Marc Andreessen — AI, crypto, 1000 Elon Musks, regrets, vulnerabilities, & managerial revolution
What this covers
Andreessen discusses venture capital's structural role within a modern economy dominated by managerial capitalism. Drawing on James Burnham, he argues that by the mid-twentieth century, most of the economy shifted from bourgeois capitalism—where founders own and run their own companies—to managerial capitalism, where professional managers control large organizations they do not own. Within this regime, managers excel at running things at scale but structurally cannot build new things; startups and venture capital exist to restore pockets of entrepreneurial creation. The conversation ranges across what actually constrains innovation, how founders should think about succession, why prior basic research determines venture returns, and the persistent mismatch between supply of capital and quality of investable opportunities.
Andreessen identifies the binding constraint on progress not as money but as the number of great entrepreneurs and the stock of underlying technological breakthrough. He examines how managing managers—rather than individual contributors—becomes the decisive skill for scaling organizations, and why founder-CEOs almost always hand succession to a capable operator who can maintain the company, then regret it. On crypto, he argues that public token pricing creates perverse incentives absent from private equity. He defends speculation as economically necessary and attacks the incumbent systems running education, healthcare, and law as cartels incapable of self-reform. Throughout, he returns to path dependence and the difficulty of judging decisions after the fact, emphasizes that contact with market reality remains essential even for long-horizon bets, and suggests that Software Eats the World is giving way to an even larger opportunity as technology infuses traditionally static sectors like real estate and government. The session includes his view that venture capital itself is oversized by a factor of five.
Andreessen argues that venture capital exists to restore pockets of 'bourgeois capitalism' within an overwhelmingly managerial economy, because managers run things at scale but cannot build new things — and that this dynamic, not money, is the true bottleneck on innovation across AI, crypto, and legacy sectors like education and healthcare.
- Managerial capitalism runs the ~99% of the economy at scale but structurally cannot build new things, leaving startups to fill the gap
- The binding constraints on innovation are great founders and prior basic research, not capital, which is in glut
- The fundamental VC role (project picking) is durable across centuries even as its details and labels change
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
When an asset has a daily, tradable price signal, people obsess over it and trade too much in ways that damage their returns; crypto tokens float publicly years before equivalent equity would, so crypto investors structured as hedge funds trade tokens too aggressively and pay themselves annually, creating misalignment with LPs and mispricing of projects still years from realizing value.
“when something has a daily price signal and where you can trade it, people tend to obsess on the daily price signal and they tend to trade it too much... Almost every investor in almost every asset class trades too often in a way that damages their returns.”
While the details of venture capital constantly change, the core activity is a durable role Tyler Cowen calls 'Project Picking' — a pattern recurring for hundreds or thousands of years in which someone exercises judgment and taste to back this idiosyncratic founder and not that one, provides endorsement/branding, and finances and assists the risky, complex, path-dependent undertaking; this role will exist in 50-200 years even if it isn't called venture capital.
“he calls it Project Picking. When you're doing new things... there's this pattern that just repeats over and over again... you're going to have a very idiosyncratic set of people who start and run that project.”
Following Burnham, there are two kinds of capitalism: bourgeois capitalism, where one person owns and runs the business with a direct link between ownership and control, and managerial capitalism, where ownership and management separate so a professional managerial class ends up in total control despite owning little; by the middle of the 20th century the economy transitioned to the managerial mode, an inevitability driven by the scale and complexity of modern organizations.
“In managerial capitalism, you have a separation of ownership and management... the managers end up in control. Even though the managers don't own the company.”
The decisive breakthrough that determines whether someone can run a company at scale is whether they can manage managers rather than just individual contributors; since even CEOs of 100,000-person companies have only eight to twelve direct reports, the ability to manage a team of managers is a generalizable, scalable skill that lets someone climb from engineering manager to director to VP to CEO.
“The very big breakthrough is whether they know how to manage managers... even the CEOs of really big companies, they're only really dealing with eight or ten or twelve people on a daily basis.”
The binding constraint on building big things is not capital but the number of great entrepreneurs and the number of big ideas for them to pursue; the deeper question is what makes a great entrepreneur — whether Elons are born or trained, already out there undiscovered, or could be produced by a new kind of entrepreneurial university — because if you show the firm ten more Elons it can figure out how to fund their companies.
“The underlying question is how many great entrepreneurs are there? And then how many really big ideas are there for those entrepreneurs to go after?”
Every startup is charting a path-dependent course through a complex adaptive system, so revisiting old decisions is unproductive: because reality has random elements everywhere, running the same scenario from point A can end at point B one time and point Z another, making counterfactual reasoning a trap that makes people less effective.
“every startup is charting a path dependent course through a complex adaptive system... You run the same scenario, start with point A, one time you end up point B, one time you end up point Z.”
