Semil Shah on Fund Size, LP Relationships and the Discipline of Seed Investing

Harry Stebbings welcomes Haystack founder Semil Shah back to 20VC. Shah says his motivation includes a fear of being unable to participate in the Bay Area ecosystem as he wants. He also wants to become an early, useful partner to founders building important companies.

Asked how he sets a fund size, Shah says Haystack’s first four funds fell short of their fundraising targets. The next two were each $50 million: the first was intended to enable meaningful seed ownership, and for the second he repeated a model he felt was working. He expected modest growth in later funds but wanted to stay below $100 million.

Seed-round labels can be confusing, Shah says. He observes that entry prices have risen and agrees that a $3 million round at a $15 million valuation felt like a common entry point at the time. His concern is the widening gap between an expensive seed and the evidence needed to raise a Series A.

Haystack had reached roughly 7% initial ownership in its two $50 million funds, according to Shah. A larger fund seeking 10% of every company would require more capital and potentially force the firm to lead every round. He prefers flexibility because eventual dilution makes exit ownership important to fund economics.

Stebbings says some fashionable seed rounds led by prominent former product executives in his portfolio moved slowly. Shah says Haystack usually avoids many such expensive rounds, although they can make sense for larger funds. A strong résumé and an ability to build a company are related but distinct qualities in his view.

Shah looks for both talent and entrepreneurial drive. In founder conversations he asks about experiences that demonstrate persistence, overcoming adversity or commitment; the experience need not come from business. He stresses that this is a personal and potentially mistaken philosophy, not a scientifically validated test.

Stebbings observes that highly competitive Series A rounds more often seem to lead to competitive B rounds than hot seeds do. Shah broadly agrees and suggests later investors may follow the validation of a prestigious A. They present this as an observation about the market at the time, not a measured causal result.

Shah resists simply doubling assets under management because he has seen how dilution and time affect returns. A much bigger fund might need larger initial stakes and a different promise to founders. He prefers to lead some rounds, co-lead others and participate as a substantial third investor when that best fits the company.

Over the previous three years Haystack had led approximately ten deals, Shah estimates. Stebbings worries that a willingness to be a passenger could let competitors take the lead. Shah replies that founders should choose partners best suited to them, and that insisting on leading everything could reduce the quality of Haystack’s portfolio.

Shah describes introducing a founder to sector specialists even while offering his own term sheet. He believes speed and conviction matter when Haystack does lead, but he also values repeated collaboration with investors he trusts. A substantially larger fund, he says, would change the collaboration and portfolio model.

Haystack generally aims for 5% to 10% ownership at entry, with about 5% as a practical lower bound, and hopes to earn the opportunity to maintain its stake through later rounds. Stebbings notes that this can be difficult in Europe when a lead investor wants most of the available allocation. Shah says his firm often invests before the larger institutional round.

Shah had previously discussed a three-year fund deployment period. He tells LPs that 24 to 30 months is a more realistic commitment for his seed strategy, with recent funds closer to 24 months. He worries that deploying substantially faster makes it difficult to return to LPs without a long relationship or established results.

Looking back at Haystack’s fourth fund, invested from 2017 to 2019, Shah says the push for more ownership came with selection trade-offs. He estimates that most mistakes were poor selection rather than adverse access, but his percentages are retrospective judgment, not a formal measurement. He says subsequent funds corrected the approach.

He points to several positions in the fourth fund that might still produce strong outcomes, without assigning a final return. At seed, he regards losses as part of the activity. He sees access and selection as equally important: investors must meet creative founders and choose well among the opportunities available to them.

Turning to fundraising, Shah says new managers can use small investments or syndicates to develop examples before approaching institutional LPs. Stebbings notes that early logos cannot establish a true long-term investing record. Shah agrees but says even a short history gives LPs a more substantive conversation about decisions, follow-on investors and capital deployed.

Endorsements from other general partners carry more weight when those people have actually worked with the manager. Stebbings describes asking each early backer for introductions to people with similar interests. Shah notes that discovery is only the start: institutional LPs may wait for a second fund to see whether a manager raises and deploys the first successfully.

Both stress the value of building relationships before a fundraising campaign. Shah maintains a one-page overview for each fund along with a deck and data room, aiming to reduce friction. Stebbings questions the value of decks; Shah thinks a deck can serve as a record, while the pitch itself should rest on a stronger body of work.

Shah says many emerging managers do not explain clearly where they are going and why. Stebbings adds that past deal exposure does not automatically establish an ability to execute a different seed strategy. Shah advises building a reputation through useful work and good investments so that LPs hear about the manager from trusted people.

For LPs who linger without deciding, Shah suggests a respectful deadline and a willingness to accept a no. Rather than waiting indefinitely, the manager can say that without a response by a specified date they will assume the opportunity is not a fit and continue the process.

Shah has not offered preferential economics for an early close and worries that such terms can set a difficult precedent. He acknowledges a first-time manager may need to make concessions to get started, but suggests limiting them to the relevant fund instead of letting them carry indefinitely into future vehicles.

An anchor investor might receive some carry in exchange for a substantial commitment and introductions, Shah says, depending on the investor and terms. He distinguishes sharing carry for one fund from selling a permanent interest in the management company. Other LPs may seek similar terms, so any exception must be handled clearly.

The pair discuss escalating carry at high fund multiples. Shah says Haystack had not used those kickers and he did not know the topic well, though he was considering them. He had generally followed a conventional two-and-twenty structure, citing Union Square Ventures as an operational example he admired.

