a16z, Fund Size, and the Math of Outsized Venture Returns

In partnership with

My first App Store game is live! Quick Think brain game - try it out and give me a 5 star rating if you like it 🙂 

Andreessen Horowitz announced $15B in new funds this week and Packy McCormick wrote an amazing overview for the launch announcement of the new fund last week which you should read. I took some of his learnings, the SVB state of the market and the Carta state of startups to bring a totally holistic view together to review the last year and how we got here.

The number itself invites a familiar reaction. Venture capital has spent years arguing that large funds and outsized returns are structurally incompatible. That argument is worth revisiting, because both the firm’s own history and the broader market data point in a different direction.

A brief look at the their track record

Andreessen Horowitz was founded in 2009 with a $300M Fund I, raised in the aftermath of the Global Financial Crisis. Fund II followed in 2010 at $650M. Fund III, raised in 2012 at roughly $1B plus a parallel vehicle, became the clearest test of whether the firm’s approach could scale.

At the time, the critique was straightforward. Funds were too large. Ownership would be diluted. The math would not work.

Fund III aged differently. As of late 2025, it has been reported at roughly 11x net TVPI including the parallel fund, with more than $7B net distributed to LPs and substantial unrealized value remaining. Coinbase alone generated approximately $7B of gross distributions across a16z vehicles. Databricks, first backed at a $44M post-money valuation, remains one of the largest unrealized positions in all of venture capital.

The takeaway is not that a16z diversified its way to returns. It concentrated.

Concentration explains the outcome

Databricks is the clearest illustration. At its most recent private valuation of $134B, Databricks is estimated to represent roughly 23% of a16z’s net asset value across funds. That figure refers to NAV, not assets under management.

If you assume firmwide AUM of roughly $60B and apply a conservative adjustment for dry powder and uncalled capital, total NAV likely sits between $40B and $45B. 23% of that implies roughly $9B to $10.5B of Databricks exposure, or approximately 7-8% ownership at current valuations.

This explains how the historical return math works at scale.

A short aside on how the return math actually works

• A venture fund targeting 3x net DPI typically needs ~4x gross outcomes after fees and carry

• Applied to $15B of capital, that implies ~$60B of gross realized value over the life of the funds

• At that scale, returns are not driven by many modest exits

• They are driven by a small number of very large outcomes with sustained ownership at $100B+ outcomes

In practice, that can look like:

• Two or three $20–30B outcomes with meaningful ownership

• Or one $100B+ outcome plus several $5–10B realizations to make it nice

This is why concentration and follow-on matter more than hit rate.

Where Databricks fits into that math

• Databricks was last privately valued at $134B

• It is widely expected to IPO in the $200B+ range

• At roughly 7–8% ownership, that implies $14–16B of value at IPO

• With staged liquidity and continued hold, Databricks alone could reasonably generate $20B+ of gross value to a16z over time

One company meaningfully clearing multiple funds is not a theoretical construct here. It is the model.

Why the scale argument is weaker than it used to be

The common objection to large venture funds assumes smaller outcomes, faster exits, and fragmented capital allocation. Recent market data does not support those assumptions.

SVB’s State of the Markets H2 2025 report shows a venture market increasingly defined by capital concentration, longer private lifecycles, and fewer but much larger category winners. Companies are staying private longer, raising more capital, and achieving materially larger enterprise values before liquidity events.

Carta’s State of Startups 2025 data reinforces the same point. While the number of venture rounds remains below the 2021 peak, total capital invested has rebounded, indicating larger checks flowing into fewer companies. AI companies alone captured roughly 44 percent of total US venture dollars in 2025, with AI taking share across every stage from seed through late stage. Ownership still accrues meaningfully to early investors who maintain conviction and follow-on rights, even as companies scale.

This is a market that favors firms capable of sustained ownership rather than rapid turnover.

The real question

This does not mean the strategy is risk-free. Larger funds require fewer mistakes, longer patience, and deeper conviction. But the data no longer supports the idea that scale alone breaks venture economics.

The more relevant question is whether a firm’s structure matches the market it is investing into. Today’s venture market is producing fewer companies, much larger outcomes, and longer paths to liquidity. That environment punishes shallow diversification and rewards concentration held over time.

a16z is betting that this is a structural shift, not a cycle. The firm’s own history suggests that, at least once before, that bet paid off.

Introducing the first AI-native CRM

Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.

With AI at the core, Attio lets you:

• Prospect and route leads with research agents

• Get real-time insights during customer calls

• Build powerful automations for your complex workflows

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉Try Attio Pro for free