Managers are very good at running things at scale but structurally cannot and will not build new things, because they lack the training, background, personality, temperament, and incentives — their number one job is not to upset the applecart, so they avoid any risk that threatens their position and compensation.
“managers don't build new things. They're not trained to do it, they don't have the background to do it, they don't have the personality to do it, they don't have the temperament to do it, and they don't have the incentives to do it.”
We live in a world of a massive oversupply of capital — Bernanke's 'global savings glut' — and not nearly enough investable projects or great entrepreneurs, so there is no real tradeoff between funding flying cars and funding art/NFTs; the supply-demand imbalance means we need more of all of it, not a choice between Edison and Beethoven.
“The world we live in is actually the opposite of that, where we have a massive oversupply of capital and not nearly enough things to fund.”
Funding projects on 50- or 100-year timeframes does not work because new things need milestones and points of contact with reality; companies that go into a tunnel for 10-20 years without exposure to customers and market pricing become self-referential bubbles that generally never produce anything, so roughly five to seven years of contact with reality is important.
“new things don't tend to go through a 20 year incubation phase in business and then come out the other end and be good. What seems to happen is they need milestones, they need points of contact with reality.”
At early stages the best companies tend to raise at less than the optimal price because the signal of who invests matters more than absolute price; most a16z Series A investments could raise at 2-4x the price elsewhere but value the signal, and founders who take the highest bid often regret it because the wrong investors panic and can shut a company down at the first sign of trouble — a self-correcting equilibrium where the best founders choose investors who understand the nature of a startup.
“At the early stages the best companies tend to raise at less than the optimal price because the signal of who invests is more important than the absolute price.”
For a decade smart Western analysts dismissed social media as narcissistic and useless ('learn what someone's cat had for breakfast'), but Erdoğan, who called Twitter the primary challenge to the survival of any political regime, was right and the analysts were wrong — and we are still only a decade into 5 billion people being online and about five years into mass social networking, so the consequences are early, like the printing press taking 200 years to fully play out.
“Erdoğan came out and he said, "I think Twitter is the primary challenge to the survival of any political regime in the modern world,"... My conclusion was Erdoğan is right and all the smart westerners are wrong.”
Per Bill Janeway, the only venture sectors that have produced strong returns over 50-60 years are computing and biotech, and this is a direct result of decades of prior federal basic-research funding — VC productizes 50 years of basic research in computer science and biology — so sectors lacking that backlog of basic research (cleantech, etc.) burned capital, and one can predict future venture returns by looking at where prior basic research was funded.
“Computer science based venture capital was able to productize 50 prior years of basic research in computer science, information science, information theory, communications theory, algorithms.”
The incumbent education system is self-destructing: K-12 prioritizes teachers over students, universities have become cartels (Stanford has more administrators than students) that are eliminating intelligence signals by dropping SAT/ACT/GRE, are intensely politicized, produce mostly fake research (per the replication crisis), and charge soaring costs — and the cartel is administered through quasi-governmental accreditation bodies staffed by the administrators themselves, controlling access to federal student loans, so an educational revolution is coming even if its timing and form are unknown.
“they’re a cartel. They operate as a cartel, they run as a cartel, it is a literal cartel. And the cartel is administered through the agencies, the quasi-governmental bodies that determine who gets access to federal student loan funding.”
Great founder-CEOs facing succession almost always hand the company to someone who can run it at scale — often the long-suffering number-two COO — rather than to a mercurial entrepreneurial type like themselves, and they often later regret it; that they do this repeatedly illustrates Burnham's thesis that large complex organizations end up run by managers in almost all cases.
“Almost always, what they do is they hand it off to somebody who can run it at scale.”
There is a distinction between bad speculation (obsessing on daily prices and day-trading) and good speculation (estimating uncertain future values, utility, and aesthetic worth); the latter is core to how all markets work, with no clear dividing line between it and ordinary investing, and the persistent taboo against it is an ancient superstition and resentment toward money rather than a real economic problem.
“If by speculation, on the other hand, you mean — look, there are different kinds of things in the world that have different possible future values. And people are trying to estimate those future values... I think that kind of speculation is good and healthy.”
The gap between the Mona Lisa's $25 material cost and its ~$10 billion cultural value exists because people care — value is aesthetic, cultural, and part of humanity's heritage; the thing that makes life worth living is that we are not merely material and we pursue higher values like aesthetics, and supporting young artists through speculative purchases can be both speculation and genuine patronage.
“what explains the spread between $25 and the $10 billion that it would go out if it ever hit the market. It's because people care. Because it's art, because it's aesthetic, because it's cultural.”
Tech as a share of GDP will only rise as it infuses every sector, and the opportunity set is shifting from the dynamic-but-small slices of GDP tech captured in the last 50 years toward the less dynamic but much larger sectors — education, healthcare, real estate, finance, law, government — which are harder to build companies in but offer much bigger payoffs; entrepreneurial capitalism, not incumbents, will deliver most of that gain.
