Author: Erik Torenberg

Translation: Shenchao TechFlow

Shenchao Guide: In the traditional narrative of venture capital (VC), the 'boutique' model is often revered, with the belief that scaling loses its soul. However, a16z partner Erik Torenberg presents a contrary viewpoint in this article: as software becomes the backbone of the American economy and the advent of the AI era unfolds, the demands of startups for capital and services have profoundly changed.

He argues that the VC industry is transitioning from a 'judgment-driven' paradigm to a 'deal-winning ability-driven' paradigm. Only 'massive institutions' like a16z, equipped with scalable platforms that provide comprehensive support to founders, can emerge victorious in a trillion-dollar game.

This is not just an evolution of the model, but also a self-evolution of the VC industry in the wave of 'software eating the world'.

The full text is as follows:

In Greek classical literature, there is a meta-narrative that transcends everything: the respect for the divine and the disrespect for the divine. Icarus was scorched by the sun not because of excessive ambition but due to a lack of respect for the sacred order. A more contemporary example is professional wrestling. You only need to ask, “Who respects wrestling, and who disrespects it?” to discern who is the hero (Face) and who is the villain (Heel). All good stories take on one form or another.

Venture capital (VC) has its own version of this story. It goes like this: “VC has been, and always will be, a boutique business. Those large institutions have become too big, and their targets have become too lofty. Their demise is inevitable, as this practice is a blatant disrespect for the game.”

I understand why people want this story to hold. But the reality is that the world has changed, and venture capital has changed with it.

Now, there are more software, leverage, and opportunities than ever before. There are also more founders building larger companies than ever before. Companies are staying private longer than ever before. And founders’ expectations of VC are higher than ever before. Today, the founders of the best companies need partners who can truly roll up their sleeves to help them win, not just write checks and wait for the results.

Therefore, the primary goal of venture capital firms now is to create the best interfaces to help founders win. Everything else—from how to staff, how to deploy capital, how large to raise funds, how to assist in closing deals, and how to allocate power to founders—derives from this.

Mike Maples has a famous saying: your fund size is your strategy. Equally true is that your fund size is your belief about the future. It is your bet on the output scale of startups. While raising large funds may have been viewed as “arrogant” over the past decade, that belief is fundamentally correct. Therefore, when top institutions continue to raise large sums to deploy in the next decade, they are betting on the future and putting their money where their mouth is. Scaled venture (Scaled Venture) is not a corrosion of the VC model: it represents the maturation of the VC model, adopting the characteristics of the companies they support.

Indeed, venture capital firms are an asset class

In a recent podcast, Sequoia's legendary investor Roelof Botha made three points. First, even as venture capital scales, the number of 'winning' companies each year is fixed. Second, the scaling of the venture capital industry means that too much capital is chasing too few great companies—therefore, venture capital cannot scale; it is not an asset class. Third, the venture capital industry should downsize to correspond to the actual number of winning companies.

Roelof is one of the greatest investors of all time, and he is a good person. But I disagree with his perspective here. (It’s worth noting that Sequoia has scaled too: it is one of the largest VC firms globally.)

His first point—that the number of winners is fixed—can easily be disproven. About 15 companies generated $100 million in revenue annually in the past; now there are around 150. Not only are there more winners than before, but the scale of those winners is also greater. While entry prices are higher, the output is significantly larger than previously. The growth ceiling for startups has increased from $1 billion to $10 billion, and now to $1 trillion or even more. In the early 2000s and the early 2010s, YouTube and Instagram were considered colossal acquisitions at a $1 billion valuation: at that time, such valuations were so rare that we referred to companies valued at $1 billion or more as “Unicorns.” Today, we take it for granted that OpenAI and SpaceX will become trillion-dollar companies, and there will be several others following suit.

Software is no longer a fringe sector made up of oddball misfits in the American economy. Software is now the American economy. Our biggest companies, our national champions, are no longer General Electric and ExxonMobil; they are Google, Amazon, and Nvidia. Private tech companies account for 22% of the S&P 500 index. Software hasn’t finished eating the world—in fact, due to the acceleration brought by AI, it’s just getting started—and it is more important now than it was fifteen, ten, or even five years ago. Therefore, the scale that a successful software company can reach is larger than ever.

The definition of a “software company” has also changed. Capital expenditures are rising significantly—large AI labs are turning into infrastructure companies, with their own data centers, power generation facilities, and chip supply chains. Just as every company is becoming a software company, now every company is becoming an AI company, perhaps an infrastructure company as well. More and more companies are entering a world of atoms. Boundaries are getting blurry. Companies are aggressively verticalizing, and the market potential of these vertically integrated tech giants is far greater than that of anyone's imagined pure software companies.

