a16z: A Study of Andreessen Horowitz

Fund Overview: AUM and a Brand, but Middling Returns

Scale: a16z currently manages roughly $35B across 27 funds, the largest being its venture funds (organized by stage) and its crypto funds (organized by theme). The product line is broad: 8 venture funds, 1 late-stage fund, 4 bio funds, 4 crypto funds, and 2 growth funds.

Returns: From publicly disclosed performance, Fund I reached a 44% IRR (far above the industry mean) and returned 2x capital to investors; no later vintage has matched it, and a meaningful share of a16z funds have failed to beat the S&P 500 IRR for their vintage year.

Seen as a timeline of fund launches, a16z's product line has expanded from a single flagship fund into a multi-fund matrix covering bio, crypto, growth, late-stage, seed, and even games:

Flagship funds: the first three vintages did well, then performance fell off a cliff

The flagship line: a16z's first three funds hit star deals across mobile internet and enterprise software — Skype, Airbnb, Slack, GitHub, Coinbase — and the returns from those stars alone have essentially exceeded each fund's size. The two vintages that followed produced no stars, and their outlook is not encouraging.

Specialist funds: the return picture is not encouraging

The specialist line: Take crypto. a16z has raised four crypto-themed funds, but apart from the first, all three later vintages have returned relatively poorly, and absolute returns on individual deals have been modest.

Coinbase remains the core of a16z's crypto franchise: Crypto Fund I performed well because it kept following on into Coinbase, the star deal originally made from the flagship funds. Beyond Coinbase, a16z has yet to replicate that success anywhere in crypto.

Fundraising: on the strength of the brand, raising has come relatively easily

The record: a16z has raised a new fund every year. Since 2019, annual fundraising has exceeded $2B, and the largest single fund has topped $4B (the fourth crypto fund).

A shift in playbook: Early on, a16z mainly raised generalist funds able to back multiple verticals. From the middle period onward — beginning with crypto (Crypto Fund I reached a DPI of 3) — a16z pivoted to raising specialist vertical funds, successively launching bio/health, CLF, games, and other dedicated vehicles.

LP mix: Per PitchBook, a16z counts 39 LPs, led by corporate pensions and family foundations. More than 15 LPs have backed a16z across five or more raises, and 5 LPs have backed more than ten.

Investment Methodology

Investment themes: software is eating the world

"Software is eating the world": From the firm's founding, Marc Andreessen set "software" as the core investment theme, arguing that in the decade after 2010 every industry would undergo a software revolution. Today a16z's themes span 7 specialist verticals (bio/health, consumer, crypto, enterprise, fintech, games, growth), 2 thematic programs (American Dynamism — backing companies "aligned with the U.S. national interest"; the Cultural Leadership Fund — backing "Black American tech entrepreneurship"), and 1 incubator (Talent x Opportunity — backing high-potential founders "in consumer or technology, deeply experienced, but outside the Silicon Valley network").

Deal statistics: a16z has made 1,590 investments in total (leading 218, or 32.6%; following on in the next round 44% of the time), with an average check of $6M, across 643 portfolio companies (of which 329 have exited).

An all-weather fund: a16z places no restriction on stage, and since 2011 has never done fewer than 60 deals a year — it has no "capital winter." Judging by yearly deal count and deal size, median portfolio-company valuation peaked at $184M in 2021; in pacing, 2013 and 2021 were the two high points (the source chart does not label yearly deal counts, so that curve is not redrawn here).

How themes are found: search mode — hunting for new technology that is feasible and fires the imagination

a16z partner Chris Dixon holds that investing, like company-building, runs in two modes: search mode and hill-climbing mode. Search mode: in a brand-new, unfamiliar territory, hunt for new technology that is both feasible and imagination-firing — this is a16z's dominant mode today. Hill-climbing mode: once you are already in a market, work out how your product breaks through into the mainstream.

