AI Fears Keep Hammering Software Stocks—Even Those Reporting Good Earnings

IBM and ServiceNow both beat Q1 estimates, but the market sold off software stocks sharply anyway as AI disruption fears intensified. ServiceNow fell 18% after $3.77B in revenue (+22% YoY) and EPS of $0.97 beat expectations, while IBM dropped more than 9% after $15.92B in revenue (+9.5% YoY) and EPS of $1.91 also topped consensus. The broader software sector weakened, with major names down roughly 8%-10% and the software ETF off 6% on the day.

Analysis

The market is no longer pricing software on current execution; it’s pricing a regime change where AI compresses willingness to pay for seat-based and workflow-based software before revenue can be rearchitected. That creates a nasty second-order effect: even “beat-and-raise” prints can become liquidity events because long-duration software names are being derated on multiple compression rather than fundamentals. The pain is broadest where products are easiest to narrate as replaceable by copilots or prompt-based workflows, which explains why workflow, design, HR, and horizontal app layers are getting hit harder than infrastructure-linked software.

The relative winner set is not just semis, but any vendor whose value proposition is tied to enabling AI deployment rather than being displaced by it. TXN’s move signals that the market is rewarding picks-and-shovels hardware exposure even when the underlying end market is mixed; the next-order trade is that enterprise software budgets may migrate from traditional SaaS line items into GPU, networking, storage, and data governance. That suggests names with AI-enablement leverage and less direct application substitution risk should continue to outperform on any software-led drawdown.

The key risk is that this becomes a forced de-grossing event across hedge funds and long-only software portfolios over the next 1-3 months, not a one-day overreaction. If the upcoming mega-cap cloud and platform earnings show that AI monetization is still not offsetting software substitution fears, the de-rating could extend another 10-15% across the group. A reversal needs either visible AI-driven net retention stabilization or management commentary that shifts the narrative from replacement to augmentation; absent that, every beat is likely to be sold.

The contrarian setup is that consensus may be overestimating near-term product replacement while underestimating enterprise procurement inertia. Most large companies won’t rip out core systems in one budget cycle, which means the real damage is likely to show up first in valuation, not in immediate revenue collapse. That makes the best long opportunities likely to be in the most oversold, cash-generative names only after earnings calls confirm that AI is additive to margins or workflow attach, rather than a blanket short on all software.

Key Decisions for Investors

• Short NOW / long NVDA or AMD as a 1-3 month relative-value trade: if software multiple compression continues, workflow SaaS likely underperforms AI compute beneficiaries; target 10-15% downside in NOW versus stable-to-higher semis.
• Initiate a basket short in WDAY, TEAM, ADBE, and FIG into the next earnings window: these names are most exposed to the 'AI can replicate the app' narrative; use a 2-4 week horizon and cover on any guidance that explicitly quantifies AI-driven attach or retention lift.
• Use call spreads in TXN for a 2-6 month bullish expression on capex reallocation into AI infrastructure: upside comes from sector rotation rather than end-demand perfection, with lower risk than chasing NVDA after a run.
• Avoid fresh longs in CRM and NET ahead of results unless valuation resets further: both can be trapped in the same de-rating basket even if execution is fine, so the risk/reward is poor until the market stops selling software on headlines.
• For long-term contrarians, begin scaling into IBM only on further weakness and only if consulting/bookings commentary proves AI is augmenting delivery economics; treat it as a 6-12 month mean-reversion trade, not a tactical bounce.