The Twenty Minute VCJake Saper, GP @ Emergence Capital: "We Sold Salesforce Early and Lost Out on Billions"
CHAPTERS
- 0:00 – 1:16
Emergence Capital’s track record and Harry & Jake’s London warm-up
Jake opens with Emergence’s long-term performance stats—capital deployed, cash returned, and how often their early bets “graduate” to follow-on rounds, unicorn valuations, and IPOs. Harry and Jake banter about their London walk, setting an informal tone before diving into investing stories.
- •Emergence deployment vs. cash returned and what that implies about performance
- •“Graduation metrics”: follow-on rounds, billion-dollar rounds, and IPO rates
- •Harry’s preference for real context over formal prep calls
- •Quick rapport-building before the Zoom deep dive
- 1:16 – 7:53
The Zoom investment: thesis, early signals, and the churn-modeling twist
Jake recounts how Emergence’s earlier work on video conferencing primed them for Zoom and why the deal was unusually concentrated and expensive for 2014. The diligence nearly derailed over churn math—until the team discovered the founder had been calculating churn incorrectly, making the business look worse than it was.
- •Emergence thesis: replacing a tired Webex; prior diligence on competitor Fuze
- •Zoom’s early PLG traction, profitability, and Eric Yuan’s domain expertise
- •Deal risk: $20M check from a $250M fund at ~$200M post on ~$2M revenue
- •Churn discrepancy discovery and the integrity decision to inform the founder
- •Layering enterprise sales on top of PLG (hiring sales leadership to scale)
- 7:53 – 18:53
Prepared minds vs. venture “packaging,” plus power-law fund outcomes
The conversation expands from Zoom into how theses actually form and when “prepared mind” helps versus misleads. Jake uses Chorus.ai, SalesLoft, and other examples to explain that even strong multiples may not move a large fund, and why power-law dynamics still dominate returns.
- •Prepared mind can accelerate insight, but doesn’t explain all returns
- •Portfolio learnings creating new theses (e.g., voice AI via Chorus.ai)
- •Fund math: why a 5–10x exit can still be needle-marginal in bigger funds
- •Winner-take-all vs. multi-winner markets (Doximity vs. tooling ecosystems)
- •Private equity outcomes: often modest for VCs, with rare exceptional cases
- 18:53 – 23:07
Market pull as the #1 variable—and how to detect it in diligence
Jake argues “market pull” is the most important ingredient: customers must be desperate, not merely interested. He shares the kinds of user statements that signal true pull and explains why breakout growth doesn’t always equal an enduring company (e.g., Zenefits vs. Gusto).
- •Market pull definition: urgency, DIY attempts, painful workarounds, time spent
- •Diligence signals: users willing to pay personally or threaten to quit without it
- •Breakout growth ≠ durability; early hype can hide fragility
- •Examples contrasting endurance vs. flash growth (Gusto vs. Zenefits)
- •AI-era pull is amplified by top-down “go buy AI” mandates
- 23:07 – 27:21
Founder vs. market vs. traction: defensibility, pricing, and “what you have to believe”
Jake ranks market pull first, founder second, and traction third—then explains why the founder’s job is converting pull into defensibility. He also outlines Emergence’s “What You Have To Believe” framework to pressure-test dilution, competition, and the path to fund-returning outcomes.
- •Ranking: market pull > founder > traction (and why)
- •Defensibility hypotheses (e.g., unique technical moats like Bolt’s Web Container)
- •“The best are often expensive”—with exceptions when non-consensus
- •“What You Have To Believe”: 3–5 deal-specific assumptions + supporting/negating data
- •Retention risk as the missing variable for many fast-growing AI startups
- 27:21 – 39:10
Emergence’s collaborative diligence: priority deals, reference calls, and speed under time pressure
Jake explains Emergence’s high-conviction, low-volume model—roughly one investment per partner per year—and how that enables a “whole partnership” diligence approach. He details their “priority deal” process, why every partner does reference calls, and how they compress time by parallelizing work (including creative COVID-era on-sites).
- •Focus strategy: B2B-only, few deals, high conviction, heavy partner involvement
- •“Priority deal” trigger that reallocates the entire partnership’s time
- •Why partner-led reference calls matter (tone, pauses, nuance)
- •Operating under time compression via parallel diligence and strong internal trust
- •COVID story: meeting a founder in a field near Denver airport to win Regal.ai
- 39:10 – 47:40
AI changes the growth bar: from triple-triple-double-double to retention-first benchmarks
They discuss how AI-era distribution and buyer urgency are redefining what “great” growth looks like (with examples of extreme ramp). Jake proposes replacing legacy SaaS heuristics with a growth-plus-retention benchmark and explains why many AI companies’ cohorts may disappoint once the initial hype fades.
- •Growth expectations reset upward in the AI era (examples: Together.AI’s surge)
- •New heuristic: fast growth plus strong net dollar retention (e.g., 120%+)
- •Retention is the unknown: many companies are too young to prove durability
- •Gross margin concerns: closed-source price competition + open-source escape valves
- •Together.AI thesis: open-source LLMs becoming a dominant market component
- 47:40 – 49:49
Where value accrues in AI: workflow stickiness, brands, and domain-specific “coaching networks”
Jake argues enduring value comes from daily-use workflows and products that become muscle memory—similar to why Salesforce remains entrenched even if not “best.” He also discusses specialization via open-source models and Emergence’s “coaching networks” concept: AI embedded as a domain-aware assistant that improves via feedback loops.