The three core themes (three new hills): A (AI) — the arrival of deep learning, machine learning, GPT-3, DALL-E; B (Biotech) — the genomics and mRNA revolutions; C (Crypto) — building trusted networks on the internet around distributed consensus.

How to spot a new theme, and the first signs of a technology revolution: track talent flows — watch where the world's smartest people, top graduates, and industry specialists are heading. The smart people split into two groups: engineers and great entrepreneurs. Engineers are problem-solvers: the denser the engineering talent in an emerging field, the likelier it is that real problems get solved. Great entrepreneurs turn ideas into shipped products fast — natural partners for the engineers.

Organization: a corporation wearing a partnership's skin

The central figures: Marc Andreessen and Ben Horowitz founded a16z; Marc is the keystone, and Ben functions more as his deputy. The two earlier co-built famous internet companies including Netscape and Opsware. Marc Andreessen is a technical prodigy and serial founder — developer of Mosaic, the first graphical browser; founder of Netscape; co-founder of the software company Opsware. Ben Horowitz is the author of The Hard Thing About Hard Things, an early Netscape employee, Opsware co-founder, and a former HP general manager.

Structure: a team of 500+: a flat pre-investment team (20%) plus a full-chain post-investment team (80%). Everyone carries the partner title, with no elaborate hierarchy. The post-investment side spans marketing, operations, the talent network, the go-to-market network, the capital network, people practices, and advisors — supporting both the investing team and portfolio companies end to end.

Fees and compensation: an above-market 3/30 fee structure paired with below-market base salaries; incentives run chiefly through carry, while management fees fund the large post-investment organization.

Decision rights: only the 20-odd general partners (GPs), each focused on a sector, hold investment decision authority.

Hiring bar: investing-team members all have founder or early-startup experience in their sectors and degrees from top schools; post-investment team members are senior specialists in marketing, recruiting, corporate strategy, operations, and related fields.

Key partner: Marc Andreessen — an internet sector player whose style runs fast and hard

A sector (Beta) player: Marc invests by first locking onto the Beta. Early on he concentrated on mobile internet and enterprise software; later he entered through vertical SaaS and pushed aggressively into AI, the metaverse, and blockchain. Why "sector player": in web3, he covered layer-1s and the application layer, yet none of these were the earliest or defining projects of the space — the pattern is find the Beta first, then blanket the sector.

Building a circle of competence in media: even in the firm's founding days, Andreessen's shelves were stacked with books on how media works, and he is skilled at managing press relationships (whether to be close to the media is always his own choice). He has also used investments in media to shape opinion; after 2016 he hoped to bypass traditional channels by investing in online media platforms, though most of those social-media bets ultimately failed.

Early-stage focus: of the 27 deals Marc has led, 59% were at Series A or earlier; early-stage deals (Seed, Series A + B) number 22, or 81% combined. Doubling down on conviction: 17 of his deals — over 60% — received follow-on investment, and 5 companies were backed across four or more rounds.

Mapped over time and sector, Marc's activity clusters along three lines: social media — BuzzFeed (2006), Pinterest (2008), Facebook (2009), Twitter (2009), Instagram (2010), Reddit (2014); enterprise software — Slack (2009), Okta (2009), Figma (2012), Samsara (2015); fintech — eBay (1995; a 2002 personal investment), Stripe (2010). Social media is the "media dream" he has never given up.

Key partner: Chris Dixon — strongly composable, crypto-native projects that rework the traditional take rate

Background: Chris joined a16z in 2013 and made the firm's first blockchain investment, Ripple, the same year. In 2021 his $350M crypto fund logged $6B in unrealized gains — a 17x — while the crypto market itself rose only 2x. a16z crypto now runs four funds totaling $7B under management.

