road workers from the toll takers.