💥 Mega-Funds Take Over Seed… | Data Driven VC Summary

"Mega-Funds Take Over Seed, Emerging Managers Rebound, Fee vs Carry, Hermes Agent Masterclass & More" Andre Retterath | June 2, 2026

1. Key Themes

Mega-Funds Have Structurally Repriced the Seed Market

The 10 largest VC funds (>$10B AUM) have nearly tripled their collective early-stage deal volume since the SaaS era — and this is not a ZIRP hangover. The data shows AI-era activity exceeds ZIRP peaks for top funds.

"a16z went from 16.6 early-stage deals/year in the SaaS era to 75.3/year in the AI era. General Catalyst went from 15.2 to 61.5/year. Both funds' AI-era activity exceeds their ZIRP peaks, confirming this is not a cheap-capital hangover."

"These 10 funds made ~140 to 150 early-stage deals/year in the SaaS era. In the AI era that figure is ~370 to 400/year."

Early IRR Is a Misleading Signal — Especially in AI-Era Vintages

Carta data across 2,276 US venture funds shows that IRR before year 4–6 is statistically unreliable, and the current AI-era vintage spread closely mirrors the dangerous 2021 setup.

"2021 Vintage at -4.3% Median IRR: Funds that printed explosive early IRR through rapid markup cycles are now dragging under stale marks. Top-quartile (75th percentile) sits at just 1.3% as of Q1 2026."

"The youngest cohort [2025] shows a 90th percentile of 54.1% versus a 25th percentile of -26.9%. AI markups in a few months can print outsized IRR before any real value is proven."

Large VC Has Become a Fee Business, Not a Returns Business

Odin's Dan Gray presents a damning structural critique: at scale, management fee income has decoupled from LP returns, and companies that raised the most private capital have delivered the worst outcomes.

"The same company held private for 14 years through multiple rounds absorbing ~$1B in VC capital generates $250M in fees, with no improved exit outcome."

"Of the 10 companies that raised more than $3B in private markets, only Robinhood shows a slight positive return vs. the S&P 500. A portfolio built on the thesis that large private raises signal quality would be at -120.5% today."

Skill Distillation: A New AI Architecture for Running Intelligence Cheaply at Scale

Tomasz Tunguz documents a three-layer agent architecture ("Pi") where frontier models write procedural playbooks that cheaper local models execute — separating institutional knowledge from the model itself.

"Because the skill library is decoupled from the executing model, the model can be swapped for whatever is cheapest each quarter with no retraining. The institutional knowledge lives in the files, not the weights."

"Standard model distillation compresses a large model's probability outputs into a smaller model's weights. Skill distillation externalises procedures into inspectable markdown files the smaller model reads and follows."

Emerging Manager Formation Is Rebounding — Into the Worst Structural Environment on Record

New fund formations ($10M–$100M) are rising after an 8-quarter trough, but the recovery is occurring precisely as mega-fund seed density hits all-time highs.

"78 New Funds in Q1 2026, Up from 58 in Q1 2025: Still well below the Q1 2022 peak of 147, but the first meaningful acceleration in two years."

"Formation Recovery Meets Peak Mega-Fund Activity: More new managers are entering the market precisely as the 10 largest funds post their highest seed deal volumes on record."

2. Contrarian Perspectives

Emerging Manager Recovery Is a Trap, Not a Signal of Opportunity

The conventional reading of rising fund formations is that the market is healthy and maturing. The article argues the opposite: new managers are flowing into a market that is structurally hostile to undifferentiated seed funds.

"More dry powder entering the emerging manager tier while mega-fund seed density hits a historical peak means more competition for the same early deals. LPs evaluating new managers should weight differentiation more heavily than at any prior point in the cycle."

The data is stark: Q1 2026 saw 78 new formations while the top 10 funds simultaneously logged their highest-ever seed volumes — a collision, not a recovery.

Brand and Fund Size Now Correlate Negatively With LP-Aligned Incentives

The prevailing LP heuristic — that bigger, more reputable firms are safer bets — is inverted by the fee economics data. Scale has turned management fees into the primary product.

"Top Firm Fee Income Up 10x Since 2005: The five largest VC fundraisers in 2005 generated ~$150M in subsequent fee income. For 2025 vintages that figure is ~$1.6B."

"Smaller, carry-dependent managers with disciplined fund sizes hold a structural alignment advantage."

IRR Before Year 4 Is a Marketing Figure, Not a Performance Signal

LPs routinely use early IRR to benchmark managers and make re-up decisions. The Carta data directly contradicts the validity of this practice.

"Early IRR in AI-era vintages mirrors the 2021 setup: rapid markups inflating numbers before fundamentals catch up. For LPs evaluating newer funds, IRR before year 4 is a marketing figure. Portfolio construction quality and entry price discipline are the only early signals worth tracking."

3. Companies Identified

Carta

Description: Equity management and fund administration platform; maintains data on thousands of US VC funds
Why mentioned: Source of net IRR percentile data across 2,276 US venture funds and new fund formation tracking
Quote: "Peter Walker, Head of Insights at Carta, published net IRR percentiles by vintage across 2,276 US venture funds ($10M to $1B+), measured March 31, 2026."

Odin

Description: VC infrastructure/SPV platform
Why mentioned: Published the structural critique of VC fee economics vs. LP returns
Quote: "Dan Gray at Odin wrote a structural critique of how venture capital's fee economics have decoupled from LP returns."

