20VC: David Tisch challenges high ownership and concentration

Summary

• Low ownership can work when outcomes expand and subsequent dilution falls. Tisch contrasts underwriting a $10 billion company with the old $1 billion benchmark, while companies that grow exceptionally quickly may raise later rounds at 3–8% dilution rather than 12–20%. Stebbings admits he underestimated outcome expansion; Tisch’s principle is that “we can’t let our business get in the way of it.”
• Ownership targets can exclude the outliers a concentrated fund most wants. Tisch thinks managers claiming 90% win rates may actually win roughly 20%, offering investments to 120–150 companies to assemble 30 positions. BoxGroup reports a 70–90% win rate with flexible ownership; the distinction is between owning substantial stakes and actually accessing the preferred companies.
• Lead-only seed investors face increasingly effective multistage competitors. Tisch credits their talent, brands and accessibility; Stebbings adds that some offer “treble the price and treble the check.” Waiting for the next round is increasingly dangerous because that first opportunity “might’ve been your only shot.”
• Being well funded does not automatically mean being undisciplined. Tisch calls The Lean Startup “a total fake” for venture-scale companies, while Stebbings worries rapid valuation step-ups distract founders. Their tranche tension is about preference and principle, not an ultimate refusal: Stebbings initially objects that the structure makes him seem less valuable, but both say they would be sad and take it.
• BoxGroup combines broad seed exposure with selective follow-on concentration. It operates two $275 million funds: approximately 150 seed investments and 20–30 follow-on checks of $5–30 million. Stebbings describes Cursor’s $60 billion sale to Elon Musk and X; Tisch confirms the investment returned its fund multiple times.
• Founder execution matters more than immediate consensus or a smooth growth curve. Clay endured four bumpy years before explosive growth; Rogo received 39 rejections from approximately 40 Series A investors before Keith Rabois led the round. Tisch admits those passes caused self-doubt, but argues companies can execute their way from overlooked to consensus.
• Follow-on access and liquidity require different judgments from contractual entitlement and personal de-risking. “The right to invest in subsequent rounds is earned,” Tisch says, although documented pro rata can strengthen founders’ negotiating leverage. BoxGroup generally avoids early secondaries because it believes the best companies compound most at the end.
• BoxGroup protects investment conviction from easy objections. It does not vote, and investors who have not spent time with the founders cannot attack a colleague’s early-stage deal, with ethical and reputational concerns treated separately. “We don’t allow grenades at early stage”: explaining probable failure is easy; finding the reason an exceptional company could work is the job.

Deep dive

1. Bigger outcomes and lower dilution change the seed equation

• Tisch rejects treating today as another 2021: “It’s a totally separate moment.” His early-stage emphasis remains company-specific rather than macroeconomic. AI’s innovation, plateau and renewed-innovation cycles now run in “two to three months,” with model releases, revenue acceleration and acquisitions repeatedly resetting sentiment.
• With $10–20 million seeds increasingly normal, Tisch’s response is blunt: “You can either come on a podcast and complain about it, or you can do the thing.” Investors cannot change competitive market conditions; they must decide whether a particular company warrants participation.
• Two conditions can preserve early-stage economics: larger outcomes and less dilution. A $10 billion outcome now needs to be conceivable where $1 billion once anchored expectations; companies that grow exceptionally quickly can raise subsequent rounds at 3–8% dilution rather than 12–20%. Compounded across rounds, that materially changes how much an initial stake retains.
• These are conditional arguments, not permission to ignore portfolio construction. Tisch says a portfolio entirely comprising $20 million seeds at $200 million pre-money would fundamentally change the business. Nor does every company compound uninterrupted: even companies he admires, including Ramp, have experienced valuation setbacks.

2. Concentration can select for the deals a manager can win

• Tisch accepts venture’s power law but questions the sequence required by concentrated seed investing: identify the true outlier, acquire 10–20%, and defeat the world’s strongest investors for that allocation. “I’ve never understood trying to compete head-to-head” against firms whose brands and founder relationships took decades to build.
• His adverse-selection charge is specific: managers saying “we win 90% of the deals we wanna win” may actually win roughly 20%. To assemble 30 companies at 10% ownership, he thinks they offer investments to upwards of 120, perhaps 150. Those 30 holdings therefore need not be their 30 preferred opportunities.
• BoxGroup’s alternative is flexibility: initial ownership ranges from 1% to 10%, and Tisch declines to invent an average. Small allocations can be disappointing without being economically wrong. Frequently they reflect arriving late, when the investor is “begging for scraps”; the actionable failure is access or decision speed.
• Tisch puts BoxGroup’s win rate at approximately 70–90%, depending on conditions. Stebbings acknowledges having rejected great businesses over ownership requirements and now accepts 1–2% positions. His admission is substantive: “I didn’t see the outcome size expanding,” making his earlier assessment of BoxGroup too rigid.

