E290: How LPs Underwrite Venture in 2026

Episode Summary

Executive Summary: Narayan argues venture capital feels more chaotic than ever because markets are being reshaped by AI, unusual compensation and funding events, and fragmented data that often misleads investors. Franklin Park’s response is to prioritize long-term founder relationships, disciplined underwriting, continuous reference checks, and specialist exposure, especially where buyers are willing to swap incumbents for better software.

Main Topics: The market is unusually noisy and regime-shifted (Priority: 5/5): Narayan says today's venture environment is unprecedented: massive AI spending, bizarre compensation events, and wildly different company trajectories make past patterns less useful. The core challenge is distinguishing temporary froth from durable shifts. Why data is an imperfect guide in venture (Priority: 5/5): He warns that venture datasets are often biased, incomplete, lagged, and unrepresentative, especially for niche slices and first-time funds. LP-level data is better than GP-reported data, but still imperfect because access and selection bias remain. How Franklin Park underwrites managers and funds (Priority: 5/5): Franklin Park uses LP perspective, large reference networks, and behavioral diligence to assess GP quality. Narayan emphasizes revealed preferences, how managers behave under stress, and continuity of presence in the market over flashy narratives. Founder magnetism, relationship capital, and long-term trust (Priority: 4/5): The discussion highlights that venture is fundamentally about whether founders want to give up equity to a particular VC. Long-term trust, founder love, and a stable alumni network matter more than pure velocity or branding. Specialists vs. generalists in a fragmented market (Priority: 4/5): Narayan argues that specialist firms may have an edge in areas like cybersecurity and drug discovery, where founder communities and sales motions are idiosyncratic. Generalist firms face pressure in a market with many first-check options. Product, distribution, and media as VC value-add (Priority: 3/5): The conversation explores whether media and distribution capabilities are now core to venture. Narayan suggests that founders mainly care about minimal brand credibility, responsive financing, sanity in process, and whether the firm can help with product or distribution. AI changes both startup creation and incumbent replacement (Priority: 5/5): He is highly convinced that small teams can now build much more, and buyers are more willing than ever to replace legacy software with better AI-native products. This creates a favorable backdrop for venture despite the noise.

Key Arguments: The market is 'untethered' because events are happening without precedent, making historical comparisons less reliable. Venture data is frequently biased by non-reporting, self-selection, lag, and missing top funds, so backward-looking analysis can mislead. LP-sourced and independent data sets are more trustworthy than GP-reported ones, but access and LP composition still bias results. For Franklin Park, the best signal is revealed behavior: how GPs act in difficult times, not what they claim in meetings. A durable VC franchise combines founder love, disciplined stage focus, operational stability, and a positive alumni network. High-velocity investing can create option value, but it also produces forgettable cap-table presence and weak underwriting. Specialist investors can outperform in verticals with unique founder and buyer dynamics, like cyber and drug discovery. The current AI wave increases both build speed and the willingness of buyers to switch from incumbents, which could unlock major value creation.

Data Points: Franklin Park AUA/AUM: $21 billion - Narayan states Franklin Park is a Philadelphia-based firm with about $21B in assets under management/advisement. Meta market cap: $1.6 trillion - Used as an example of how AI talent spending may be rational when viewed against future upside. Hypothetical Meta future market cap: $10 trillion - Narayan suggests Meta's AI strategy makes sense if AI drives the company toward this scale. Example compensation package: $1.5 billion - He cites a reported package for a single person leaving a company he co-founded as evidence of unprecedented market behavior. HF0 accelerator program length: 12 weeks - He references a San Francisco AI accelerator that puts founders into focused 'monk mode' for 12 weeks. HF0 ARR growth example: $500,000 to $10 million ARR - Example of companies scaling rapidly during the accelerator program. HF0 initial economics: $1 million for 20% - Former model mentioned for the accelerator's financing terms. HF0 later economics: 3% for services - Narayan notes some founders now prefer services-only arrangements over taking the million dollars. Addepar platform assets: ~$7 trillion - Referenced as a large LP-driven data source that is converging with MSCI Burgiss-type data. OpenAI hypothetical valuation: $1 trillion - Used in a discussion of how a few big wins can dominate a portfolio's return profile. Potential venture fund outcome: ~2x - Narayan notes a fund could still return around 2x even if many bets go to zero, depending on one or two major winners.

Pivotal Quotes: "Why do founders want to give up part of their precious life’s work to somebody else?" — Narayan: He frames this as the core venture question beneath all market noise and data analysis. "We think that groups that have engendered a lot of founder love by virtue of being in the market, having board seats, being in the trenches for decades... has a lot of merit." — Narayan: Explaining why long-standing relationships and credibility matter in manager selection. "The ability to build and the ability or the willingness for buyers to think about swapping incumbents... are both at all times high." — Narayan: His most concrete conviction about the current market opportunity, especially in AI/software.

Implications: Investors should expect more noise, less reliable historical signal, and greater importance of relationship-based sourcing and behavioral diligence. For founders, AI-native speed and buyer openness make incumbents more vulnerable and specialist, trusted partners more valuable.