What it takes to build a top decile Seed Window fund

01 - The gap

A hole in the valuation ladder

Across the funding stack every round's valuation range overlaps the next: Pre-seed ($5m-$10m) into Seed ($10m-$25m), Series A ($40m-$120m) into Series B ($100m-$300m), Series B into Series C ($250m-$600m), and on to IPO. There is one exception. Seed valuations top out around $25m and Series A valuations begin around $40m, leaving a $25m-$40m band that belongs to no named round, which we have defined as the Seed Window.

The Seed Window sits precisely in the valuation gap, of $25m-$40m, a stage the ladder itself leaves unnamed.

02 - Backing execution before it is priced

What the stage is

There is a moment in a deep technology company’s life that rarely makes the headlines. The science works. The prototype runs. The first pilot is live. And then the company faces the long, unglamorous middle, where a working technology has to become a dependable, adopted one, through manufacturing, integration, reliability, regulation and procurement. This is the Seed Window, and it has become longer, more capital intensive and more important than it used to be.

It is tempting to describe the appeal as "the technical risk has passed." The technical risk is declining, but on its own it is not an edge. The science is validated during the execution period between the Seed and Seed Window rounds. By the Seed Window, that technical derisking is reflected in the entry price. Therefore, the Window’s advantage cannot simply be the milestones already achieved and priced in.

The true edge is what the price has not yet absorbed, execution evidence. At the Seed Window the science has been validated, but company-building execution is only beginning. The focus shifts to whether the science works in practice alongside other execution metrics such as integration, scalability, operational readiness, and early commercial validation.

The Seed Window is the point where a company executing is visible.

The harder question is whether a fund built to do that can still generate the power law returns that define a top tier fund. That depends on four variables: entry valuation, initial ownership, dilution after investment, and the share of the portfolio that converts into major outcomes.

03 - The model

The fund size sets the ownership constraints

A $25m fund is deliberately sized around a concentrated portfolio of ten companies. After ~$4m for fees, establishment, and expenses, about $21m is investable, average cheques near $2.1m, with no capital reserved for follow on. That structure sets the first hurdle immediately: to return the $25m committed while investing only $21m, the portfolio must produce about 1.19× gross on invested capital to return the fund.

A $2.1m cheque buys about 10.5% at a $20m post-money, 7% at $30m, and 5.25% at $40m, but only 3.5% at $60m. That maps directly onto the valuation ladder above: the ownership range the fund needs is achievable across the upper end of Seed and the $25m-$40m Seed Window, but begins to fall away as valuations move into Series A territory. Acquiring a meaningful 5-10% consistently is only possible when entry valuations cluster around $20m-$40m. At $60m, a 5% stake costs $3m, and ten of those would need $30m before fees. The strategy is therefore a core of $20m-$40m entries, with selective room up to $60m where conviction, a lower cheque or lower ownership justifies it. The top of the range can be part of the strategy, but it cannot be the average.

04 - Dilution

Initial ownership is not exit ownership

A fund that does not follow on is diluted at every later round. Two things need to be kept apart: round dilution, the share of the company sold in each financing, and ownership retention, the cumulative fraction of an initial stake that survives those rounds. Using the midpoints of the dilution ranges above, roughly 21.5% at Series A, 18.5% at Series B, 16% at Series C and 11.5% at Series D, retention compounds to about 78.5% × 81.5% × 84% ≈ 53.7% after three rounds, and × 88.5% ≈ 47.6% after a fourth. The model therefore uses 47.6% ownership retention as its central case.

The consequence is easy to forget: a fund that buys 7% should not build its returns around owning 7% at exit. Without follow on, 3.33% is the closer number.

05 - The power law

Validation does not make modest outcomes enough

For an initial cheque only investment, the return is governed by one relationship:

At 47.6% retention, a company has to grow about 2.1× just to return the cost of the investment. Beyond that, the leverage is entirely in the company's growth: a 5× company, 2.38×; a 20× company, 9.52×; a 50× company, 13.80×. Initial ownership decides how much is invested and how many independent positions the fund can hold, but the multiple is set by how much the company grows and how much ownership survives dilution.

Technical validation may make the risk more legible. It does not make a modest outcome enough.

A company that triples from entry returns about 1.43× after dilution; one that grows fivefold, about 2.38×. The Seed Window remains a power law strategy, and the best funds still require companies capable of growing 20×, 50× or 100× from the point of investment no matter the stage.