3. Plenty of capital creates choices, not automatic indiscipline

• Stebbings worries about investors losing a round at 400, then immediately offering another at 800 despite no company progress: excess fundraising could distract founders. Tisch counters that capital, runway and momentum can improve recruitment and reassure customers. Whether those advantages become distraction depends on the founder.
• “I think The Lean Startup was a total fake,” Tisch says. Stebbings offers 37signals as a counterexample; Tisch distinguishes bootstrapped businesses from venture-scale ambitions. His claim is that venture-backed winners ultimately need substantial capital: “The difference between the word, like, lean and the word well-funded is not lack of discipline.”
• Their tranche tension is about preference and principle, not an ultimate refusal. Stebbings initially argues that paying more than another investor makes him seem less valuable; Tisch separates that reaction from the business decision: “I will be sad and take it.” Stebbings ultimately echoes, “I will be sad and take it.” Getting the cheaper tranche requires persuading the founder; failing to persuade them is the investor’s problem.
• Tisch is less forgiving when another VC structures a deal to exclude existing relationships or push collaborators into a more expensive tranche. “I also have a long memory.” Those decisions can influence future recommendations to founders: the business involves recurring dependencies, and short-term allocation tactics can damage long-term relationships.

4. Multistage firms have removed the comfortable waiting strategy

• Tisch does not want a lead-only seed business competing for 10–15% ownership against major platforms. He emphasizes their investing ability and improved accessibility. Stebbings insists price matters too, citing a loss to a more junior investor at a large firm offering three times the valuation and check.
• Previously, firms could wait for the A or B. Now they may never get another opening, prompting immediate, aggressive follow-on rounds with little additional progress. Tisch’s framing: “That might’ve been your only shot to do the deal.” Losing the seed does not guarantee a more attractive entry later.
• Stebbings challenges LP enthusiasm for sub-$100 million, concentrated Silicon Valley seed funds. Tisch describes three misaligned constituencies: founders possess leverage, GPs struggle to maintain static strategies, and LPs want exposures that may be unavailable. Though many LPs say they dislike products at multistage firms, Tisch says most venture allocations still go to established multistage firms for scaled capital and exposure to leading companies. BoxGroup’s answer is explicit: a $275 million seed fund investing in roughly 150 companies without fixed ownership requirements.

5. Cursor’s speed does not make slower compounding obsolete

• Asked about liquid returns, Tisch names Cursor: partner Claire knew Michael through Dorm Room Fund, and BoxGroup invested in its earliest round. Stebbings describes a $60 billion sale to Elon Musk and X after four years; Tisch confirms multiple fund returns and calls it “the biggest venture M&A deal in history.”
• The same fund contains roughly 10 other companies Tisch finds fascinating, although he does not know whether any will match Cursor. Clay took four bumpy years to reach explosive growth; Tisch says BoxGroup was in the seed round of “Base10,” then calls “Baseten” a full outlier after inference became a space. The transcript uses both names, so their relationship is unclear.
• Rogo supplies the sharper example of consensus arriving late. After its seed, Tisch put his reputation behind the AI-for-finance company and introduced it to approximately 40 Series A leads; 39 passed. Keith Rabois led the A; Tisch says Sequoia and Kleiner also invested. What was rejected became a consensus company.
• Asked whether those rejections made him doubt himself, Tisch answers, “Yeah.” He also thinks investors misread Rogo at that particular moment. His conclusion preserves both possibilities: early-stage information is imperfect, and companies outside fashionable categories can keep improving until their execution changes how the market sees them.

6. Founder quality means navigating uncertainty rather than predicting it

• At Harvard and MIT, Tisch asked undergraduates who could envision the world in 10 years: no hands. Five years produced one; two years finally produced some. With predictions spanning AI catastrophe to abundance, he distrusts confident forecasts. The founder requirement is navigating conditions that change quickly and remain deeply uncertain.
• Early signs of exceptional execution still exist: recruiting outlier employees rapidly or accessing customers an observer cannot explain. Those signals do not eliminate future setbacks. Tisch thinks today’s founders are more technical and, in his formulation, smarter; there is “less opportunity for mediocrity to succeed.”
• Stebbings observes fewer sales-led founders. Tisch asks whether go-to-market might become differentiating again in two years; if so, investors should fund those people now. Early-stage businesses may take one to four years to reach broader markets. “You can’t, as an early-stage investor, solve an immediate opportunity or need.”
• On Harry’s, Tisch praises Jeff and Andy while explaining why BoxGroup stopped funding brands. The issue is evidence at seed: resonance, whether customers buy, and customer LTV or CAC-to-LTV data are not yet established. He finds Series A brand investing interesting across categories such as beverages and personal care, but does not consider BoxGroup equipped to invest before momentum.

