The Twenty Minute VCWhy VC Today is Worse than 2021
CHAPTERS
- 0:00 – 0:49
VC mood check: doubling down on giants & vertical SaaS skepticism
The episode opens with two framing takes: big money may flow to the largest incumbents, and vertical SaaS/vertical AI may be over-romanticized. The hosts tee up a broader theme—today’s VC environment feels late-cycle, with high expectations and fragility if growth decelerates.
- •“Double down on the biggest companies” as a 2025 money-making play
- •Skepticism that verticals stay attractive in an AI era
- •Warning: aggressive risk-taking looks smartest right before a crash
- •Early signal that VC is feeling “as tough as it’s ever been”
- 0:49 – 3:04
Everett Randle joins Benchmark: why top firms reshuffle talent
Harry, Rory, and Jason react to Everett Randle joining Benchmark and what it says about elite-firm recruiting. They discuss how firms like Benchmark rarely add partners, how they source talent, and why ambitious investors move quickly across top platforms.
- •Benchmark’s partner additions are rare and signal confidence
- •How top firms recruit: overlap on deals, call references, offer equal partnership
- •Career velocity: “VC shuffle” across brand-name firms as a strategy
- •Benchmark/Kleiner talent benches remain deep despite departures
- 3:04 – 5:38
Is VC still the best-paid job in tech? The new comp reality
The group debates whether venture remains a top compensation path versus Big Tech—especially for elite AI engineers. They contrast liquid, near-term stock packages with VC’s long-duration carry model.
- •Backdated carry and strong carry pools can be highly attractive
- •Big Tech AI comp can dwarf VC on a liquid, 4-year vesting basis
- •VC wealth is often “paper” wealth for a long time
- •Winner-take-all labor markets now apply to engineers and investors
- 5:38 – 6:59
Carry payouts, delayed distributions, and the ‘get rich slow’ program
Rory and Jason explain how carry timing works in practice and why waiting years (or decades) for distributions is common. The conversation highlights how staying private longer changes liquidity expectations for investors and fund managers.
- •First carry checks can arrive early, but long droughts are common
- •Post-bubble periods can produce 10-year stretches of minimal payouts
- •Paper markups vs real liquidity: the core emotional/financial tension
- •Longer private-company timelines push returns further out
- 6:59 – 7:45
Revolut at $75B: private markets delay the public exit
The Revolut valuation jump becomes a case study in how mature companies can remain private despite IPO readiness. The hosts discuss why private capital keeps funding “public-ready” businesses and what that implies about today’s market structure.
- •Revolut’s scale: major revenue, profitability, and growth—IPO-ready
- •Private markets increasingly ‘own’ companies that could be public
- •Late-stage rounds become public-style investing with private terms
- •Valuation context vs UK banks underscores the TAM bet baked in
- 7:45 – 17:02
The TAM myth vs TAM exhaustion: founders expand markets, but not infinitely
Jason argues TAM exhaustion is becoming visible across portfolios, while Rory emphasizes starting with a niche wedge and expanding into adjacent markets. They debate whether “founders create the TAM” is over-stated, and how market size caps outcomes.
- •Jason’s heuristic: prefer companies at $100M ARR with tiny real market share
- •Rory’s counter: best wins start niche, then expand into adjacent empty space
- •Market size sets the prize; CEO execution determines who wins it
- •Thinly sliced SaaS markets can cap outcomes regardless of founder quality
- 17:02 – 18:46
AI verticals: hype, lofty ARR assumptions, and why legal is the battleground
The conversation shifts to vertical AI, with Jason warning that investors are funding too many niche plays on aggressive forward assumptions. Legal emerges as a key example: attractive workflow/data structure, but potentially smaller TAM than excitement suggests.
- •Concern about pricing AI verticals on unrealistic ARR trajectories
- •Legal AI: sticky enterprise workflows, but TAM may be overestimated
- •Valuation risk: paying for a future that may not materialize
- •B2B vertical AI could see many losses with a few big winners
- 18:46 – 30:30
The ‘Covid mistake’ in AI: everyone is in-market now—don’t extrapolate forever
Jason introduces a crucial idea: AI is pulling buyers into the market simultaneously, compressing multi-year adoption into a short window. Rory agrees this could distort growth signals and lead to over-extrapolation—similar in shape (not cause) to Covid-era demand spikes.
- •AI creates a temporary “everyone is shopping now” dynamic
- •Growth rates today may reflect market-wide urgency, not steady-state demand
- •Vendor switching and business process change are expensive and slow
- •Winners must be captured quickly before the market settles
- 30:30 – 32:10
Sell vs hold: M&A decisions under TAM constraints and exit-bar inflation
Jason shares a new rule of thumb: if TAM isn’t truly accelerating, taking a strong acquisition offer may be rational even if revenue can still grow. They connect this to the modern exit bar—where $100M ARR is no longer “enough” for venture-scale outcomes.