This raises the question of why the second point—that too much capital is chasing too few companies—is incorrect. Output is much larger than before, competition in the software world is much fiercer, and companies are going public later than ever before. All of this means that great companies simply need to raise far more capital than ever before. Venture capital exists to invest in new markets. Time and again, we learn that, in the long run, the size of new markets is always much larger than we expect. The private market is mature enough to support top companies achieving unprecedented scales—just look at the liquidity that top private companies can access today—both private and public market investors now believe that venture capital's output scale will be astounding. We have consistently underestimated how large an asset class VC can and should become, and VC is scaling to catch up with that reality and the opportunity set. The new world needs flying cars, global satellite grids, ample energy, and intelligence that is so cheap it need not be measured.

The reality is that many of today's best companies are capital-intensive. OpenAI needs to spend billions of dollars on GPUs—more than anyone imagines can be acquired in computing infrastructure. Periodic Labs needs to build automated laboratories at an unprecedented scale for scientific innovation. Anduril needs to build the future of defense. All of these companies need to recruit and retain the best talent in a historically competitive talent market. The next generation of large winners—OpenAI, Anthropic, xAI, Anduril, Waymo, etc.—are all capital-intensive and have completed huge initial funding at high valuations.

Modern tech companies often require hundreds of millions in funding because the infrastructure needed to build world-changing cutting-edge technology is incredibly expensive. During the internet bubble era, a “startup” entered a blank slate, envisioning the demands of consumers still waiting for dial-up connections. Today, startups enter an economy shaped by thirty years of tech giants. Supporting “Little Tech” means you must be prepared to arm David to fight against a few Goliaths. Companies in 2021 did indeed receive excessive funding, with a large portion flowing into sales and marketing, trying to sell products that haven't improved tenfold. But today, funding is directed towards research and development or capital expenditures.

As a result, the winners are much larger than before and need to raise far more capital, and often have from the outset. Therefore, the venture capital industry must necessarily scale significantly to meet this demand. Given the scale of the opportunity set, scaling makes sense. If VC scale is too large for the opportunities that venture capitalists invest in, we ought to see that the largest institutions perform poorly. But we have not seen this scenario. In fact, top venture firms have consistently achieved very high multiple returns during expansion—so have the LPs (limited partners) who are able to enter these firms. A famous venture capitalist once said that a $1 billion fund could never achieve a 3x return: because it's too large. Since then, certain companies have exceeded 10x a $1 billion fund. Some point to poorly performing institutions to denounce this asset class, but any industry that conforms to a power-law distribution will have huge winners and a long tail of losers. The ability to win deals without relying on price is why institutions can maintain ongoing returns. In other major asset classes, people sell products or loan from the highest bidder. But VC is a typical asset class that competes along more dimensions than just price. VC is the only asset class where significant continuity exists among the top 10% of firms.

The last point—that the venture capital industry should downsize—is also incorrect. Or at least, it would be a bad thing for the tech ecosystem, for the goal of creating more generational technology companies, and ultimately for the world. Some complain about the second-order effects brought on by the increase in venture capital funding (and there are indeed some!), but it has also coincided with a significant increase in the valuations of startups. Advocating for a smaller venture capital ecosystem likely advocates for smaller startup valuations, and the result may be slower economic development. This may explain why Garry Tan recently stated in a podcast: “Venture capital can and should be ten times larger than it is today.” Admittedly, if there was no competition, some individual LP or GP became the “only player,” that might be beneficial for them. But if there are more venture capitalists than today, that is clearly better for founders and for the world.

To further elaborate on this, let’s consider a thought experiment. First, do you think there should be many more founders in the world than there are today?

Second, if we suddenly had many more founders, what kind of institutions could best serve them?

We won’t spend too much time on the first question because if you are reading this article, you likely understand that the answer is obviously yes. We don’t need to tell you much about why founders are so great and so important. Great founders create great companies. Great companies create new products that improve the world, organizing and channeling our collective energy and risk preferences towards productive goals, creating disproportionate new enterprise value and interesting job opportunities in the world. And we cannot possibly have reached a state of equilibrium whereby every person capable of founding a great company has already done so. This is why more venture capital helps unleash more growth within the startup ecosystem.

But the second question is more interesting. If we woke up tomorrow and the number of entrepreneurs was ten or even a hundred times what it is today (spoiler alert, this is happening), what should the entrepreneurial institutions look like? How should venture capital firms evolve in a more competitive world?