“more and more now, we're going to be going after the less dynamic but much larger sectors of GDP. Education, healthcare, real estate, finance, law, government are really starting to come up for grabs.”
Per Andy Rachleff, venture capital is always overstaffed and overfunded — by roughly a factor of 5, so it should be about 20% of its current size in people, funds, and money — because the world has a massive imbalance of too much capital chasing too few productive opportunities; the Swensen endowment model allocates ~4% to venture, and 4% of the global asset base is a gigantic number that floods the sector.
“Venture capital is always over staffed and over funded. His estimate is that it is overfunded by a factor of 5. It should probably be 20% of the size that it is.”
The best case for a large modern organization is a bourgeois entrepreneurial CEO managing a team of managers who handle the scale work — getting the benefits of scale while still building new things — whereas the degenerate version is a manager-CEO running product-builders, in which case the company won't actually build its products and those people will leave to start their own companies.
“The best case scenario is a bourgeois capitalist, entrepreneurial CEO, managing a team of managers who are doing all the managerial stuff required at scale.”
AI's impact on coding means there's a real possibility every application category gets upended within five years, shifting software from the old database/front-end/forms model to a model where apps are a real-time dialogue between human and machine, requiring an entirely new stack to be built.
“I think there's a real possibility that basically every application category gets upended in the next five years. I think the whole model of how applications get built across every domain might just completely change.”
The histories of great companies are systematically distorted because there were early tenuous moments that could easily have gone the other way — Netflix nearly sold to Blockbuster, Google nearly sold to Yahoo — so outside criticism of management teams is unreliable since a thousand factors invisible from the outside make internal decisions hard.
“Netflix could have sold out to Blockbuster early on, and Google could have sold out to Yahoo. And we never would have even heard of those companies.”
NFTs are not worthless 'baseball cards'; they are a broad mechanism for digital ownership, and the most-criticized form — NFTs as creative art — is simply digital art, so every criticism of it applies equally to buying and selling paintings, photographs, or sculpture, and art has been a fundamental, valuable part of every economy for thousands of years.
“they're just art. That's just digital art, right? And so every criticism people make of that is the same criticism you would make of buying and selling paintings.”
Whenever anyone asks whether they should start a company, the answer is always no, because starting a company is a gigantic, emotional, irrational commitment whose profound implications for how you live your life make the romanticized process far harder than it appears.
“Whenever anybody asks me if they should start a company, the answer is always no. Because it's such a gigantic, emotional, irrational thing to do.”
Modern tech founders are throwbacks to the older bourgeois capitalist model, drawing a straight line from Thomas Edison and Henry Ford to Steve Jobs, Larry Page, and Mark Zuckerberg — founders who are CEOs and at least at the start owned 100% of their companies, even though they typically must raise money.
“You can just draw a straight line from Thomas Edison, Henry Ford to Steve Jobs, Larry Page, and Mark Zuckerberg. That's that model, it's a founder, it's a CEO, at least when they started out owning 100%.”
Because computer science now applies across many markets rather than being just an industry that makes computers — the 'software eats the world' thesis — there is a payoff to software in sectors like financial services and healthcare, making Andreessen more optimistic than Janeway about a broader spread of categories yielding special companies even without a large prior basic-research base, as SpaceX showed in aeronautics.
“computers used to be just an industry where people made and sold computers. But now you can apply computer science into many other markets, financial services, and healthcare, and many, many others, where it can be a disruptive force.”
The formal public/private investing distinction is already dissolving into shades of gray: late-stage private companies increasingly have semi-liquid traded stock via secondaries and tenders, new mechanisms like crypto tokens are emerging, and a company can be private in equity yet effectively public through publicly-traded bonds and disclosed financials — so the definition of 'public' will broaden and the line will fuzz out.
“In the last 20 years, it's become much more common for especially later stage private companies to have their stocks actually trade. Actually be semi liquid and trade either through secondary exchanges or tender offers.”
Over a 20-year horizon a16z's most likely failure cause is simply making bad investments, but the scariest, because it is exogenous and out of their control, is that there isn't enough underlying technological change — not enough prior basic research paying off — so the sectors entrepreneurs operate in just don't pay off, as could happen if they hit five cleantech-like sectors in a row.
“The other downside case would just be that there's just not enough technological change happening. There wasn't enough investment in basic research in the preceding 50 years.”
Few VCs become entrepreneurs because building a company is harder than investing and demands a level of personal commitment comfortable VCs are reluctant to return to, and because the activities are fundamentally different: founding is a full-contact sport of a hundred decisions a day requiring a bias to action, whereas investing is analytical, clinical, outside-in, with long decision cycles and infrequent trading.