Shah recalls an impressive introduction to a university endowment that still did not lead to a meeting, despite his record and prior contact. He took the quick rejection as an early warning that his fourth fundraising campaign would be difficult. Stebbings is surprised; Shah appreciates that the person responded promptly.

His larger lesson was not to assume every LP will support a manager in the same way after committing. He says he chose not to accept two existing backers into later funds because of behavior he considered incompatible with how he works. He emphasizes that these anecdotes do not describe most LPs.

Asked whether LPs really stay for three funds, Shah says he does not know. He expects more turnover because institutions have different liquidity needs and varying visibility into their existing portfolios. His expectation of tighter commitments and stranded capital is a forecast from the conversation’s market environment, not a current finding.

Stebbings worries that LP employees may favor famous firms because the career risk of backing a small manager is higher. Shah says his experience differs: many allocators had used a barbell approach, and he expected some to shift toward smaller vehicles as concerns about very large funds grew. Both are expressing views, not verified allocation trends.

They note that new sources of capital were showing interest in venture. Shah considers long-established endowments and foundations among the stickiest partners, but acknowledges that individual institutions will respond differently in a downturn. He then speculates about opportunities for firms that can combine early access with follow-on capital.

Stebbings compares the likely winners to boutique luxury brands and large retailers with deep capital, worrying most about undifferentiated firms in between. Shah cautions that a single exceptional investment can change a fund’s trajectory. Fundraising conditions may be discussed, but predicting which investment will transform a franchise is much harder.

They discuss firms whose reputations and realized results may not move in step. Shah cites longstanding LPs who he says were pleased with Mayfield, and mentions Crosslink’s historical ownership in Chime as an example of a less discussed investor finding a major position. These are their accounts rather than independently verified return figures.

Shah believes some LPs erred by accepting expanding fund sizes and broad allocation discretion within a single pool. Separate vehicles for distinct activities, he argues, can provide better discipline. Stebbings responds that LPs often face a difficult choice between accepting a larger fund and leaving a valued franchise altogether.

Shah says he heard that some long-standing institutional investors had reduced exposure to large funds and shifted earlier. Stebbings criticizes rigid allocation buckets that can push LPs to select an available fund merely to fill a category. Shah agrees the approach can be limiting but notes that allocators outside investment hubs must still present a coherent coverage plan.

Shah suggests that recognizable venture firms often appear together on important technology-company capitalization tables, though he acknowledges the pattern might change. He believes a network of experienced firms can help companies develop toward large outcomes. This is his interpretation of the market, not evidence that a specific investor guarantees an exit.

Asked to design a hypothetical endowment allocation, Shah would spend extended time in major startup hubs, build relationships, seek references and recruit direct-investment expertise. He would also try to join a small number of established venture franchises. Stebbings says access is difficult and would instead consider anchoring promising investors leaving larger firms.

Shah identifies another excess of the boom: investors lowered the discipline of using each financing as a real milestone. He says some companies raised large cash balances without product-market fit. Stebbings recounts investors proposing to return money in exchange for founder liquidity; Shah sees such arrangements as a possible way to address stalled situations.

He also believes some founders failed to set their own milestones or keep investors informed. Regular updates, he says, help investors understand a company well enough to advocate for it. His broader principle is to treat other people’s capital with respect and offer transparent information, a responsibility he applies to both founders and fund managers.

In the quick-fire round Shah chooses an article about a Nintendo game creator whose childhood experiences outdoors influenced his games. He finds it striking that digital worlds can draw inspiration from nature. He says he enjoys gaming himself, mostly casually.

His biggest contemporary concern is the combination of tighter money and multiple risks. He expects a prolonged adjustment, but treats that as his outlook at the time. Haystack was then conducting a third-party audit and revaluing its portfolio; he says there is no universal markdown percentage and methodologies should reflect the holdings.

He names several angel investors he admires, including Kyle Parrish, Guy Podjarny and Charlie Songhurst, and mentions Scott Belsky and Tony Fadell. Stebbings notes that Songhurst makes many investments, which also affects diversification. These are the speakers’ assessments, not a ranking based on comparative results.

Shah says the hardest part of building Haystack was almost continuous fundraising across its first four funds; he took only about six months off from fundraising during that period. Even after the fifth fund made the process easier, he says he remained anxious about raising capital and deploying it well.

He identifies HashiCorp as his largest returner and estimates that a $25,000 initial investment produced roughly $30 million to $35 million when shares were distributed around the end of the lockup. He estimates the $1 million first fund was marked at roughly 30 to 40 times capital at the time. These are his approximate historical figures, with different realized and marked components.

His biggest miss, he says, was OpenSea. Nick Tomaino repeatedly shared the opportunity, but Shah did not understand the product and passed despite seeing signs of use. He says its ultimate outcome was uncertain and describes the lesson as paying closer attention to what customers actually use, rather than relying solely on his initial understanding.

Shah credits investor Paul Martino with simple advice: do good deals. He shows poker chips made to mark shares received through acquisitions, illustrating how unpredictable investment outcomes can be. These are personal reminders rather than a systematic performance analysis.

His most recently announced investment at the time was Impart Security, founded by a team with experience at Signal Sciences. Haystack liked the team’s product skills and approach to collaboration among security teams, though Shah says the company could change direction. He has a direct investment interest in the company.

Stebbings thanks Shah for the conversation and their friendship, and Shah returns the thanks. The extended exchange is a complete interview rather than a collection of unrelated clips.