← Back to list
Colossus (Invest Like the Best / Business Breakdowns)Podcast2 May 2023Source: joincolossus.comHost: Patrick O'Shaughnessy

Bessemer Venture Partners - Building a VC Firm that Lasts Centuries - [Invest Like the Best, EP.327]

In plain words

This piece explains how century-old VC firm Bessemer thrives with a 'no boss' model where partners vote on deals independently. They believe vertical market software (like restaurant or legal software) still has room to grow, and AI will speed up improvements but not disrupt the industry. Key holdings: Shopify (raised $100M but only burned $2M before IPO), Toast (cheaper Android tablets replacing expensive legacy systems, plus payment bundling), and Procore (first internet software for construction, founded by an industry insider).

AI SummaryAI-generated · may contain errors · verify against the original

At a Glance Three partners at Bessemer Venture Partners discuss the century-old VC firm’s unique operating model: through high partner autonomy and a commission-based incentive system, each partner can independently determine their investment style (e.g., Jeremy Levine focuses on vertical industry s

~11 min full read · 9 sections
Deep Analysis

Bessemer Venture Partners - Building a VC Firm that Lasts Centuries

At a Glance

Three Bessemer Venture Partners partners (Jeremy Levine, Kent Bennett, Brian Feinstein) detail the unique operating model of this century-old VC firm: a structure of high partner autonomy, commission-based incentives, and a "boss-less" setup without a founder at the helm, enabling Bessemer to consistently deliver strong investment performance across generations. Core thesis: vertical market software (VMS) remains in the "7 PM" rather than "11 PM" stage of its lifecycle, and AI large language models will accelerate rather than disrupt the existing software landscape.


Theme 1: The "No-Boss" Operating Model of a Century-Old VC — Individual Accountability + Collective Support

Jeremy Levine argues that Bessemer's core competitive advantage lies in "nobody owns it, and nobody manages it" (referring to the investment decision-making process).

  • Historical Origins: Bessemer traces its roots to 1907. After Andrew Carnegie, Henry Phipps, and Henry Frick founded Carnegie Steel, Phipps allocated part of his wealth to a family trust, investing in "high-risk, high-growth" opportunities. Initially, it invested in industrial companies such as Ingersoll Rand and International Paper. In the 1970s, it shifted toward the technology sector and formally established Bessemer Venture Partners.
  • Decision-Making Mechanism: Every partner is a member of the investment committee and holds equal voting rights. Voting uses a 1–10 scale, with an average score of ≥5.5 required for approval. In 22 years, only one investment has been voted down. Partners make decisions independently, documenting who made the call and who takes credit or blame.
  • Incentive Structure: In their early years, junior partners are primarily compensated based on their own investment performance. Over time, they accumulate equity and become "true co-owners." Approximately three-quarters of Bessemer's partners started as analysts, taking 5–9 years before gaining access to a checkbook.
  • Conflict Resolution: All key meetings include an independent psychologist/coach with no conflicts of interest. The core value is "intellectual honesty" — investment memos are not meant to persuade others but to faithfully record one's own judgment.

> Kent Bennett adds: "We constantly and freely rethink the entire model because nobody owns it. There is no founder who makes you feel like you 'can't offend the person in the corner office.'"


Theme 2: The Prism of Investment Style — Product Advantage, Capital Efficiency, and the Path to Market Leadership

The three partners each described distinctly different investment tastes, yet all agreed they represent "three sides of the same coin."

Partner Core Preference Key Judgment Logic Representative Case
Kent Bennett Absolute Product Advantage "Every failed investment lacked a clear product advantage" — verifiable through 10 customer calls Toast (10x better + 10x cheaper)
Jeremy Levine Capital Efficiency "I like investing in companies that don't need my money" — looks for early-stage "scrappiness" and a frugal culture Shopify (burned only $2M before IPO)
Brian Feinstein Path to Market Leadership "In vertical markets, the first and second players take all the treasure" — prefers greenfield opportunities Procore (the first internet presence in the construction industry)
  • Jeremy Levine's "Capital Efficiency" Paradox: The most successful investment, Pinterest (Series A at $8-9M, eventually raising $1B), was actually a counterexample to capital efficiency, illustrating the need to maintain a flexible mindset.
  • Brian Feinstein's "Path" Framework: Greenfield (ideal) > Replacement Cycle (requires 10x better or 10x cheaper) > Service Digitization (new AI opportunity).

Theme 3: Vertical Market Software (VMS) — Lifecycle at "7 PM," AI Opens the Third Wave

Kent Bennett believes VMS is still at "7 PM" rather than "11 PM" in its lifecycle, with AI opening a new window.

  • Historical trajectory: VMS has undergone three waves — pure SaaS (MindBody) → embedded payments (Toast) → embedded finance/transactions/payroll (Procore, Clio, etc.). Bessemer currently has 70+ VMS investments, completed by nine different partners.
  • Market opportunity: Brian Feinstein divides the opportunity into three buckets:

1. Greenfield: Increasingly scarce, but larger businesses can be built in smaller markets through higher take rates (from 0.5% to 2%)

2. Replacement cycles: "Tens of billions of dollars in ARR" in PE/Constellation/legacy software await replacement by 10x products

3. Service digitization: AI will convert "tens of billions of dollars in service spending" into software — e.g., Even Up uses LLMs to automatically generate legal demand letters

  • AI's impact on VMS: Kent Bennett views AI as a "fully commoditized tool" that will not create network effects but will accelerate product adoption and customer onboarding. Dozens of Bessemer portfolio companies are already developing AI features, such as natural language database queries.

