Software stocks took a hard turn lower as fears about artificial intelligence’s impact on legacy business models escalated into a full-fledged exit. Microsoft (MSFT) shed roughly $440 billion in market value after reporting strong results, the second-worst single-session wipeout on record behind last year’s hit to Nvidia (NVDA). Salesforce (CRM) sank 11%, Adobe (ADBE) slumped after a fresh downgrade, and Europe’s SAP (SAP) fell more than 6% in its steepest drop since 2020. The selloff widened despite a comparatively modest 1.4% move in Meta (META), sharpening a line between AI spenders without near-term payback and platforms seen as monetizing AI with less capital strain.
The core problem is no longer whether AI is the future—it is whether today’s AI spending earns its keep. Investors who applauded the early wave of AI announcements now want math: attach rates, pricing, gross margin deltas, and time-to-payback on billions of dollars in accelerated compute and data center buildouts. Microsoft’s shares sank even with beats on revenue and profit as its capital intensity and the open question of Copilot’s net uplift overshadowed the print. If the AI line of business is additive, it should show up in unit economics or in margin expansion that offsets the depreciation cycle. Instead, the narrative is turning on its head: the bigger the AI ambition, the higher the demanded proof. Until there is clarity on paybacks, the market is disciplining the spenders.
What changed is the perceived threat to the software playbook. For two decades, the model was clear: sell seats, upsell modules, expand price 5% to 10% a year, and defend with switching costs. AI raises three disruptive possibilities. First, smart assistants could reduce seat counts by automating work previously done by incremental users. Second, cross-suite agents could compress vendor sprawl, cutting the number of logos in an enterprise stack. Third, AI features positioned as add-ons may face pricing pushback if customers view them as table stakes rather than premium. That triad is the bear case that landed hardest on Salesforce and Adobe. Salesforce’s 11% slide reflects doubt that AI will produce a durable uplift in average revenue per user without cannibalizing existing modules. Adobe’s downgrade amplified concerns about core creative software facing a wave of AI-native challengers and price elasticity limits on novel features.
SAP’s more than 6% drop—its biggest one-day decline since October 2020—shows the anxiety is global, not just a Silicon Valley mood swing. European software names, already trading with a quality premium for predictable maintenance revenues and mission-critical footprints, are now being reassessed on two vectors: exposure to seat-based pricing and the scale of AI reinvestment required to defend moats. Premium multiples are compressing as investors recalibrate for slower net expansion and the prospect that incremental AI revenue arrives later, at lower margins, or both. The old math of paying 8 to 12 times forward sales for durable compounding assumes steady attach and limited disruption; in an AI transition where product roadmaps are in flux, that assumption no longer holds. The reset did not wait for earnings calls. It arrived via positioning and a collective rethink of what the next three years look like for enterprise software.
One important signal: Meta’s comparatively small decline underscores that not all AI stories are priced the same. Markets are favoring cash-efficient AI applications with visible monetization over infrastructure-first strategies that demand massive capex before yields are proven. Meta can point to AI improving engagement, ad targeting, and time spent today—drivers with short feedback loops. That doesn’t make the company immune to AI risk, but it places it on the side of the ledger where spending can be toggled against near-term return. Investors appear more willing to hold platforms that convert AI into operating leverage than those layering AI on top of already high opex and capex. The message for software CEOs is blunt: if your AI plan requires a new data center line item, be ready to map each dollar to revenue or margin in a timeframe equity holders can live with.
This selloff landed despite beats because guidance and capital allocation now dominate the narrative. A classic beat-and-raise quarter is no longer enough if it sidesteps the central question: will AI expand total addressable market and margins, or will it compress pricing and pull forward costs? Investors want specifics, not slogans. What is the net retention rate with AI upsells? Are AI features cannibalizing existing SKUs or increasing wallet share? How much gross margin drag should be expected from AI inference in product? Microsoft’s experience is instructive. Strong top-line and cloud growth could not offset concern that the cadence and scale of AI investment have outpaced the near-term monetization curve. The equity market is a forward-looking machine; it is now discounting the possibility that the payoff window is longer and messier than management teams suggested last spring.
Software has been prized for recurring revenue and pricing power. AI tests both. Early feedback from procurement circles suggests buyers are resisting broad-based price hikes justified by AI enhancements unless paired with measurable productivity gains, headcount savings, or compliance benefits. That sets up a tug-of-war on list prices and discounting as vendors try to monetize AI modules. If the only way to drive adoption is bundle-and-discount, gross margin trajectories get tougher. Conversely, vendors that can show hard ROI—shorter sales cycles, fewer licenses required elsewhere, automated workflows that retire third-party tools—will have leverage to defend uplift. The sector bifurcation will likely track who can document that proof on customer scorecards and in KPI disclosures. Without that, net expansion slows, making rich multiples indefensible.
Stability will come from transparency. The market wants a dashboard: AI attach rates by product, per-seat uplift, incremental compute cost per user, and the depreciation profile of AI capex. Clear guidance on when AI shifts from investment drag to contribution margin will matter more than the next demo reel. So will commentary on churn risks if AI features cannibalize legacy modules. Software leaders need to articulate cannibalization strategies deliberately—better to self-disrupt and capture the wallet than leave a gap for newcomers. This is also a moment to revisit capital returns. If the capex curve is flattening or if AI spend can be throttled, signaling discipline via buybacks or a steadier margin path can rebuild trust. Vague assurances about transformative potential are out; line-of-sight metrics are in.
The next catalysts will be unforgiving. Adobe, Oracle (ORCL), and other enterprise bellwethers will be judged on AI monetization math as much as on bookings. Expect higher volatility around guidance as management teams recalibrate language and disclose more granular AI KPIs. Institutional positioning has already shifted toward balance-sheet strength and cash flow visibility; it will take concrete evidence of AI-driven net retention and margin durability to draw that capital back into high-multiple software. Until then, the sector sits in a prove-it phase where every quarter becomes a referendum on AI unit economics. The market’s message is consistent across MSFT, CRM, and SAP: spend if you must, but show your work.
