Most important take away

Venture is a power-law industry where roughly 30 companies a year generate more than half of the total exit value, so an LP’s job is not to find every great manager but to ensure every manager they back is great — even if that means waiting years between new commitments. The prevailing fear that mega-funds can no longer deliver fund returners is overstated: Vencap has identified 45 investments that returned $1B+ to a single fund (most in the last 7-8 years), and fund sizes today must be compared against exit sizes 10-15 years from now, not today’s.

Summary

Actionable insights and patterns from David Clark’s 32 years as an LP:

On manager selection (career/investing advice):

• Know your lane. Be ruthless about staying focused on what you’re good at; don’t chase every shiny manager. Doing nothing for years is acceptable if the signal-to-noise ratio is poor.
• Output over process. You can only judge an investment process by its outputs over a long enough horizon. Anyone can articulate a compelling process; only realized DPI proves it works.
• Intercept at Fund III. Vencap’s data shows Fund I success has a strong randomness component, but once a manager lands a top 1% company, replication becomes more likely. Roughly half of Vencap’s 12 core managers were first backed at Fund III.
• Two reasons to leave an existing manager: persistent underperformance, or poor succession. Team breakdowns are the #1 silent killer of great firms.
• Reference beyond co-investors. Don’t only ask GPs who have worked with a manager; ask GPs in the same sector who haven’t — and find out why.

On the venture industry’s structure:

• The power law dominates everything. ~30 companies a year drive >50% of all VC exit value globally; the same firm names appear in those cap tables repeatedly.
• DPI reality check: of ~1,200 funds raised 2000-2014, more than 50% failed to return 1x, only 6.6% returned 3x net, and just 2.6% returned 5x. A 5x fund is roughly a 1-in-50 event.
• Vencap’s core-manager mature funds blend to ~3.5x net, with <3% of those funds showing TVPI below 1x — the model is built on minimizing losses while still accessing power-law upside.

On fund size and exits:

• Compare today’s fund sizes to exit sizes 10-15 years out, not today’s exit sizes. Technology’s share of the economic pie is still growing.
• Incumbent dominance is the real long-term risk, but tech still moves in paradigms (mainframe → client/server → internet → mobile → AI, possibly blockchain) and paradigm shifts dislodge incumbents.

On liquidity:

• Liquidity windows are brief — “weeks where years happen.” The way LPs hedge against bad timing is consistent deployment across every vintage; don’t try to time the market.
• The Sequoia evergreen-fund idea (holding public winners) is directionally right — top 1% companies often keep compounding post-IPO (Nvidia, Square/Cash App). Vencap itself sells distributed public stock because that’s not their job, but they let GPs use judgment on timing.
• M&A is structurally constrained (Khan, UK CMA blocking Figma/Adobe). Founders must increasingly build standalone, durable businesses, which will further concentrate returns.

On fund mechanics LPs scrutinize:

• Deployment pace: Vencap pushes for 3-year investment cycles. Their worst-ever fund was a 1999 vintage fully deployed in 15 months.
• Fees and carry: they tolerate up to 2.5/25 (they haven’t seen 3/30) as long as net performance stays top-quartile. They prefer tiered carry for better alignment.
• Fund size discipline: an early-stage fund should still be able to return the whole fund from one investment; a later-stage fund should be able to return at least half.
• Portfolio shape: target ~50/50 early-vs-growth allocation; growth fund DPI comes back faster, and in Vencap’s data the TVPI of growth funds vs early-stage funds from their core managers is statistically indistinguishable.
• Benchmarking: Vencap sends anonymized peer benchmarks (by IRR, TVPI, and DPI, grouped in 3-year vintage cohorts) to managers every 3-6 months so re-up decisions are never a surprise.

On geographies:

• 70% US, 10% Europe, 10% China (down from ~20% over the last decade). They follow the top 1% companies, not geographic mandates. UK/European government pressure to back local managers regardless of performance is a known trap.

On returns going forward:

• More pain still to come for existing funds. Newer managers are holding marks at last-round values; established managers have written down more aggressively. Loss ratios (currently artificially low) will revert to the historical ~60% for early-stage.
• Doug Leone’s “boutique to commoditized” thesis applies to late-stage/crossover capital (where dollars compete entry valuations to efficiency), but seed/Series A/early-B remain a craft business where excess capital is detrimental.

