a16z and others are devouring the seed round: a decade of data from 20 top VCs, fully unpacked

Original Author: Pavel Prata

Original Translation: TechFlow

Foreword: Mega-funds managing over $10 billion are flooding into seed rounds at an unprecedented pace. Murph Capital pulled data from Harmonic to analyze the early-stage investment behavior of 20 top mega-funds across three cycles: the SaaS era, the zero-interest-rate era, and the AI era. The conclusions are far from simple. While mega-funds' seed-to-Series B conversion rates are indeed 3.7-4.2 times the market average, this advantage is rapidly diluted when they deploy capital at scale. Opportunities still exist for emerging managers, but they must choose their赛道 carefully.

A month ago, I posted a tweet asking a simple question: Are mega-funds really taking over seed rounds, or does it just feel that way? With 65,000 views and hundreds of DMs, it was clear the question struck a nerve.

Emerging Managers (EMs) wrote in saying they felt the pressure but couldn't quantify it. LPs asked: if a16z and Sequoia are already playing, does it still make sense to invest in seed funds? Even GPs at mega-funds wanted to know how aggressively their competitors were deploying in early stages.

@pavelprata tweeted: Are mega-funds really taking over seed rounds? I decided to study the early-stage behavior of the world's largest VC funds ($10B+ AUM) to answer a simple question: Should EMs worry about their structural advantage?

A broad consensus quickly formed, with which I largely agree:

• Mega-funds have significantly increased their seed round allocations, roughly tripling over the past decade.
• The market is large and fragmented enough that their share remains relatively small, concentrated mainly in the top quartile.
• Their core motivation isn't immediate capital returns, but rather gaining early access to talent, acquiring high signal-to-noise data, and minimizing the risk of missing the next generational opportunity.

But consensus is just the starting point. Behind the big picture lies a more interesting and uneven landscape that isn't visible without the data.

So we pulled data from Harmonic, tracking the performance of 20 mega-funds across three eras (SaaS, Zero Rates, AI) to honestly answer: What is really happening in the seed market? Where exactly are mega-funds going? What impact is this having on pricing? Should EMs genuinely be concerned?

Intuition vs. Data

First, the research framework.

We relied on public information, supplemented by real-time data from Harmonic (covering over 30 million companies and 190 million people). Time-wise, we analyzed the past decade, divided into three eras:

• SaaS Era (2015-2019): 5 years of a normal market cycle. Cloud, SaaS, platforms, and fintech were dominant narratives. Interest rates were normal, and the market was disciplined.
• Zero-Rate Era (2020-2022): 3 years of near-zero interest rate policy. Capital was almost free, with various investors flooding into early-stage seeking returns. Tiger Global and SoftBank seemed to appear in every meaningful funding round. The seed market was severely overheated, but in a chaotic and structurally illogical way.
• AI Era (2023-2026): From the launch of ChatGPT to today. A massive technological shock has spawned a new type of company for which mega seed rounds are the norm.

Technically, we focus on Seed rounds, but operationally, we included Pre-Seed and Seed Extension. The reason is simple: the boundaries between these early stages are often blurred or variable, making strict segmentation disingenuous.

Let's dive in. Honestly, before starting the research, I had a strong intuition that mega-funds were appearing more frequently on early-stage radar. This intuition stemmed largely from social media—the logos of a16z, General Catalyst, and Sequoia appearing increasingly frequently in seed round announcements, each accompanied by high-profile media campaigns. The data confirms this:

• In the first 6 months of 2026, a16z participated in approximately 48 seed round transactions, leading 46% of them. This is a systematic seed strategy, not isolated bets.
• The most striking aspect is the check size: the median for a16z-led rounds is $10.5 million, a figure more reminiscent of a classic Series A than a traditional seed round.
• Adding General Catalyst and Sequoia, these three giants completed 87 seed deals in just 5.5 months, averaging one early-stage investment every 1.5 business days.

@a16z tweeted: We are thrilled to lead the seed round for Westmag. An underappreciated advantage of investing across the entire hardware stack is getting first-hand exposure to the supply chain challenges plaguing the industrial base…

Meanwhile, Carta's latest data shows that, from a valuation perspective, seed round valuations are inflating rapidly. While some might attribute this solely to a few aggressive players, the fund math for most EMs still forces them to operate at or below the median to secure sufficient initial ownership and maintain a viable return path.

The logic for mega-funds is entirely different. With accumulated AUM, brand premiums, and premium deal flow, price discipline is no longer a real constraint. This gap is tearing the market into two distinct tiers, which we broadly call "Classic Seed" and "Super Seed":

• The 90th percentile for seed round valuations soared to $93.7 million in Q1 2026, nearly double that of four years ago.
• Over the past year, valuations above the median have risen by at least 53%.
• The bottom has barely moved: the 25th percentile crept from $18 million to $22.7 million.

@PeterJ_Walker tweeted: Top 5% of seed valuations now frequently exceed $175 million, tripling over the past 12 months. Has a bit of that 2021 absurdity feel (even as an AI believer).

But all of this is still circumstantial evidence. It points to a major trend without providing definitive answers about what is really happening in the early market and how systematic the presence of mega-funds truly is.

That's why we decided to dig deeper. We analyzed the individual dynamics of each fund across the three eras, deconstructed their behavioral patterns, and examined what this shift ultimately means for EMs.

Deconstructing the Deal Machine

Caption: Comparison of early-stage transaction counts for 20 mega-funds across three eras.