> Jeremy Levine cautions: "AI feels like a platform-level opportunity, but VR and blockchain initially felt the same. Nine months is too early to conclude, five years to gain confidence, and ten years to confirm."


Theme 4: The AI Era — Excitement Meets Caution, with Defensiveness Rooted in Classic Software Strengths

Kent Bennett, the team's "AI bull," argues that LLMs will accelerate everything, but long-term defensiveness still hinges on classic software fundamentals.

  • Excitement:
  • Speed far surpasses the mobile era: The iPhone launched in 2007, but Pinterest did not have an app until 2012 (five years later); in contrast, AI features can go from idea to prototype within a week
  • Recommendation for companies: Find a "20-year-old hacker" to experiment over a weekend, and host internal hackathons (once a month, as the technology changes weekly)
  • Concerns:
  • Threat to "complacent" incumbent software companies — "anyone not paying attention should be afraid"
  • New entrants can disrupt the existing landscape with "better, more efficient mousetraps"
  • Defensiveness assessment: Jeremy Levine believes AI will not change the nature of defensiveness — "speed to market, owning the market, and building the best product and team" are the true moats. Proprietary data may be useful, but as large models grow bigger, it may no longer be necessary.

Theme 5: Partner Exit Mechanism — No Fixed Tenure, Annual Evaluation

Brian Feinstein notes that one of Bessemer’s healthiest traits is "no tenure" — current performance is assessed annually, and results from 10 years ago are irrelevant.

  • Exit culture: Over 20 years, approximately 50–70 professionals have left, with the vast majority succeeding in the VC industry (Sarah Tavel → Benchmark, Christina/Chris → a16z, etc.). Bessemer exports more talent to the broader VC industry than all other firms combined.
  • Operational approach: Extremely cautious and low-key, emphasizing "not a fit" rather than "not good enough" — Bessemer’s individual decision-making model is not suitable for everyone.
  • Jeremy Levine’s "legacy" story: A former subordinate, Sarah Tavel (now at Benchmark), once published a blog post thanking him without his consent, prompting him to resolve to "find 10 more people younger than me to help."

Mentioned Positions

Position Guest Sentiment Key Data
Shopify Bullish (Jeremy) Raised $100M, burned only $2M; $73M still on balance sheet at IPO
Toast Bullish (Kent/Brian) 10x better + 10x cheaper (Android tablet vs $20K legacy system); bundled payments
Procore Bullish (Brian) Founder Touhy came from the industry, not Silicon Valley; started in Santa Barbara
Pinterest Bullish (Jeremy) Series A $8-9M, ultimately raised $1B; invested in 2011, iPhone App launched in 2012
MindBody Bullish (Brian) Initially invested at $12M ARR, 2.5-3x ARR multiple; discovered payment bundling opportunity
Even Up Bullish (Brian) Uses LLM to auto-generate legal demand letters (previously took weeks of manual work)
Clio Bullish (Bessemer) Member of vertical software investment portfolio
X-Time Bullish (Bessemer) Member of vertical software investment portfolio
Yelp Bullish (Jeremy) User-generated content investment thesis
LinkedIn Bullish (Jeremy) User-generated content investment thesis

Judgments Worth Remembering

1. Jeremy Levine: "In 22 years, I've only seen one investment vetoed" — Bessemer's voting mechanism (score 1-10, average ≥5.5 passes) gives partners near-complete autonomy in decision-making, embodying the core principle of "individual responsibility + collective support."

2. Kent Bennett: "Our best investment roadmaps come from investments that turned out 'much better than expected'" — Bessemer's roadmap is not derived top-down but emerges from "accidental discoveries" of trends within the portfolio. For example, MindBody's payment bundling inspired nine partners to invest in 70+ VMS companies.

3. Brian Feinstein: "A Centaur ($100M ARR company) is more meaningful than a unicorn" — because "we can't recall a single $100M ARR software company that went under." This is a true measure of industry health, not valuation bubbles.

4. Jeremy Levine: "The secret to capital efficiency is investing in companies that don't need your money" — Shopify raised $100M but burned only $2M; Pinterest eventually raised $1B but achieved user growth at near-zero cost — "The media loves to write about how much money was raised, but the real question is, 'Relative to the amount raised, how big a business did you build?'"

5. Kent Bennett: "AI features can go from idea to prototype in a weekend — that's 5 years faster than the mobile era" — Pinterest launched its app 5 years after the iPhone debuted, while companies in Bessemer's portfolio "built a natural language query database feature just last week."

6. Brian Feinstein: "The third wave of VMS is service digitization — turning billions of dollars in service spending into software" — e.g., Even Up uses LLMs to automatically generate legal demand letters, a process that previously required weeks of manual labor.

7. Jeremy Levine: "AI's defensibility comes from classic software advantages — speed to market, owning the market, the best product, and the best team" — AI is a "fully commoditized tool" that does not create network effects, and proprietary data may not be a long-term moat.

8. Kent Bennett: "Bessemer 'exports' more talent to the entire VC industry than the rest of the industry combined" — Sarah Tavel (Benchmark), Christina/Chris (a16z), and others all worked at Bessemer before leaving, making Bessemer a "no-risk choice" for attracting top talent.