Murph Capital

Description: Venture capital firm
Why mentioned: Conducted the mega-fund seed activity analysis using Harmonic data
Quote: "Pavel Prata from Murph Capital used Harmonic data to analyse early-stage deal activity across 10 mega-funds ($10B+ AUM)."

Theory Ventures

Description: Venture capital firm led by Tomasz Tunguz
Why mentioned: Source of the skill distillation / Pi agent architecture framework
Quote: "Tomasz Tunguz at Theory Ventures documented a personal agent architecture called Pi."

Kruncher

Description: AI-first private capital CRM and intelligence platform
Why mentioned: Newsletter sponsor; positioned as a tool for automating VC workflows with 450+ configurable signals, MCP server integration, and LP reporting
Quote: "Automate 80% of your firm's workflows from screening to LP reporting."

Harmonic

Description: Startup and funding data provider
Why mentioned: Data source used by Murph Capital to track mega-fund early-stage deal activity
Quote: "Pavel Prata from Murph Capital used Harmonic data to analyse early-stage deal activity."

Robinhood

Description: Consumer fintech / brokerage platform
Why mentioned: Named as the sole positive outlier among the 10 companies that raised >$3B in private markets
Quote: "Of the 10 companies that raised more than $3B in private markets, only Robinhood shows a slight positive return vs. the S&P 500."

4. People Identified

Pavel Prata

Description: Investor at Murph Capital
Why mentioned: Conducted and published the mega-fund seed dominance analysis using Harmonic data
Quote: "Pavel Prata from Murph Capital used Harmonic data to analyse early-stage deal activity across 10 mega-funds."

Peter Walker

Description: Head of Insights at Carta
Why mentioned: Author of both the net IRR vintage analysis and the emerging manager fund formation data
Quote: "Peter Walker, Head of Insights at Carta, published net IRR percentiles by vintage across 2,276 US venture funds."

Dan Gray

Description: Writer/analyst at Odin
Why mentioned: Author of the VC fee economics critique showing decoupling of management fees from LP returns
Quote: "Dan Gray at Odin wrote a structural critique of how venture capital's fee economics have decoupled from LP returns."

Tomasz Tunguz

Description: General Partner at Theory Ventures
Why mentioned: Architect of the "Pi" skill distillation agent system; coined the term "skill distillation" as a distinct approach from classical model distillation
Quote: "Tomasz Tunguz at Theory Ventures documented a personal agent architecture called Pi, where frontier models author procedural skill files that smaller local models execute."

CyrilXBT

Description: Crypto/AI practitioner and content creator
Why mentioned: Published the comprehensive Hermes Agent setup and deployment masterclass
Quote: "CyrilXBT published a complete masterclass on building a fully autonomous Hermes Agent operation."

Andre Retterath

Description: Author of Data Driven VC newsletter; venture investor
Why mentioned: Newsletter author and curator of all insights in this edition
Quote: "Hi, I'm Andre and welcome to my newsletter Data Driven VC which is all about becoming a better investor with data and AI."

5. Operating Insights

For VC Firms: Encode Proprietary Processes Once, Run Them Cheaply Forever

The skill distillation architecture (frontier model writes the playbook → cheap local model executes it) offers a practical cost-efficient AI infrastructure model. The moat is procedural quality, not model selection.

"For VC firms building internal AI tooling, skill distillation is a practical architecture: encode proprietary processes once using a frontier model, then run them cheaply on local models indefinitely. The competitive moat is the quality of the captured procedures, not which model you use."

For Emerging Managers: Structural Differentiation Is Now a Survival Requirement, Not a Nice-to-Have

The data makes the competitive math explicit — generalist seed positioning is untenable when 10 funds are collectively executing 370–400 early-stage deals per year.

"Emerging managers need a specific, defensible sourcing edge: geographies or sectors where these 10 funds lack density, or pricing discipline that wins deals they will not chase. Generalist seed funds without a structural moat are entering the hardest competitive environment on record."

For Hermes Agent Operators: Start Day 1 — Memory Is the Compounding Asset

The Hermes framework's persistent memory (SQLite-backed) creates an asymmetric advantage that accrues over time and cannot be replicated retroactively.

"A Hermes agent at day 90 has processed hundreds of sources, tracked dozens of decisions, and built a detailed picture of what works in a specific operation. That accumulated memory is not replicable by starting later. Day 1 is the only day the compounding starts."

6. Overlooked Insights

The CLAUDE.md as a Single Leverage Point Across All Automated Workflows

In the Hermes Agent setup, configuration quality — not model choice or skill quantity — is identified as the primary determinant of output quality across every automated task. This is a rarely discussed but critical detail for anyone building agentic systems.

"Every skill reads a single configuration file before executing. A precisely written CLAUDE.md shapes every output across every automated workflow. Vague configuration produces generic outputs regardless of skill quality."

This has broader implications: for any multi-agent or skill-based AI system, the system prompt / master config file is where leverage is concentrated — not in the number of tools or the quality of the underlying model.

Even the "Disciplined" Mega-Funds Have Permanently Shifted Upward

The article's focus on aggressive funds like a16z and General Catalyst may obscure a subtler and more durable signal: even funds historically known for selectivity have structurally increased their seed activity, suggesting the shift is market-wide, not firm-specific.

"'Stabiliser' funds like Sequoia (19.6 to 50.6/year) and Lightspeed (11.6 to 32.1/year) operate at 2 to 3x their SaaS-era baselines. Even the most disciplined funds (Bessemer, Index, Lux) show durable upward shifts."

If Sequoia and Bessemer have permanently repriced their seed cadence upward, it signals a structural reconfiguration of the entire asset class — not a cyclical response to AI hype.