- •M&A choice depends on whether TAM growth outpaces revenue growth
- •You can keep growing while enterprise value stops compounding
- •Exit thresholds rising toward $1B+ revenue changes the whole calculus
- •Overfunding niches increases downside risk for founders and investors
- 32:10 – 41:34
Vertical SaaS ‘truth’: why Toast is the benchmark and AI must raise spend per customer
The hosts pressure-test vertical SaaS economics using Toast as the reference outcome. The key question becomes whether AI can 10x customer spend (e.g., $10k/year to $100k/year) by replacing labor—otherwise venture math may not work at today’s entry prices.
- •Classic vertical SaaS math: 10k customers × $10k/year = $100M ARR
- •Modern venture needs bigger outcomes; $100M ARR is often insufficient
- •AI bull case: replace humans → dramatically higher ACV per customer
- •Entry valuation discipline matters more than ever
- 41:34 – 46:25
OpenAI, Microsoft, and Oracle: cloud power dynamics & offloading balance-sheet risk
OpenAI’s potential to spend more with Oracle than Microsoft becomes a lens into partnership leverage and capital intensity. Rory and Jason argue OpenAI is skilled at shifting infrastructure risk to others, while incumbents weigh what’s economically rational.
- •Microsoft’s stance: avoid economically irrational spend; keep strategic upside
- •Oracle’s incentive: buy a seat at the table, even with higher leverage risk
- •OpenAI’s strategy: commit demand while pushing capex risk downstream
- •Risk-return asymmetry for compute providers vs model/platform owners
- 46:25 – 56:32
Poolside builds a data center: the shocking capital intensity of competing in AI
Poolside’s decision to build (not just rent) massive compute capacity raises the stakes for model-layer competition. The group discusses why a software company might be forced into fixed-asset strategy—capacity scarcity, rising competitive bars, and the need to secure GPUs.
- •Poolside positioning: enterprise coding/LLM stack and runtime ambitions
- •Building a data center signals capacity scarcity and elevated competitive bar
- •From SaaS-light to infrastructure-heavy: ‘boiled frog’ of capital intensity
- •If true broadly, model competition may demand multi-billion-dollar scale
- 56:32 – 1:03:28
Bubble vs bust in AI: overbuild risk, inference-driven demand, and surviving corrections
They explore what a real “bust” would look like: overbuilt capacity, falling asset prices, and years of reduced investment—without AI ‘not working.’ Jason adds a concrete demand driver: future apps consuming orders of magnitude more inference via always-on workflows.
- •Bust mechanics: mild glut → price drops → no one builds new capacity
- •AI can still be transformative while adoption timing disappoints
- •Inference demand may expand dramatically (always-on, multi-pass workloads)
- •Investor strategy: capture upside while remaining resilient to a correction
- 1:03:28 – 1:13:47
Why VC today feels ‘worse than 2021’: pacing, LP pressure, cycles, and venture vs public returns
The conversation turns inward to venture fund dynamics: compressed fundraising cycles, lack of temporal diversification, and whether venture is earning its illiquidity premium. Rory argues long-run data supports VC, but cyclical overfunding and recent underperformance vs public markets create pressure.
- •18–24 month fundraising cadence increases cycle risk and crowding
- •In bull markets, the most aggressive looks smartest right before the crash
- •Recent venture returns lagging public markets threatens capital allocation
- •True VC cycles are multi-decade; underfunded windows are rare recently
- 1:13:47 – 1:20:29
OpenAI erotica & content moderation: permissiveness, privacy, and moral boundaries
A lighter topic reveals serious governance challenges: what happens when AI systems generate sensitive content and effectively “author” it. They discuss why content moderation is uniquely hard for LLMs, plus the personal/privacy discomfort of AI chat histories.
- •Erotica demand is predictable; the harder issue is broader boundary-setting
- •LLMs ‘write’ content, raising liability and governance stakes vs social platforms
- •Content moderation is politically and technically difficult to “solve”
- •Privacy: discomfort with others seeing chat histories reflects trust gaps
- 1:20:29 – 1:34:11
Lightning round: Replit to $1B ARR, Deel vs Rippling, and what ‘early’ even means now
They close with rapid-fire bets: whether Replit can hit $1B ARR quickly, and which is the better ownership bet—Deel or Rippling. The debate circles back to the episode’s core tension: the bar for venture outcomes keeps rising, and even $1B ARR can feel “early” in today’s market framing.
- •Replit bull case: exploding software creation; dev shops/agency spend as TAM
- •Skeptic view: churn, prosumer limits, and cohort maturation may slow growth
- •Deel vs Rippling: acute international pain vs defensibility and installed base
- •Meta takeaway: VC time-to-exit elongation distorts what “success” means
