AI productivity tools are overhyped and overfunded. Investors should look elsewhere

The article argues that most AI productivity tools launched in 2023-2024 are likely to fail, with AI ecosystem revenue estimated at ~$1T in net new additions since ChatGPT (Nov 2022) but characterized as “amongst the riskiest” and potentially low-quality. It cites Northzone analysis that AI applications drive ~$150–200B in ARR, with AI coding representing ~20–30% of that, yet warns standalone tools like generic note takers face existential pressure from open-source and rapid model/application evolution. Northzone positions the next investment focus on autonomous AI “systems of work” (including leading rounds since early 2026 of several hundreds of millions of dollars), implying a shift in venture capital from basic productivity to higher-value innovation.

Analysis

The key market mechanism is bifurcation: standalone application-layer software is becoming easier to replicate, while value migrates to distribution owners, infrastructure, and products that can prove closed-loop task completion. That is bearish for the long tail of mid-cap SaaS and venture-backed AI apps because AI lowers switching costs, compresses feature moats, and forces more spend into compute just to defend seat-based pricing. The first visible pressure should show up in gross-margin commentary and slower net retention over the next 1-3 earnings cycles, not in immediate topline collapse.

Second-order effects favor incumbents that can bundle AI into existing workflows and absorb model costs, including MSFT, GOOGL, and to a lesser extent AMZN. The losers are not only point-solutions but also the channel partners that sold implementation around them: systems integrators, niche data tools, and some cloud resellers tied to usage-heavy but low-ARPU apps. Over 6-18 months, the likely outcome is consolidation rather than category destruction; the market may underappreciate how quickly venture-backed AI apps get forced into either M&A or marginless growth.

Contrarian view: the consensus may be too quick to declare a broad AI-app graveyard. Enterprise buyers still pay for auditability, compliance, and human-in-the-loop liability shields, which slows autonomy adoption outside coding and back-office workflows. The real falsifier is evidence that top-tier AI-native apps can sustain premium pricing with stable retention despite model commoditization; if that happens, the short thesis on application-layer software is overdone.

Key Decisions for Investors

• Short WCLD or IGV on strength into the next software earnings window; target a 1-3 month mean reversion trade if guidance shows slower AI monetization than usage growth. Cover if software NRR and gross-margin commentary remain stable across multiple names.
• Pair trade: long MSFT / short a basket of unprofitable AI-native app names or WCLD. Thesis is that distribution and bundling win while standalones face pricing pressure; risk/reward improves if the long leg is kept to a market-neutral sizing.
• Add on pullbacks to AMZN or GOOGL as infrastructure/hosting beneficiaries of higher inference and workflow automation intensity over 6-18 months. Falsify if cloud growth decelerates despite rising AI traffic or if pricing competition offsets usage.
• Avoid fresh exposure to late-stage AI productivity venture rounds and IPOs until they prove durable retention after the next product cycle; treat any new funding as a watch item, not a quality signal.