Come to win, not just to lose out

Marc Andreessen likes to tell the story of a famous venture capitalist who said the VC game is like being at a conveyor belt sushi restaurant: “A thousand startups go by, and you meet them. Then occasionally, you reach out and pick a startup off the conveyor belt and invest in it.”

The type of VC Marc describes—well, for most of the past few decades, almost every VC has been this way. Back in the 1990s or 2000s, winning deals was that easy. Because of this, the only truly important skill for a great VC is judgment: the ability to differentiate between good companies and bad ones.

Many VCs still operate this way—basically the same way VCs of 1995 did. But a massive change has occurred beneath their feet.

Winning deals used to be easy—just as easy as picking sushi off a conveyor belt. But now it is incredibly difficult. People sometimes describe VC as poker: knowing when to pick companies, knowing what price to enter, etc. But this may obscure the full-blown war you must wage to secure the rights to invest in the best companies. Old-school VCs miss the days when they were the “only players” and could dictate terms to founders. But now there are thousands of VC firms, and founders can obtain term sheets more easily than ever.

The paradigm shift is that the ability to win deals is becoming as important as picking the right companies—even more important. If you can’t get in, what good does picking the right deal do for you? Several factors have contributed to this change. First, the surge in venture capital firms means that VC firms need to compete with each other to win deals. With more companies than ever competing for talent, customers, and market share, the best founders need strong institutional partners to help them succeed. They need institutions with resources, networks, and infrastructure to give their portfolio companies an advantage.

Second, as companies stay private longer, investors can invest later—by which time companies have received more validation, making deal competition fiercer—and still obtain venture-style output returns.

The last reason, and the least obvious, is that picking has become slightly easier. The VC market has become more efficient. On one hand, there are more repeat founders continuously creating iconic companies. If an Elon Musk, Sam Altman, Palmer Luckey, or a genius repeat founder starts a company, VCs will quickly line up to try to invest. On the other hand, companies are scaling to crazy sizes faster (due to staying private longer, the upward room is also larger), meaning that the risk of product-market fit (PMF) is lower compared to the past. Finally, with so many great institutions today, it’s also much easier for founders to reach out to investors, so it’s hard to find deals that other firms aren’t pursuing. Picking remains the core of the game—choosing the right evergreen company at the right price—but it is no longer the most significant piece of the puzzle.

Ben Horowitz hypothesizes that the ability to consistently win automatically makes you a top firm: because if you can win, the best deals will come to you. Only when you can win any deal do you earn the right to pick. You may not have chosen the right one, but you at least had the opportunity. Of course, if your firm can consistently win the best deals, you’ll attract the best pickers to work for you because they want to get into the best companies. (As Martin Casado said when recruiting Matt Bornstein to join a16z: “Come here to win deals, not to lose them.”) Thus, the ability to win creates a positive feedback loop that improves your picking ability.

For these reasons, the game has changed. My partner David Haber described the shift that venture capital needs to make in response to this change in his article: “Firm > Fund”.

In my definition, a fund has a single objective function: “How do I generate the most carry (performance fee) with the fewest people and in the shortest time?” Whereas a firm, in my definition, has two goals. One is to deliver outstanding returns, but the second is equally interesting: “How do I build a source of compound competitive advantage?”

The best firms will be able to invest their management fees into reinforcing their moat.

How can I help?

I entered the venture capital field ten years ago, and I quickly noticed that among all VC firms, Y Combinator plays a different game. YC is able to secure favorable terms from excellent companies on a large scale while seemingly also being able to serve them on a large scale. Compared to YC, many other VCs play a commoditized game. I would go to Demo Day and think: I’m at the poker table, and YC is the casino dealer. We were all happy to be there, but YC was the happiest.

I soon realized YC has a moat. It has positive network effects. It has several structural advantages. People used to say that venture capital firms couldn't have moats or unfair advantages—after all, you’re just providing capital. But YC clearly has one.

This is why YC remains so strong even after scaling. Some critics dislike YC scaling; they think YC will eventually fail because they feel it lacks soul. For the past ten years, people have been predicting the demise of YC. But that hasn’t happened. During that time, they replaced their entire partner team, and death still didn't occur. A moat is a moat. Just like the companies they invest in, scaled venture firms possess moats that are not just about branding.

Then I realized I didn't want to play the commoditized VC game, so I co-founded my own firm, along with other strategicassets. These assets are tremendously valuable and create strong deal flow, allowing me to taste the fruit of the differentiated game. Around the same time, I began to observe another firm building its own moat: a16z. Therefore, when the opportunity to join a16z arose a few years later, I knew I had to seize it.

