David Tisch on BoxGroup's Contrarian VC Strategy

Ask a concentrated venture fund how often it lands the deals it wants, and the answer is almost always the same: north of ninety percent. David Tisch, who has spent seventeen years investing in very young companies and co-founded the seed fund BoxGroup, thinks that number is close to fiction. "Every concentrated manager will say that we win 90% of the deals we wanna win, and I think that's total BS. I think it's, like, 20%," he tells Harry Stebbings on The Twenty Minute VC (16:35).

That gap between the story funds tell about themselves and the arithmetic underneath it is the real subject of this conversation. Tisch explains the mechanism plainly. A concentrated fund usually needs something like 10% to 20% ownership of a company to make its model work (13:00). If that fund wants to end up with 30 portfolio companies it believes in, it has to offer that 10% deal to something like 120 to 150 companies to get the 30 who actually say yes (17:03). The founders who turn a fund down are not randomly distributed. Often they are the most sought-after ones, with the most other offers. So a concentrated portfolio is not a list of the 30 best companies a fund saw. It is a list of the 30 best companies that also agreed to give up double-digit ownership. "I think the best companies, the ones you're most excited about, don't always make it into the portfolio," Tisch says (17:03).

BoxGroup's answer is to stop chasing a fixed ownership number altogether. The firm runs two $275 million funds, a core seed vehicle and a follow-on fund, and expects the seed fund alone to end up invested in roughly 150 companies (31:16). Ten investors, six partners, and checks that range from 1% of a company up to 10%, with everything in between (23:54). The logic: a sliver of a giant outcome beats a large slice of a deal you never got into. BoxGroup was an early investor in the AI coding tool Cursor, which sold to Elon Musk's X for $60 billion, a deal Tisch says returned the fund "multiple times" on its own (31:58).

Tisch is candid that he used to think this was the wrong way to build a firm. He once turned down good companies over ownership concerns and now calls those "terrible decisions" (25:47). What changed his mind was watching outcome sizes expand. A company that might have been written off as a billion-dollar business a decade ago is now the kind of outcome an early-stage investor has to "squint" to imagine reaching $10 billion (10:31). At the same time, the companies that do break out are raising later rounds at far less dilution than they used to, something like 3% to 8% per round instead of the historical 12% to 20% (10:56). Stack a few rounds of lighter dilution on top of a bigger eventual outcome, and a small initial stake compounds into real money. Venture, Tisch argues, "is a pure power law dynamic industry," meaning almost all the returns come from a handful of enormous winners rather than being spread evenly across a portfolio, so getting into the right company matters more than owning a large piece of an average one (12:50).

This is also why Tisch is unsparing about The Lean Startup, the influential 2011 book and methodology that told founders to build minimum viable products, test cheaply, and avoid raising or spending too much before finding product-market fit. "I think the lean startup was a total fake. It was a, like, stupid, like, brainwash that happened for 10 years," he says, challenging anyone to name a company that actually built a venture-scale business by running the lean playbook (18:46). His point is not that discipline is bad. It is that being well-funded and being undisciplined are not the same thing, and an entire generation of founders was told to stay small when the companies that actually mattered got big by spending aggressively.

Begging for Scraps, Gracefully

Being a price-taker rather than a price-setter is, by Tisch's own account, often uncomfortable. He describes founders tranching their rounds, taking an initial check at one price and a later slice at double the valuation once bigger names have validated the deal, leaving earlier investors like BoxGroup negotiating for the cheaper tranche after the fact. "I would like the lower tranche please," he says, half-joking about the posture this forces on him (21:07). Pressed on whether he resents it, he is unambiguous about where the resentment actually lands: not on the founder making a rational choice, but on the other investor who restructured the deal to squeeze him out. "I also have a long memory," he notes (21:37).

That same unsentimental read extends to who gets to stay in the lead-investor business at all. Tisch argues that pure-play, lead-only seed funds, the kind many limited partners say they want exposure to, are a dying category, because they cannot compete with multi-stage firms that will write a bigger check at a higher price and still want the deal (26:50). Firms once content to wait and invest at the Series A or B now have to fight for entry wherever they can get it, because waiting means losing the company to someone else entirely (28:05).

BoxGroup's self-reported win rate on deals it actually bids for is 70% to 90% (24:39), a number Tisch attributes mostly to speed: the single biggest reason the firm loses a deal is getting to the founder too late, not losing a straight fight on terms (24:39). The clearest illustration of how misread early-stage judgment can be is Rogo, an AI company built for finance. After BoxGroup's seed investment, founder Gabe took the company out to raise a Series A from roughly 40 lead investors. Thirty-nine said no. Keith Rabois led the round that one firm agreed to do, and Rogo has since been funded by Sequoia and Kleiner Perkins as what Tisch calls "a consensus company" (42:20). The lesson he draws is not that the 39 were foolish, but that early-stage investing is not a market with perfect information, and conviction has to survive being wrong in public, repeatedly, before it pays off.

Earning the Right to Follow On

On pro-rata rights, the contractual right to invest more in a company's future rounds, Tisch rejects both extremes common in venture Twitter debates. He does not think pro-rata is an automatic entitlement, and he does not think funds that loudly renounce it are being principled either. "I believe the right to invest in subsequent rounds is earned. Whether you have it in the documents or not, it's earned," he says (46:05). Having the right documented mainly helps the founder, he explains, because it gives them leverage to push back on a new investor demanding an outsized allocation in the next round. But enforcing that right against a founder's wishes is close to reputational suicide: a fund that fights a founder over pro-rata will likely lose anyway, and will certainly lose the relationship.

A similar relationship-first logic shapes how Tisch thinks about founders selling stock early, known as secondary sales. He argues these sales can raise a founder's appetite for risk rather than signal doubt, because removing personal financial pressure, debt, housing stress, the need for a safety net, frees a founder to keep swinging for a bigger outcome rather than selling the company early out of fear (51:48). He draws a sharp line between that and a fund trimming its own position to make its numbers look better to its own investors, which he sees as a fundamentally different, less defensible motive (52:21).

Inside BoxGroup's own decision-making, Tisch has built a culture meant to protect a conviction once it exists. Partners meet 50 to 100 founders a week as a firm (61:00) and decide twice weekly, but a partner who has not personally spent time with a founder is not allowed to criticize that deal in the room. "We don't allow grenades at early stage. You're not allowed to throw a grenade on someone else's deal," he says (62:23). His reasoning is almost mathematically blunt: since the large majority of early-stage companies fail, anyone can construct a plausible case for why a given company will not work. That exercise is trivial and, in his view, useless. The actual craft, the one BoxGroup is built around, is the harder and rarer skill of looking at an unproven founder and unproven market and finding the reason it might succeed. "The magic of our business is squinting and seeing why something is going to work," he says (63:12). It is a modest-sounding claim for a firm that was early into a company that sold for $60 billion, but it is also, on the evidence of this conversation, the whole argument: venture rewards the investor willing to say yes before the proof exists, not the one with the tidiest portfolio math.