“actually starting and running a company is a full on contact sport, it's a hundred decisions a day... Being an investor is different. It's much more analytical, clinical, outside-in.”
The role of startups and entrepreneurial capitalism is to bring back enough of the old bourgeois capitalist model to build new things within a 99% managerial economy; VC enables this at the point of inception and private equity at the point of transformation, so if VC were ever snuffed out the economy would become 100% managerial and there would be no innovation forever.
“the role of startups, the role of entrepreneurial capitalism, is to basically bring back the old bourgeois capitalist model enough... to at least be able to build the new things.”
Restricting private-company investing to high-net-worth accredited investors who can bear losses, while barring normal investors, cuts off growth investing as an opportunity for ordinary people and thereby makes wealth inequality worse — a counterargument to the investor-protection rationale that will keep playing out as both sides moderate.
“there's a counter argument that says, then you're cutting off growth investing as an opportunity for normal investors, and you're making wealth inequality worse.”
A real risk to a16z is that innovation gets outlawed in big, important areas, as it already is in nuclear: the Nuclear Regulatory Commission has not authorized a new nuclear design in nearly 50 years, making new nuclear effectively illegal in the US, and Andreessen sees little prospect of nuclear fusion becoming legal despite excellent technology and genius entrepreneurs.
“the Nuclear Regulatory Commission has not authorized a new nuclear design since its inception nearly 50 years ago. So it's just illegal to build new nuclear in the US.”
Healthcare has lots of activity but it is extraordinarily difficult to find positive outcomes — anything that actually causes people to live longer or be healthier longer; analyses show massive investment and public health-insurance support don't move health outcomes, so new market-driven ways of doing things will be needed for people who care about the reality of their health.
“It's very hard to find anything that causes people to live longer. Or to be healthier longer... massive investment and public support for health insurance and all these things. And the health outcomes basically don't move.”
When the companies VC funds reach scale they get pulled into the managerial matrix, which is exactly when they stop being able to build new things, causing the smart and aggressive people there to leave, return to VCs, and start new bourgeois capitalist companies — a cycle that keeps the economy vital.
“they tend to get pulled into the managerial matrix, which by the way, is when they stop being able to build new things, which is what causes the smart and aggressive people at those companies to leave and then come back to us and raise money and start a new bourgeois capitalist company.”
Twitter's core innovation — a one-way public follow graph (publish-subscribe) — ought to be titanically valuable as the most complete expression of what people care about and the primary channel through which every creator interacts with their audience, yet the economic opportunity around it is far larger than anyone has captured, with monetization paths like creators promoting and selling tickets, events, and content directly.
“the public-follow graph... ought to be just absolutely titanically valuable, that ought to be the most valuable content, loyalty brand signal in the world.”
a16z does not speculate on crypto; it applies the classic venture capital playbook with a 5-20 year hold, backing sharp founders building real products into real markets, and is agnostic as to whether the venture is a company or a tokenized network — even writing into term sheets that a C-corp can flip into a tokenized network at any time.
“we apply the classic venture capital 101 playbook to crypto... we always write in the term sheet that they can flip it into being a tokenized network anytime they want to. We don't distinguish between companies and networks.”
a16z treats AI not as a separate fund but as integral to its core software fund, because AI is the next turn on software; separate funds are only created for areas structurally different in how their industries work, whereas AI is the future of the core of the firm.
“AI basically is the next turn on software. And so I view it as the opposite of what you said, it is the most integral thing that we're doing.”
Rather than elongating timeframes, the more interesting experiment for venture is funding companies with far larger amounts of capital ($2-20 billion rather than $20 million) on normal 5-10 year horizons, since world-changing companies like Tesla and SpaceX show that some fundamental things can be built in the entrepreneurial model with large amounts of money.
“It's possible that there are pretty big fundamental things that could be built with larger amounts of money in this kind of entrepreneurial model.”
The stress of a founder differs in kind from that of an investor: for a founder everything that goes wrong is on you and there is no buffer, whereas investors have a layer of buffer because the crisis is not in their own company, making investor stress more diffuse and easier to manage.
“when you're a startup founder, it's all on you. Everything that happens is on you, everything that goes wrong is on you... With investors, there's just a layer of buffer.”
a16z deliberately caps assets under management to the opportunity set rather than pursuing AUM growth, because growing AUM is not a goal; they would not double AUM in exchange for 5% lower returns, instead aiming to maintain return levels, eat market share, and fully fund the available good founders and spaces.
“we're not trying to grow assets under management, that's not a goal... what we wouldn't want to do is double assets under management in return for 5% lower returns.”
a16z, founded in 2009, now has over 500 people, which is small for a tech company but big for a venture capital firm.
“We're over 500 people now at the firm which is small for a tech company, but it's big for a venture capital firm.”