If you believe in venture capital as an industry, you—by definition—believe in power-law distributions. But if you really believe that the VC game is governed by power laws, then you should also believe that venture capital itself will follow a power law. The best founders will cluster around those institutions that can most decisively help them win. The best returns will concentrate in these institutions. Capital will follow.

For founders seeking to build the next iconic company, scaled venture firms offer an exceptionally attractive product. They provide the expertise and full-service support needed for rapidly expanding companies—hiring, go-to-market strategies (GTM), legal, financial, public relations, government relations. They provide enough capital to actually get you to your destination, rather than forcing you to penny-pinch while struggling against well-funded competitors. They offer immense reach—connecting you to every key player you need to know in business and government, introducing you to every important Fortune 500 CEO and every significant world leader. They provide access to talent that is a hundredfold, with a network of thousands of top engineers, executives, and operators globally ready to join when your company needs them. And they are everywhere—meaning anywhere for the most ambitious founders.

Meanwhile, for LPs, scaled venture firms are also an exceptionally attractive product on the most important simple question: Are the companies driving the most returns choosing them? The answer is simple—yes. All the big companies are working with scaled platforms, often at the earliest stages. Scaled venture firms have more swings at bat to capture those important companies and have more ammunition to persuade them to take their investments. This is reflected in the returns.

Excerpted from Packy's work: https://www.a16z.news/p/the-power-brokers

Consider where we are right now. Eight of the ten largest companies globally are VC-backed and headquartered on the West Coast. Over the past few years, these companies have accounted for much of the global new enterprise value growth. Meanwhile, the fastest-growing private companies globally are predominantly VC-backed West Coast companies: those companies born a few years ago are rapidly heading towards trillion-dollar valuations and the largest IPOs in history. The best companies are winning more than ever, and they all have the support of scaled institutions. Of course, not every scaled institution performs well—I can think of some epic collapses—but almost every great tech company is backed by a scaled institution.

Either scale up or specialize

I don’t think the future is only for scaled venture firms. Just like all areas touched by the internet, venture capital will turn into a “barbell”: one end will feature a few ultra-large players, while the other end will see many small, specialized firms, each operating in specific domains and networks, often collaborating with scaled venture firms.

What is happening in venture capital is what typically occurs when software eats the services industry. One end has four or five large, powerful players, often vertically integrated service firms; the other end contains an extremely differentiated long tail of small providers that have emerged thanks to the industry being “disrupted.” Both ends of the barbell will thrive: their strategies are complementary and empower each other. We also support hundreds of boutique fund managers outside the firm and will continue to support and work closely with them.

Both scaled and boutique will do well, it's the firms in the middle that will have trouble: those whose fund sizes are too large to bear the cost of missing giant winners, yet too small to compete with larger institutions that can provide better products structurally to founders. What is unique about a16z is that it straddles both ends of the barbell—it is both a set of specialized boutique firms and also benefits from a scaled platform team.

The institutions that can best work with founders will win. This may mean enormous reserve capital, unprecedented reach, or a vast complementary service platform. Or it may mean irreplaceable expertise, excellent consulting services, or sheer unbelievable risk tolerance.

There’s an old joke in the venture capital world: VCs believe every product can be improved, every great technology can be scaled, every industry can be disrupted—except their own.

In fact, many VCs dislike the existence of scaled venture firms altogether. They believe that scaling has sacrificed some soul. Some say Silicon Valley has become too commercial and is no longer a haven for misfits. (Anyone claiming there aren't enough misfits in tech surely hasn't attended a tech party in San Francisco or listened to the MOTS podcast.) Others resort to a self-serving narrative—that change is “disrespectful to the game”—while ignoring that the game has always served founders and always will. Of course, they would never express the same concerns about the companies they support, whose existence is predicated on achieving substantial scale and changing the rules of their respective industries.

To claim that scaled venture firms are not “real venture capital” is like claiming that NBA teams shooting more three-pointers are not playing “real basketball.” You may not think so, but the old rules no longer dominate. The world has changed, and a new model has emerged. Ironically, the way the rules change here mirrors the way the startups that VCs support change the rules of their industries. When technology disrupts an industry and a cohort of new scaled players emerges, something is always lost in the process. But more is gained. Venture capitalists intimately understand this trade-off—they have always supported this trade-off. The disruption that venture capitalists wish to see in startups should also apply to venture capital itself. Software has eaten the world, and it certainly won't stop at VC.