06 - The hurdle

Top decile is a right tail, not a better average

Returning the fund takes about 11.9x aggregate investment multiples across ten $2.1m positions. Top decile is another matter. Carta's 2017 vintage funds, measured at the end of 2025, showed a median of 1.89× net, a top quartile of 2.53×, and a top decile threshold of 4.08× net, attributing that top tier funds own one or two exceptional companies.

That hurdle is roughly five times the aggregate performance needed to return capital. Top decile returns are not produced by lifting the median outcome; they come from a radically stronger right tail.

07 - Why it's a stage

The same cheque buys a different company

The Seed Window round is a distinct stage for a simple reason. The funds on either side do not lead it, and that under competition is what keeps the entry price low.

A Seed fund's $1m cheque or less is built for $20m post-money rounds or less; brought to a $30m Window round it buys 3.3%, not a lead. A Series A lead fund writes much larger cheques for defined traction with rounds priced at 3-5x above the Seed Window. There is room for a Series A investor’s cheque in a ~$5m-$15m Window raise, but it is more likely to be buying optionality than committing with conviction: investing significantly less than its typical cheque and rarely leading the round. Therefore, one is too small to lead, the other too big and too early. A fund built for the Window can lead it, at ~$30m, before the company reprices.

For the capital that could back the company, then, the real choice is when to enter: now, at the Window, or later, at the Series A. For the same company, the two compare like this.

The same company, entered now vs at Series A · multiple shown for a $1.5bn exit

Round	Post-money	Cheque	Exit ownership	Multiple per winner
Seed Windowenter now	$30m	$2.1m	3.3%	~23.8×
Series Await for proof	$90m	$9.0m	6.1%	~10.1×

The Series A fund ends up owning more about 6.1% at exit against the Window's 3.33%, because it enters later, with a bigger cheque and slightly less dilution ahead of it. But it pays three to five times the price, and on the same $1.5bn outcome it makes about 11× where the Window fund makes about 23.8×. Owning more of a company you paid up for is not the same as making more on it: per winner, the Window fund returns roughly 2.3× the multiple.

08 - The portfolio

Three winners, not ten

Concretely, a ten company portfolio that clears the bar might look like this, four total losses, one impaired, two modest successes, and three winners doing almost all the work. The maths can show what the portfolio needs to look like, but it cannot produce the winners: the result still hinges on the manager identifying the small number of companies capable of becoming exceptional.

Illustrative top decile portfolio · 47.6% ownership retention

Companies	Company growth	Investment MOIC	Aggregate
4	0×	0×	0
1	0.5×	0.24×	0.24
2	3×	1.42×	2.86
1	20×	9.52×	9.52
1	30×	14.28×	14.28
1	65×	30.94×	30.94
10	-	-	57.83×

That produces roughly $121m gross and about 4.09× net, just above the threshold. But this is one construction of many. The same bar could also be cleared by a single company producing roughly 55× investment MOIC alongside modest contributions elsewhere in the portfolio; three is illustrative and not required, which is the power law’s real point. In dollar terms here, at a $30m entry the three winners exit near $600m, $900m and $1.95bn and return about $115m between them, leaving the other seven to contribute only ~$6.5m. The fund does not need ten successes, or even three, it needs an exceptional right tail, however few names it comes from. The difficulty is not the maths but finding them.

As a counterexample, a "healthier" portfolio with only two losses, several companies returning 2-3×, and a single 5× produces about 15.6x aggregate multiples, roughly 1.3× gross on committed capital. A better middle protects the downside, but only the right tail creates top decile returns.

09 - Conclusion

What it comes down to

The maths establishes the opportunity. A concentrated Seed Window fund can still produce top decile returns, entering before the major repricing, acquiring enough ownership for individual winners to matter, and does not need the whole portfolio to work. It needs a small number of companies to become exceptional.

That shifts the real question away from portfolio construction and onto selection. By the Seed Window, technical feasibility is no longer enough to distinguish the best companies. Most of the companies worth considering can already demonstrate that the technology works. The harder task is determining which are beginning to show real world execution and the evidence that customers will build around it.

That evidence is different from activity. Pilots, letters of intent, innovation programmes and demonstrations can all create the appearance of momentum without creating dependency. The stronger signals are harder for a customer to produce and will be explored in a future article.

The science tells you an idea can work. The execution signal tells you it can be a company. The Seed Window is the moment it becomes visible.