Crypto investment philosophy:

• Hunt for exponential forces: Moore's Law in hardware, compounding in finance, composability in software — all are signals of coming rapid growth. Web3's open-source composable protocols and token incentives can bootstrap quickly and generate network effects;
• Watch for native applications: every computing wave passes through a skeuomorphic era and a native era; skeuomorphic design thinking is inherited from the previous domain. Web 1.0 (Read Only) digitized prior-era activities such as letter-writing and mail-order. Web 2.0 (Read & Write) grew native applications like social and e-commerce. Web 3.0 (Read, Write & Own) is still in its skeuomorphic era, but native applications are emerging (DeFi, NFTs);
• Overturn the internet's revenue split: Chris sees Web 2's take rates as Web 3's opportunity — by removing the intermediary layer, Web 3 can cut take rates sharply. He therefore focuses on games, streaming services, and the creator economy: fields with high Web 2 take rates that Web 3 can visibly optimize. In his analysis of token network effects, a traditional network's total utility equals its application utility, while a token network's total utility = financial utility + application utility, the two trading off as the user base scales — token incentives can be used to solve a network's cold-start (bootstrapping) problem.

An owned-media power user: consistent with a16z's media doctrine, Chris is a famous crypto KOL with over a million followers across Twitter, Medium, and other channels; his essays routinely set the industry's agenda. His known investments span L1s (Aptos 2021, Avalanche 2018), NFTs (OpenSea 2017, Yuga Labs 2021), CEXs (Coinbase 2012), and DeFi (Uniswap 2018, dYdX 2017).

Sourcing ("find them"): building brand volume through aggressive owned media

From a16z's earliest days, Marc Andreessen was certain that marketing was critical to success — the way to catch up with the established firms quickly. He deliberately pushed team members to become agents for their portfolio companies and KOLs in their fields, using individual people as the industry lever that helps portfolio companies and lifts returns. He also hired numerous news-media editors into a16z.

Owning the channel: in 2010, a16z created the first marketing-partner role in the history of American venture capital, hiring Margit Wennmachers, co-founder of the top PR firm OutCast, to build the firm's own media arm. Compared with a traditional VC, a16z looks more like a media company: its site publishes a constant stream of substantive content — podcasts with prominent guests inside and outside the industry, original and curated reading collections organized by theme into a public knowledge base. The site carries news, commentary, research, and podcasts, and runs multiple newsletters by theme: enterprise tech, fintech, bio/health, American Dynamism, games, and more.

Turning GPs into KOLs: a16z's GPs deliberately cultivate KOL personas, and their channels command enormous share of voice. Starting in 2014, a16z GPs blogged prolifically to promote blockchain ideas and cryptocurrencies, manufacturing narratives that steered market attention. Chris Dixon is a famous crypto KOL who publishes continuously on social media; Marc Andreessen is likewise a celebrity KOL of the venture world whose tweets have repeatedly set off market-wide debate.

Judging ("see them clearly"): non-consensus, plus structurally positive and ephemerally negative opportunities

Structurally positive opportunities: industry efficiency or demand improves 10x, or cost falls to a tenth. Ephemerally negative opportunities: the market misprices something short-term on sentiment, while the long-term fundamentals are unchanged — a likely reversion to normal valuation.

Investing in non-consensus: Andreessen holds that success = being right × being non-consensus, and says all of his big money has come from that quadrant. In his words: non-consensus can be read as crazy. You need to invest in things that look crazy, because most breakthrough technologies and companies looked crazy at the start — the personal computer, the internet, Bitcoin, Airbnb, Uber, 140 characters (Twitter), and so on. The personal computer — he founded browser company Netscape and took it public; the internet — invested in Skype, Instagram, Github; Bitcoin — went heavy on Bitcoin and the crypto sector, with a 100x on Coinbase; Airbnb — Series B investor; Uber — did not invest; Twitter — Series A investor.

Judging: courage is the core quality of a great founder

The screening funnel: a16z invests at roughly 1-in-100 (of about 2,400 startups met, roughly 20 get funded), and no more than half of those ever deliver outsized returns. Andreessen estimates that of the 4,000 companies seeking funding in Silicon Valley each year, only 200 get financed — and 15 of them generate 95% of all returns.