7. Follow-on concentration is selective, and access must be earned

• Alongside its $275 million seed fund, BoxGroup runs a $275 million follow-on fund making 20–30 investments, almost entirely within its portfolio. Checks range from $5 million to $30 million, generally around the A and B; aggregate investment in one company has reached approximately $40 million. Substantial commitments include Ramp, Clay, Base10, Rogo and Tandem.
• Missed follow-on opportunities have two explanations Tisch keeps separate: founders may not permit more than pro rata, or BoxGroup may misread progress and fail to lean in. “There are so many opportunities” where it reacted too slowly. He declines to single out one company as the definitive mistake.
• “The right to invest in subsequent rounds is earned,” whether documented or not. Fighting a founder who asks an investor to step aside risks losing economically through mechanisms that remove the right, or reputationally through damaged trust. Tisch also dismisses blanket public declarations renouncing pro rata as “marketing garbage.”
• Documentation can nevertheless help founders negotiate. If a new lead demands 22% while existing investors require 6% for pro rata, the founder has grounds to reopen the math. Tisch similarly questions firms claiming to “need” 20% after accepting 5% elsewhere: “need” often means preference, and relationships influence how aggressively it is enforced.

8. Founder liquidity and fund liquidity solve different problems

• “I work for our LPs and my customers are founders,” Tisch says. BoxGroup generally avoids selling early because it joins a founder’s journey, whether four, 15 or 20 years. Secondary sales involve both financial risk and the possibility of damaging a relationship if the founder opposes them.
• Tisch believes the best companies compound most at the end: moving from $2 billion to $4 billion to $8 billion is easier than getting from zero to $1 billion. If conviction remains, selling to de-risk sacrifices that compounding. “We’re in the business of risk,” rather than improving fund optics by trimming winners.
• Founder secondary does not create an equivalent rationale for investor selling. Clearing debt or having $10 million, $20 million or $30 million can reduce personal stress and increase willingness to pursue a larger outcome. Employee tenders can also improve retention; Tisch values portfolio companies that make such liquidity regularly available.

9. Fund four reveals the strategy behind the starter fund

• Tisch thinks venture has “less noise” but more competition: fewer new firms, with inactive managers remaining visible long after meaningful deployment stops. Stebbings pushes back that spin-outs feel more numerous. Tisch’s distinction is between names still on the map and firms actively investing or successfully raising another fund.
• For new managers, his question is “What’s fund four look like?” A $20 million, $40 million or $80 million seed fund may be an entry point toward much larger ambitions. Increasing AUM, partners and organizational scale can fundamentally change stage, ownership and check size; advice should address that destination.
• Stebbings challenges BoxGroup’s consistency claim given its $550 million across two funds. Tisch concedes scale changed while defending the same early-stage craft. He also rejects treating every scaled firm alike: Benchmark, USV, Conviction, Ribbit and Thrive illustrate different ways to build substantial investing businesses.

10. The decision process protects conviction from easy negativity

• BoxGroup has 10 investors and six partners, meeting approximately 50–100 companies weekly. Tisch guesses it sees 50–75% of the opportunities it would want to see. Finding founders early—and becoming someone they take seriously—is the obsession; the strongest referrals come from founders with high-quality taste and networks.
• Winning can improve tactically by understanding the service each founder wants: transactional assistance, companionship or experienced judgment. Picking is harder. Tisch admits he lacks a reliable technique for improving it: after enough repetitions, “your taste is your taste,” and the presence or absence of investment butterflies still matters.
• The firm meets Tuesdays and Thursdays but does not require decisions to wait for committee. “We don’t vote.” Someone must genuinely want the investment; a partner announcing a deal already completed can be perfectly acceptable. Conviction begins with the person who has spent time with the founders.
• “We don’t allow grenades at early stage.” Colleagues without founder exposure cannot derail a deal with generic objections; ethical or reputational information is different. Tisch says 70–90%—“whatever the number”—of early-stage companies do not work, and argues that the failure rate is more applicable at seed than Series A. Explaining why something fails is easy; identifying why it could work is the scarce contribution.

11. YC is differentiated, while the strongest firms keep compounding

• Asked which seed fund he would back, Tisch chooses Y Combinator: “the only differentiated product in venture.” He credits its brand, attraction of talent and business model. Stebbings agrees, emphasizing ownership terms other investors cannot obtain, while Tisch stresses the product itself rather than ownership alone.
• Tisch refuses to select one universal Series A or growth winner, naming multiple strong firms and emphasizing fund size, stage exposure and blended products. Stebbings gives explicit choices: YC at seed, Benchmark at Series A, Sequoia at Series B onward. Tisch challenges the implied discount to Sequoia’s early-stage capability.
• His underlying call is consolidation: the best firms keep attracting exceptional founders and scaled LP capital, reinforcing returns. He estimates perhaps 20–30 competitive names, with around 10 outliers among them. The same “haves and have-nots” dynamic discussed for startups increasingly applies to their investors.

12. Technological progress still has to become social adoption

• Tisch wants autonomous driving adopted faster and calls its safety profile “pretty solved.” Stebbings challenges that with ambiguous cities and difficult weather. Tisch concedes imperfection; his conditional example is whether an 80% reduction in automobile deaths would justify adoption. That is his argument for action, rather than a verified outcome established in the discussion.
• Their closing tension concerns changing institutions: Stebbings thinks building an AI-first school is easier than reforming stale systems; Tisch wants both. His broader hope is practical decision-making and, if abundance arrives, benefits distributed beyond a few. AI’s potential effects on health and science remain a hopeful possibility, not a confident forecast.