- •Workflow stickiness as a moat (why rip-and-replace is hard)
- •Brand and habitual usage as underappreciated AI defenses
- •Specialized LLMs likely proliferate—especially via open source
- •“Coaching networks”: AI copilots that learn from outcomes and improve recommendations
- •Domain data and feedback loops create differentiation even vs. frontier models
- 49:49 – 56:17
Why AI won’t replace vertical SaaS: buy vs. build, maintenance, accountability, and pricing shifts
Responding to claims that companies will “build all their tools,” Jake gives three reasons vendors persist: opinionated solutions, ongoing maintenance, and the buyer’s need for accountability. The discussion then moves into how pricing may evolve from seats to usage to true outcomes—and why outcomes-based pricing is hard when humans remain in the loop.
- •Vendors sell an opinionated solution, not just code
- •Software gets easier to create but harder to maintain amid rapid change
- •Enterprises want a “throat to choke” (support, uptime, guarantees)
- •Pricing spectrum: per-seat → usage → outcomes; outcomes is directionally right but messy
- •AI-enabled services make outcomes pricing easier (e.g., end-to-end delivery models)
- 56:17 – 1:03:40
Incumbents vs. startups—and Jake’s ‘FTX moment’ worry for AI agents
Jake explains how he’s become less convinced incumbents will capture most AI value, emphasizing startups’ focus advantage. He shares a bullish view on Salesforce’s entrenched workflow position, a bearish view on IBM due to mainframe displacement, and warns that AI agents could trigger backlash if deployed without guardrails.
- •Focus beats distribution in many early AI categories (startups running faster)
- •Salesforce bull case: massive daily workflow entrenchment
- •IBM short thesis: AI enabling COBOL/mainframe migration to the cloud
- •Adoption realism: likely a trough of disillusionment after experimentation
- •“FTX moment” risk: agents taking harmful actions could slow enterprise rollout
- 1:03:40 – 1:09:47
Why Emergence hasn’t lost partners: internal growth, incentive design, and avoiding orphaned companies
Harry probes partner departures across venture; Jake attributes Emergence’s stability to promoting from within and creating real paths to equal partnership. He explains how the industry’s common incentive structure produces churn and “orphaned” portfolio companies—and why signaling risk is real when multi-stage firms treat seed as an option.
- •Common VC hiring model creates check-writing incentives and short time horizons
- •Unequal partnerships and retained founder carry drive top talent to spin out
- •Founder harm: ‘orphaned deals’ lose internal champions and pro-rata support
- •Signaling risk: multi-stage seed “options” that don’t double down
- •Emergence approach: partner-led seed with meaningful ownership and commitment
- 1:09:47 – 1:15:19
Selling too early: Salesforce lesson, managing public positions, and LP expectations
Jake discusses how exit decisions can dwarf entry decisions, citing Emergence selling Salesforce soon after IPO and missing enormous upside. He then outlines how the firm manages public positions with board-level information, quarterly reviews, and a comparison of different selling strategies (lockup vs. active management vs. peak).
- •Salesforce: an iconic example of selling far too early post-IPO
- •Public stock management tied to board access and information advantage
- •Quarterly sponsor-led reviews within legal trading windows
- •Counterfactual analysis: lockup sales vs. actual vs. peak-price outcomes
- •LP dynamics: charities/endowments often prefer long-term value maximization over quick liquidity
- 1:15:19 – 1:20:13
Reserves, ‘mirage PMF,’ bridge rounds, and when IPOs might return
The discussion turns to follow-on strategy: why reserves matter, how bias creeps in, and how Emergence sanity-checks inside rounds by re-engaging the full partnership. Jake introduces “mirage product-market fit” (growth masking misalignment or weak unit economics) and gives examples where bridges saved a company—and where they likely don’t.
- •Reserves can meaningfully impact outcomes, but late follow-ons can be costly mistakes
- •Bias risk in follow-ons; full-partnership reviews for tougher ‘save’ scenarios
- •‘Mirage PMF’: growth that hides fragmented use cases or poor economics
- •Bridge rounds: usually risky, occasionally lifesaving (example: Intacct)
- •IPO timing guess: likely next year due to macro/political uncertainty
- 1:20:13 – 1:33:55
Quickfire: contrarian beliefs, deal tactics, and a human-centric AI future
In quickfire, Jake shares personal contrarian views (like teaching anyone to sing), public market picks, and stories about winning and losing competitive deals. He closes with a reflection inspired by Sam Altman: as AI accelerates, the most durable advantage may be understanding people—how they think, feel, and are persuaded.
- •Contrarian belief: anyone can learn to sing (and why Jake believes it)
- •If holding one stock for 10 years: Microsoft as ‘index’ on B2B software
- •Competitive deal stories: Ironclad loss and the Assembled mock board meeting win
- •Carry structure: retiring founders forfeiting carry to enable true equal partnership
- •AI future: less rote work, more human connection, influence, and creativity
