Portfolio review takes on a new urgency for VCs as AI makes 2024’s great deals look like 2026’s clear mistakes

AI is compressing the “half-life of a thesis,” with VCs estimating roughly 10–20% of venture portfolios (in addition to normal failure rates) are especially vulnerable to foundation-model shifts. Examples like Perplexity are cited as getting “caught” as incumbents (e.g., Google) rapidly catch up in AI-powered search. Investors are urged to re-underwrite portfolios by assessing what durable value remains (people, product, or distribution), potentially accelerating exits or returning 60–70 cents on the dollar rather than letting companies burn out to zero.

Analysis

The real market mechanism is a faster depreciation schedule for software moats. If model capability is outrunning product cycles, then the discount rate investors should apply to app-layer growth is effectively higher: a 2024 feature set can be commoditized before it scales, which pressures long-duration EV/Revenue names and raises the bar on retention, integration depth, and proprietary data.

That is constructive for control-point platforms like GOOGL, because the winner-take-most layer shifts upward to distribution and query ownership rather than standalone AI interfaces. The second-order loser is the venture-funded “wrapper” cohort: lower follow-on capital, more down rounds, and more M&A at distressed prices should trickle into public comps through softer software multiples and tighter growth underwriting over the next 1-3 quarters.

The contrarian miss is that not all software is equally exposed. Regulated, workflow-embedded businesses can still defend pricing if AI reduces labor costs faster than it erodes switching costs; in that case, the market may be over-discounting the whole category. What would falsify the bearish software view is stable net retention and accelerating AI attach rates in the next two earnings cycles; if that happens, the thesis of rapid obsolescence is too aggressive.

Key Decisions for Investors

• Go long GOOGL on any post-news weakness; 3-6 month thesis is defensive share capture as AI search commoditizes third-party interfaces. Risk/reward improves if ad growth holds while perceived AI threat broadens.
• Pair trade: long GOOGL / short ARKK for 1-3 months as a relative expression of platform moat versus venture-style duration risk. Target ~10-15% relative outperformance; stop if ARKK breadth and high-multiple software momentum re-accelerate.
• Avoid adding exposure to AI-native app-layer winners with unclear proprietary data advantages; use rallies to trim positions in expensive software until next earnings confirms retention. Falsifier: sustained >100% net retention and accelerating usage monetization.
• Watch private-markets spillovers as an alert, not a trade: if venture markdowns force secondary sales or layoffs, that is a medium-term negative for software hiring and customer spend, but it is not yet a clean public-market catalyst.