The founder profile a16z prefers: an entrepreneur smart enough to conceive a breakthrough idea and brave enough to build it — or a highly original college dropout attacking a market that is small and unprofitable for now. Broken into four dimensions: Idea (Breakthrough & Crazy) — ideas keep flowing; whether the first one lands does not matter; Market — seek underestimated markets with a high ceiling that nobody notices early; Entrepreneur (Brilliance & Courage) — a preference for dropouts, with courage as the first requirement; Business model — offer a service people genuinely love, one routinely misread as unprofitable in its early days.

Judging: the "Onion Model" — how a16z de-risks an investment

When underwriting a company, a16z peels risk like an onion — assessing it layer by layer and stripping it away as it goes:

Helping ("be useful"): opening the full post-investment stack to portfolio companies

A complete post-investment chain: a marketing team (wins the company attention) + a talent team (wins it people) + a go-to-market team (wins it customers) + a research team (solves its strategy problems). a16z has also built a talent network of over 20,000 people spanning large corporations, investors, media, executives, and technologists.

PR for Airbnb: shortly after a16z led Airbnb's Series B, the company was hit by damaging news of a renter trashing a host's home. That same evening, Marc went to founder Brian Chesky's office to coach him through the PR response and the follow-up.

The battle for GitHub: in 2010, against a crowded field of rivals, a16z won the chance to put $100M into GitHub — called the fiercest investment contest of the five years around it. GitHub co-founder Chris said a16z's services were the most compelling: 80% of its people work on value-add for portfolio companies.

Crypto as a case in point: after entering the crypto market in 2013, a16z persuaded a roster of heavyweight capital-markets and government veterans to join as its regulatory team:

Helping: Anti-Benchmark

Why anti-Benchmark: when Marc and Ben were building LoudCloud, Benchmark was an investor — and unfriendly to technical founders. Partner David Byrne once asked Ben, "when are you going to get a real CEO?"

Case Studies

Signature win: pressing the bet on Coinbase for a 100x at listing

The company: Coinbase, founded in 2012, is America's largest, earliest, and most compliance-forward cryptocurrency exchange.

The team: a16z's verdict on Brian is "the definition of equanimity" — able to ignore short-term market swings, hold to his values, and keep building; its verdict on Fred is "a true crypto futurist," with deep industry understanding and a constant push to advance the field. CEO Brian Armstrong holds a BS and MS in computer science from Rice; co-founder Fred Ehrsam studied computer science at Duke, traded at Goldman Sachs, and is now co-founder of the crypto fund Paradigm.

The timeline: between 2013 and 2020, a16z partner Chris Dixon led six investments into Coinbase. At the listing, a16z was the largest outside shareholder with 15.4% — a stake worth over $10B on day one, an unrealized 100x.

Shared talent: deep personnel ties run between Coinbase and a16z — two members of Coinbase's management once worked at a16z, two board members are current a16z employees, and Coinbase's head of product is now an EIR at a16z. Specifically: Chris Dixon (current a16z partner; Coinbase board observer), Katie Haun (current a16z partner; Coinbase board member), Mike Lempres (current a16z operating advisor; former Coinbase chief policy officer), Balaji Srinivasan (former a16z partner; then Coinbase CTO), and Asiff Hirji (former a16z operating advisor; then Coinbase president and COO).

Signature miss: path dependence — betting on Picplz, passing on Instagram

Picplz: a photo-sharing tool founded in 2009 by Dalton Caldwell and shut down for good in 2012. Caldwell had previously built Imeem (acquired by MySpace), and a16z was his angel investor ($5M).

Instagram (formerly Burbn): began as a microblog with photo-sharing, then pivoted to a pure photo-sharing tool. Founded by Kevin Systrom — ex-Odeo and Google, first-time founder — with a16z as the seed-round institution ($250K). Instagram was soon acquired by Facebook, handing a16z a 312x return of roughly $78M.