On succession (key for evaluating firm durability):

• Best in class: Accel (multiple generational transitions), Sequoia (Valentine → Moritz → Botha), Foundry Group (honest about being a finite-life partnership).
• The signal: senior partners stepping aside on a real timeline, fresh blood being credited for the top 1% companies they source, and economics flowing to the next generation.

On democratization (a career-relevant observation):

• Venture remains exclusive both for capital access and for who gets into the industry. Clark advocates broadening GP sourcing — he himself only got in because someone took a chance on a graduate from a small village in Northumberland in 1992.

Tech/market patterns mentioned:

• AI as the current paradigm shift; question is whether incumbents (Google, Microsoft) with compute, data, and distribution moats can be unseated as in prior shifts.
• Bytedance and OpenAI as evidence that >$100B outcomes can still emerge despite strong incumbents.
• Blockchain/crypto called out as a potential second paradigm running in conjunction with AI.

Chapter Summaries

Becoming an LP (1992-onward): Clark stumbled into venture answering a vague newspaper ad in Oxford as a fresh graduate. Got hooked four or five years in when handling a Netscape stock distribution and realizing the second-row seat to technology was unmatched.

Is being an LP harder today? Yes in volume of managers, no in approach — stay in your lane, be ruthless about focus, and accept missing some good managers. The 2010-2012 post-GFC window let Vencap add several top-tier names whose Ivy League LPs were paralyzed by the denominator effect.

Power law and the case for venture: ~30 companies/year drive >50% of all VC exit value. The same names show up. Top-quartile compounding still massively beats public markets despite venture’s poor median.

The “fund size is too big” debate: Vencap data: 45 investments have returned $1B+ to a single fund (one at $15B); most in last 7-8 years. Compare fund sizes to exit sizes 10-15 years forward, not today’s.

Incumbents and paradigm shifts: Clayton Christensen and Thomas Kuhn frame the discussion. AI (and potentially blockchain) is the new paradigm; incumbents have longer half-lives but not infinite ones.

Liquidity and M&A: Liquidity is concentrated in brief windows; vintage diversification is the LP’s defense. M&A is structurally constrained, pushing companies to build standalone businesses, which concentrates returns further.

Geographic allocation: 70% US / 10% Europe / 10% China. Follow the top 1% companies, not geographies.

Selection process: Deal flow is entirely outbound. They screen by identifying top 1% companies and tracing back to early-stage investors. Cold-inbound managers get a reply but no meeting. Persistence wins access — moments of firm dislocation (succession, LP turnover) create entry points.

Process vs. output debate with Stebbings: Stebbings argues process (investment decision-making) IS the product; Clark counters that you can only judge process by output, and there’s no single right way to run a partnership (Kosla, Founders Fund/Singerman as examples).

Re-up process: Continuous diligence on 90% of existing managers; full IC paper for every commitment as documentation. The two reasons to drop: performance or succession. They share quarterly anonymized peer benchmarks so managers see exactly where they stand.

Fund mechanics: Push for 3-year deployment; tolerate up to 2.5/25; require single investment can return the fund (early) or half (later); ~50/50 early/growth allocation. Their best-performing core fund of the last 15 years is a growth fund.

Returns outlook: More pain coming as newer managers mark to last round while established firms write down. Loss ratios will revert toward 60%. Late-stage will commoditize; seed-Series-A/early-B remains a craft.

Succession case studies: Accel, Sequoia, and Foundry Group as positive examples. Failures stem from founders holding economics and seats too long.

Mistakes and learning: Decision reviews 4-5 years post-investment. Biggest learning: reference managers in the same sector who haven’t co-invested, not just those who have.

Quick fire: Changed his mind on LP direct co-investments (now selectively positive for established companies). Would democratize venture access. Famously declined Benchmark’s $5M minimum in the mid-90s when a UK pension constraint capped them at $4M.

Closing: Wants to be useful to the next generation at Vencap in 10 years rather than running the firm — succession lived rather than preached.