Every fund tells its LPs it is concentrated. Then you look at the portfolio page and there are fifty-eight logos. The word has stopped meaning anything in Indian venture, so we should say what we actually mean: a concentrated portfolio is one where every company can receive the senior attention of a partner for the duration of the engagement. That number, for an operator-led fund, is small.
The arithmetic of attention
Start with a partner's working week. Forty hours of deep work is a high estimate for someone who is also raising the next fund, picking up calls from existing LPs, and seeing new deals. Of those forty, maybe twenty-five are available for portfolio support — the rest go to sourcing, diligence, and partnership running. A portfolio company that receives anything less than three hours of focused partner attention per month is, in practice, not getting attention.
Attention per company per month, by portfolio size
60-company fund (4 partners)
Sub-threshold. Default to associate coverage.
1.6 · 1.6
30-company fund (4 partners)
Threshold. Partner attention possible but thin.
3.2 · 3.2
18-company fund (4 partners)
Genuine partner relationships sustainable.
5.5 · 5.5
12-company fund (4 partners)
Approaches board-member level engagement.
8.2 · 8.2
Hours / company / month
Assumes 25 hours / partner / week available for portfolio (the rest goes to sourcing, fund operations, IR). A useful unit of partner attention is roughly 3 hours / month of focused time. The range reflects whether the partner is supporting one company or four.
The math is simple and unforgiving. A 60-company fund with four partners has roughly 1.6 partner-hours per company per month. The work of helping a founder hire well, debate a roadmap, or walk through a P&L cannot be done in 1.6 hours. So it isn't, and it shouldn't be priced as if it is.
Concentration as a return strategy, not a preference
Concentrated funds are sometimes pitched as a virtue. They are not. They are a return strategy with specific implications that LPs need to understand and underwrite. The math behind the strategy is well-trodden: in a power-law-distributed asset class, fund returns are driven by a small number of outsized winners, and the fund's ownership in those winners matters as much as the existence of them.
What concentration buys at the fund level
12–15%
Initial ownership
Across most companies, vs 6–8% at platform funds.
8–10%
Post-dilution at exit
After two to three follow-on rounds.
30–40%
Reserves vs initial
Dry powder for the winners we already own.
1.4×
DPI sensitivity
Estimated change in DPI per 1pp ownership lift in top decile.
Said differently: if our model unit-economics for the fund work at 12–15% initial ownership and 30–40% reserves, we cannot also write eighteen pre-seed cheques and forty seed cheques. The portfolio shape is a constraint, not a slogan.
What concentration is not
Concentration is not 'taking fewer meetings.' We see more deals than most platform funds in our cohort because operator-led discovery surfaces things other funds cannot see. Concentration also is not 'fewer categories.' We invest across consumer, vertical AI agents, and selective infrastructure plays because the underlying thesis — operator-led building in India in the AI era — cuts across these. What concentration is, very specifically, is fewer yeses, with more conviction behind each.
“The right number of investments for a fund is the number where the last cheque the partner wrote would still get her attention on a hard day. For us, that number is eighteen.”
How we say no
We say no often. We try to say it cleanly. Because we have committed to a small portfolio, the no is not a soft no — it is a real no, and we owe founders the reason. The most common categories of no, in our last twelve months of deal flow:
- Strong founder, wrong stage — the company has product-market fit and is ready for a Series A; we are a pre-seed/seed fund.
- Strong category, no operator advantage — we have no partner who has shipped in that space and would not be additive on the cap table.
- Strong narrative, weak unit economics — the AI-amplified version of a category we know does not work without AI.
- Strong everything, full thematic — we already have a similar bet in the portfolio and cannot add a second without prejudicing the first.
If a founder receives one of these nos, we try to send a warm intro to the fund we think would actually be the right partner for them. The opportunity cost of a bad investor is too high for us to pretend not to have an opinion.
That is the doctrine. Eighteen companies. Four partners. Real attention. A no that is honest about why. Concentration, in this fund, is not a preference — it is the price of the promise we have made.