Why a16z doubled down on Picplz rather than Instagram:

• Picplz got there first: Picplz committed to being a pure photo-sharing tool earlier than Instagram, and happened to be in a fundraising window; the original Instagram investment had not been made on the photo-sharing thesis at all;
• Picplz had a serial founder: unlike Instagram's, Picplz's founder had built a company before — and sold it — so he carried proof of success;
• The reality it ignored: Picplz took six months and two platforms to reach 100,000 users; Instagram got there in its first week.

Extended Reflections

Does VC have Vintage Years? (1): industry Beta erupts in bursts, and a16z caught America's mobile internet

Industry Beta is born in Vintage Years: the birth of industry Beta comes in distinct windows. The great semiconductor companies were founded in 1965–1970, the great IT companies in 1983–1989, the great PC software companies in 1995–1999, and the great mobile-internet and enterprise-software companies in 2008–2013. (The source report cites Social Capital's bubble chart of the top 100 U.S. technology companies by market cap since 1960 as evidence for these four founding windows; that third-party chart is not redrawn here.)

Fund I's excellence was a Vintage Year effect: across a16z's whole portfolio, its best investments were founded almost entirely in 2008–2013, clustered around mobile internet and enterprise software — a16z stood at the right moment when a sector inside its circle of competence erupted. Later vintages found progressively fewer great companies, which weighed heavily on returns.

Does VC have Vintage Years? (2): the whole industry's performance swings with the year

The fund industry has Vintage Years: U.S. VC industry IRR data show that from 2004 through 2016, the best mean and median IRRs cluster in the 2009–2013 vintages. 2010 and 2013 were the two best years, with industry mean IRRs of 18.98% and 21.16% and median IRRs of 14.91% and 13.36% (figures as in the source text; they differ slightly from the vintage convention used in the chart).

a16z's fund performance correlates with Vintage Years: Fund I (2009) and the Fund III parallel vehicle (2012) posted the firm's best IRRs, mapping exactly onto the peaks in industry mean IRR — fundamentally because those two moments produced the most great companies (Instagram in 2009, Coinbase in 2013).

Mobile internet carried less momentum than the PC internet: VC funds founded after 2010, a16z chief among them, have struggled to surpass the mid-1990s generation typified by Benchmark — at bottom because America's PC-internet Beta was bigger than its mobile-internet Beta.

The power law in VC: 6% of companies produce 60% of returns, and the best funds are not better at avoiding failure

6% of companies produce 60% of returns: for a fund, the most successful deals (those returning over 10x) account for less than 5% of deal count and of dollars deployed, yet deliver 60% of returns — and within the firm, those deals are made by a small handful of people.

VC is an odds business: comparing deal outcomes, great funds and average funds pick wrong companies at essentially the same rate; the great funds are simply better at hitting outstanding companies rather than mediocre ones, with 10x-plus deals contributing over 90% of a great fund's total return. Put differently, funds returning above 5x and funds returning 1–2x fail at similar rates — top funds win on bigger wins, not fewer losses.

Is a heavy post-investment platform worth it? (1): without a big Beta, its ROI is poor

Heavy services mean heavy costs and heavy AUM requirements: a16z fields a 425-person post-investment team; Benchmark, by contrast, runs on fewer than 10 partners. Their management fees and assets under management differ by a factor of seven.

Heavy services sit oddly with a16z's Beta-investing DNA: a16z is at bottom a Beta investor, while heavy post-investment service is fundamentally an Alpha-manufacturing play — a mismatch with the firm's genes, and of little help in actually judging the Beta.

a16z covers more of the good companies: the heavy-service model has let a16z touch more of the past decade's quality deals than Benchmark, with more star cases — a16z performed well in both enterprise software and mobile internet, while Benchmark's circle of competence sits mainly in mobile internet.

But the economics have worsened: after 2013, no Beta on the scale of mobile internet appeared in America, and enterprise software never matched its magnitude. The heavy-service playbook stopped working — neither a16z nor Benchmark could reliably hit big deals — and at that point the spend became a16z's burden.