Looking at the average, a typical mega-fund in the SaaS era completed 10.6 early-stage deals per year. By the AI era, this jumped to 23.9 deals, an average growth of 2.37x across the entire cohort.

The most interesting aspect is what happened after the zero-rate era ended. If this growth was purely a byproduct of free money, it should have reversed after rate hikes. However, among the 20 funds in our dataset, the average annual deal count in the AI era is nearly identical to the zero-rate era: 23.9 vs. 24.3. In fact, only 3 funds reduced their early-stage investment pace. This proves the shift is structural, although a few outliers pull the overall numbers up:

• a16z: 16.6 → 49.7 → 76.8 deals/year
• General Catalyst: 15.2 → 33.0 → 62.1 deals/year
• Khosla Ventures: 14.6 → 21.0 → 30.9 deals/year

At least three fundamental drivers underlie this:

AI-era companies are inherently more capital-intensive. GPU infrastructure, data pipelines, and research scientists earning $300k–$500k annually create a completely different baseline cost. What cost $500k in the SaaS era (two engineers plus AWS) now requires $2 million to $5 million in the AI era. The expanding median check size partly reflects genuine R&D expenditure, not just valuation inflation. Moreover, early-stage dynamics in the SaaS era were fundamentally exploratory (allowing founders to iterate, pivot, and spend years finding PMF), whereas the first-mover advantage window in AI is much shorter. If your model works, you pull ahead rapidly, and the window closes faster.

The competition for founders shifts pricing power. At the outset of a revolutionary technology cycle, high capability paired with top talent is invaluable. The best AI founders can choose between a16z, Sequoia, and Lightspeed at the seed stage, building a cap table that helps them raise larger subsequent rounds faster. Often, pricing power shifts from investors to founders: rounds get bigger not because the company objectively needs more capital, but because founders can demand it and get it.

Fund size math is very telling. The combined AUM of the top 5 funds in our cohort grew from roughly $34 billion to $249 billion, about a 7x increase over the decade. Meanwhile, their seed transaction counts only grew 2-4x. AUM expanded much faster than seed activity, meaning seed checks now represent a smaller fraction of these funds' portfolios.

Take a16z: In 2015, it managed about $4 billion; now, it manages $90 billion (including the latest $15 billion raise, the largest single VC fund in history). A $6 million seed check represents only 0.01% of $90 billion in AUM. Mathematically, the fund has no incentive to haggle over every million in valuation. Conversely, in an increasingly concentrated market, the risk of missing a generational opportunity is catastrophic.

Therefore, we can state with high confidence: The influx of mega-funds into seed rounds during the AI era is not speculative behavior from the free-money period but a strategic imperative. Massive capital flowing into mega-funds, coinciding with the emergence of a new class of companies and talent worth competing for at the earliest stage, jointly drove this shift.

Group Analysis Based on Growth Rate

Caption: 20 funds grouped by growth trajectory.

During the zero-rate era, all 20 mega-funds in the dataset increased their early-stage deals, without exception. After the pandemic, the Fed cut rates to near zero, massive LP capital flowed into VC pockets, and total US VC fundraising reached a staggering $169.5 billion in 2021.

Flush with dry powder, some mega-funds tested the seed stage waters; others actively retreated from late-stage rounds (where valuations were extremely inflated at the time) and also moved downstream.

But in the AI era, with rates stabilizing above 5%, the market became highly fragmented. Macro divergence split the funds into three behavioral paths:

Accelerators

Their AI-era transaction volumes even exceed those of the zero-rate period:

• a16z (75.3 deals/year)
• General Catalyst (61.5 deals/year)
• Khosla Ventures (31.5 deals/year)

These funds didn't just stay active at the seed stage after cheap capital vanished; they doubled down, aggressively expanding their presence.

Stabilizers

AI-era volumes are slightly below zero-rate peaks but remain well above SaaS-era levels:

• Sequoia (19.6 → 49.3 → 50.6)
• Accel (15.2 → 43.3 → 34.7)
• Lightspeed (11.6 → 41.7 → 32.1)

The zero-rate spike has peaked and receded, but baseline activity has permanently risen to 2-3 times historical levels. There's no going back.

Disciplined

Steady growth across all three eras:

• Bessemer (9.4 → 23.0 → 20.9)
• Lux (7.2 → 14.3 → 14.7)
• Index Ventures (10.0 → 23.3 → 17.6)

They avoided the zero-rate surge and the AI explosion, but their baseline has permanently shifted upward. From 10 deals per year in the SaaS era, they now stabilize at 15-21 deals.

The only exceptions are three funds: Founders Fund, NEA, and Greylock. They either reduced or kept early-stage activity flat from the SaaS to the AI era.

Founders Fund is perhaps the only institution that made a philosophical, active choice. Peter Thiel's contrarian framework, heavily influenced by Girard's mimetic theory, treats crowded market consensus as a clear signal to look elsewhere. So, while the other 17 mega-funds rushed to the seed stage, Founders Fund went the opposite way, focusing on large, concentrated late-stage bets, channeling capital into generational outliers like OpenAI, Databricks, and Anduril.

Greylock remains deeply committed to the 'first check' tradition but chooses to play a high-conviction game. It doesn't operate a high-volume deal machine; instead, it focuses on fewer, higher-conviction bets, sometimes even incubating companies within its own offices.

NEA's large multi-stage mandate makes its seed-stage volatility harder to analyze in isolation. Without hard data, we will refrain from speculation.

Core Allocation vs. Side Hustle

Caption: Changes in the proportion of early-stage deals as a percentage of total investments for each fund.

Absolute numbers don't answer a key question