Are mega-funds really taking over seed?
I decided to look at the behavior of the world's largest VC funds ($10B AUM) at early stages and answer a simple question: should EMs worry about their structural edge?
So I used
@harmonic_ai and looked at all pre-seed, seed, and seed extension rounds across 3 eras:
- SaaS Era (2015β2019): 5 years of a normal market. Cloud, SaaS, fintech were the dominant theses.
- ZIRP Era (2020β2022): 3 years of zero interest rates and free capital.
- AI Era (2023β2026): from ChatGPT to the present day.
We focused on one core metric: average number of early-stage deals per year for each fund in each era.
Here are a few insights we found interesting:
1/ In the SaaS era, a typical mega-fund made 10β20 early-stage deals per year. This was moderate, targeted seed activity β a complement to the core Series A/B and later strategy.
2/ In the ZIRP era, everyone scaled up. Each of the 10 funds increased their early-stage deals/year (some by 2β3x), because capital was free, competition at later stages was fierce, and seed felt like a cheap entry point.
3/ Then came the AI era and it became clear this was no temporary effect. Even as rates rose and capital became more expensive with the end of ZIRP,
@a16z and
@generalcatalyst posted peak early-stage activity.
>
@a16z: 16.6 β 48.7 β 75.3 deals/year. A 4x increase from the SaaS era.
>
@generalcatalyst: 15.2 β 31.7 β 61.5 deals/year. Also 4x.
The most interesting finding, though, is 3 distinct behavioral models:
1/ "Accelerators" - deals/year in the AI era exceed ZIRP levels:
@a16z (75.3/yr),
@generalcatalyst (61.5/yr),
@khoslaventures (31.5/yr). These funds didn't just stay active in seed after free money ended β they doubled down.
2/ "Stabilizers" β deals/year in the AI era are slightly below ZIRP peak, but well above SaaS-era levels:
@sequoia (19.6 β 49.3 β 50.6),
@Accel (15.2 β 43.3 β 34.7),
@lightspeedvp (11.6 β 41.7 β 32.1). The ZIRP spike moderated, but activity levels remain sustainably 2β3x above the SaaS era. There's no return to the old normal.
3/ "Disciplined" β steady, gradual growth across all eras:
@BessemerVP (9.4 β 23.0 β 20.9),
@Lux_Capital (7.2 β 14.3 β 14.7),
@IndexVentures (10.0 β 23.3 β 17.6). No ZIRP spikes, no AI explosions β but the baseline has durably shifted upward.
So in the SaaS era, these 10 funds collectively made roughly 140β150 early-stage deals per year. In the AI era β around 370β400. And I think they just set up a new, sustained baseline, not just doubled after a ZIRP-peak era.
For an LP evaluating an emerging seed manager, this is the most important context.
The early-stage market your GP is investing in is one where 10 funds with $10B in AUM are doing dozens deals a year.
An emerging manager needs to be able to articulate exactly where, in that market, they have the right to win.