Is a heavy post-investment platform worth it? (2): it helps the brand — the cost is, in essence, a marketing fee

How a VC builds brand: online through media, offline through post-investment service. Media (high traffic) — blogs, podcasts, cultivated personas, YouTube — builds the firm's media presence; the payoff is that more founders know the firm, and think of it first when raising. Post-investment service (high conversion) — resource-rich support builds word of mouth among founders; the payoff is that a founder holding multiple term sheets leans toward the firm with the best services.

The strong labels a16z now carries: sector labels — devoted champion of social media, AI, crypto, and life sciences (from media plus services); a political label — a "patriotic," racial-equality-minded fund (from media); a culture label — respects founders and offers them the best talent development (from services).

Can VC AUM expand without limit? (1): fund sizes grow with time, but every era has a sensible range

Top-quartile funds mostly run $600–800M: U.S. VC return data show fund AUM rising steadily over the years — a function of monetary debasement and the industry's scaling. But each era has its own dynamically sensible AUM range for returns: today, $600–800M is the range that best balances scale against performance, and once AUM crosses $1B, performance goes out of control.

Can VC AUM expand without limit? (2): growing AUM warps a fund's behavior

a16z's deal sizes grew and its rounds drifted later: at founding, a16z invested mainly at Seed and Series A; by 2016, growth-stage deals had become its largest bucket (the source chart shows 2016 as the dividing line between "mostly early-stage" and "mostly growth-stage," with growth-round investments peaking at 85 in 2021; since only some yearly values are labeled in the source, that multi-series chart is not redrawn here). Deal size has stepped up roughly every four years, from $5M at the start to an average of $25M today.

Neither heavy services nor bigger checks improved returns: a16z's later deals returned markedly worse than its earlier ones even as check sizes kept growing — eroding its identity as an early-stage fund (see Exhibit 3 for the core deals and returns of the first five funds).

Can Beta be manufactured? Top-down vs. bottom-up

Top-down, riding the wave: a16z — a media company hiding behind an investment firm — excels at finding angles top-down and using media distribution to talk sectors upward, as in its role stoking the Web3 boom; it habitually sells its industry theses and investment philosophy to the mass media.

Bottom-up, riding through the cycle: Benchmark works mainly bottom-up, insulated from economic cycles and sector fashions, never bending its methodology to the environment; its core is to start from the person and the business, abstracting one universal method for reading both.

Top-down or bottom-up: top-down demands an expensive brand, and in practice a16z's star deals still came mostly from mobile internet and enterprise software, while crypto's boom and bust proceeded largely untouched by a16z's messaging. Bottom-up demands a stable pipeline of quality deal flow and a universal method — and is inefficient for funds at large scale.

Reusing insight globally: bonus, or must?

The decade's biggest Beta was in China: since 2009, the largest industry Beta in the world has been China's mobile internet. Over the past decade China produced 4 companies worth over $200B and 6 worth over $50B; the U.S., by contrast, produced no company founded in the past decade worth over $200B — enterprise software and mobile internet were its biggest Betas, but with less growth than China's.

Globalization was never in a16z's genes: shaped by their politics, Marc and Ben placed the overwhelming majority of a16z's investments in the U.S.; Marc carried a deep bias against China and did not begin investing there until 2021. Yet at a16z's scale, China's mobile internet may have been the only market that could sustain its returns while it expanded aggressively.

An open question for today: would this play out the same way? The constraint that kept a16z domestic was ultimately political, and the world that produced it has shifted. The AI cycle — unlike the mobile internet — is global from day one: frontier models, open-source ecosystems, developer communities, and AI-native products cross borders faster than any prior platform, and exceptional founders and researchers are more globally distributed than ever. If the biggest Beta of the last cycle sat inside a single geography that a U.S. fund could choose to ignore, the biggest opportunity of this cycle may be globalization itself — backing the teams, talent, and markets that operate across borders. Whether a returns-first fund can still afford to be a single-market fund is, once again, the open question.