Wall Street Divided on Software Stocks as AI Fears Collide With Earnings

The market’s sending mixed signals on software — and one side’s bound to be wrong. Over the past three months, Wall Street has raised two-year forward earnings estimates by 5%, even as AI worries crushed a group of vulnerable software names by 50% in the past year. That gap suggests either fundamentals will catch up to sentiment — or investors are walking away from companies that may be stronger than they appear.

The optimists’ case: Goldman Sachs strategists led by Ben Snider say industries most exposed to AI disruption still delivered double-digit profit growth in Q4. Analysts even raised 2026 earnings-per-share forecasts after results beat expectations — a sign that the feared AI hit hasn’t yet shown up in financials. Still, sentiment’s sour. A record 35% of investors in Bank of America’s global fund manager survey say firms are overspending on AI — even as compressed valuations leave software stocks looking cheap if forecasts hold up.

• That tension has JPMorgan calling for a rebound after what it describes as “extreme” price action, favoring MicrosoftMSFT
  and ServiceNowNOW
  .
• Meanwhile, the broader selloff adds pressure as firms replace lost equity pay with cash — cutting into free cash flow just as AI dealmaking heats up.

Tuning Into the Flip Side

Polar Capital’s Nick Evans, whose global technology fund beat 99% of peers over the past year, is rejecting the optimism. He’s slashed most application software exposure, arguing the sector faces an existential AI threat and that only a few players will survive. The sales backdrop supports his caution as AI vendors say the easy wins are gone, with deal cycles stretching from 60–90 days last year to roughly six months now. Regal CRO Alex Levin said enterprise sales that once closed after a single demo now involve legal and finance teams, stricter ROI checks, and hesitation as AI evolves too fast to pin down.

• Evans likens today’s software landscape to 2000s newspapers, saying AI coding tools are now replacing and reshaping products just as new tech once upended old media.
• A US software ETF is already down 22% this year, a sharp contrast to chip stocks surging on AI-driven computing demand.

The illusion of choice: Evans has rotated heavily into semiconductors and leaned into networking, fiber, and data-center power — while keeping selective exposure to infrastructure software like CloudflareNET

and SnowflakeSNOW

that underpin AI systems. That positioning aligns with Morgan Stanley, which recently raised its 2026 hyperscaler capex forecast to $740B from $570B, citing demand for compute that still far exceeds supply. The open question is not whether AI will disrupt software, but whether earnings can rise fast enough to keep investors from losing faith.