The article elaborates on the transformation of a16z’s investment philosophy from “software devouring the world” to supporting “small technology”, as well as the changing role in the venture capital industry, emphasizing the importance of startups.
In 2009, Marc Andreessen and Ben Horowitz founded Andreessen Horowitz (a16z) at the deepest trough of the financial crisis, bucking the trend at a time when Silicon Valley was panicking and capital was receding. That year, only two new venture capital funds were born in the United States, one was initiated by legendary investor Ron Conway, and the other was them.
At that time, no one could have predicted that this fledgling foundation would grow into a synonym for “venture capital branding”, “full-stack services” and “policy games” more than a decade later, and no one realized that a16z would not only be at the forefront of betting on technological change, but would gradually move to the center of the stage – speaking out in Washington, setting up points on the European policy battlefield, and setting new boundaries in the global venture capital order.
From “software is eating the world” to “VCs themselves should be productized”, Marc Andreessen is revising the script of the industry himself. He constantly reminds people that most of the companies that eventually become the dominant force in the industry have faced the fork in the road of “almost not it”; And those forces that can ultimately shape the future order often come from the margins, from doubts, from “small technologies” that were misunderstood or even ignored at the beginning.
For Marc Andreessen, entrepreneurship is never a straight upward path, but more like “falling up the stairs” – new difficulties follow at the point where it seems to solve the problem again and again. Almost every company that has grown into a world-class company has almost taken a key turn to another path.
The birth of a16z not only started in the cold winter of capital, but also at the end of a decade-long process of disillusionment and recovery of the Internet bubble. Andreessen traced back history and said that between 1995 and 2000, the dot-com bubble continued to expand and finally collapsed in 2000, and it took three to four years for the market to gradually recover, and it was not until 2003~2004 that it improved slightly.
Even so, society as a whole is still suspicious of emerging technology companies. In 2005, Yahoo bought Flickr and Delicious for $25 million each, but was described by the media as a “waste of public money”; When social networks first emerged, they were generally regarded as jokes, and almost no one believed that content like “what did your cat eat for breakfast” could become serious business.
How can product managers do a good job in B-end digitalization?
All walks of life have taken advantage of the ride-hail of digital transformation and achieved the rapid development of the industry. Since B-end products are products that provide services for enterprises, how should enterprises ride the digital ride?
View details >
This skepticism continued until after the 2008 financial crisis, when the venture capital industry came to a near halt. Marc recalled that during that whole year, there were only two new venture capital funds in the United States: one was launched by legendary investor Ron Conway, and the other was a16z. Compared to Conway, a16z is almost entirely a “grassroots” fund that started from scratch.
Marc said that he and Ben did not become monks halfway, but took the initiative to choose a new stage after starting a business and completing multiple cycles. They started forming teams, visiting LPs, building relationships, and everything started from scratch.
Miscalculations and reversals in the mobile era
As a longtime member of Facebook’s board of directors, Andreessen also reflects on the outside world skepticism faced by Facebook in the early days. He pointed out that at that time, the public was not worried about the “danger” of technology, but generally thought that they were “useless”. Facebook was seen as a joke, and the company was initially supported almost entirely by “leftover ads” in Microsoft’s Bing ad network, with very low CPMs and little monetization.
But that doesn’t mean Mark Zuckerberg lacks monetization. In fact, there were several early attempts to acquire Facebook, the most notable of which occurred around 2006. At that time, Yahoo, led by CEO Terry Semel, offered to complete the acquisition for $1 billion. Marc said that the agreement was largely verbal, but the sudden financial crisis changed everything.
Against the backdrop of Yahoo’s rapidly shrinking advertising market, Mark found a reason to exit the deal. Andreessen believes it was supposed to be a “home run-level” deal, but years later, Yahoo’s internal acquisition analysis was leaked, showing that they were actually very bullish on Facebook’s potential, but seriously underestimated its future growth space.
In 2012, Facebook went public, and once again faced a decline from the outside world. Marc recalls that at the time, the iPhone was already mainstream, but the media almost unanimously agreed that Facebook would fail because it “couldn’t adapt to mobile advertising.” The logic seems to hold – smaller screens, limited advertising space, and inevitable decline in revenue.
The reality is the complete opposite. Facebook’s mobile user stickiness and usage time are longer, and its ad targeting ability has been greatly improved due to its rich user data. Marc emphasizes that while there is now consensus, almost no one believed that moving ads from desktop to mobile would be successful and less likely to become more effective.
Ironically, just a few years later, Facebook ads went from being “ineffective” to “mind-controlling.” Marc noted that political advertising came into focus, especially after the 2016 US election. At one point, the media accused Russia of influencing a $3 billion campaign on Facebook for just $80,000.
He recalled attending a speech by Hillary Clinton at Stanford, where he publicly declared, “Trump can be president because Putin owns Facebook.” Marc bluntly said that this statement sounds extremely absurd.
He asked: If these ads are really so effective, why not rely on them to sell toothpaste? Why is it useful for only one election? Why aren’t businesses reusing the same “magic”?
He further pointed out that the “psychological profiling method” promoted by organizations such as Cambridge Analytica did not ultimately prove to be effective. These methods are deified at a specific point in time and then quickly abandoned, leaving only a short-term overreaction of public opinion.
At the beginning of its inception, Marc Andreessen admitted that they did not have a clear expectation for the size of the future fund. The first fund raised $300 million, which was rare at the time. But Andreessen is more concerned with the structural trends emerging in his time: the number of software companies is increasing and growing at an astonishing rate.
As early as the mid-90s, he had doubts about the rotation structure of traditional venture capital in an exchange with KP’s John Doerr. At that time, the mainstream investment path was to go public after three rounds of A, B and C, with an overall financing amount of between 30 million and 40 million US dollars. The listing threshold is typically $50 million in annual revenue, corresponding to a valuation of about $500 million.
Take Amazon and Netscape as examples, both of which had a market capitalization of $4~$500 million when they went public. But Andreessen notes that some businesses clearly have the potential to go beyond this “ceiling.” The discussion with John at the time was based on Cisco, who had realized that there could be tech companies with valuations of more than $100 billion in the future. And this was an astonishing ceiling at the time.
So the question arises: if an investor invests $50 million in Series C and gets the same return as $5 million in Series A, should the logic of venture capital also shift from “round participation” to “continuous betting on the endgame”? This judgment was later validated by the development of companies such as Facebook. DST’s high-valuation investment in Facebook has opened a new paradigm of “growth venture capital” and prompted a16z to establish a “stage-agnostic” investment philosophy early in its creation.
In their initial fundraising materials, they stated that whether it is a seed round or a growth round, the key is to bet on the winning company. Instead of sticking to the label of the investment stage.
Operating model: generalist VS vertical
The team at a16z in the early days had a generalist structure, with Marc and Ben working on both ToC and ToB projects. At that time, mainstream venture capital was still operating between the two categories of consumer and enterprise software, and some established institutions were also involved in biotech. However, as the differentiation between biomedicine and IT intensified, many funds were forced to split organizationally, and technology and life science projects gradually formed independent teams.
Marc admits that initially they followed Benchmark’s generalist partner model. But soon after, the evolution of the tech industry itself forced them to make major adjustments.
Since 2009, so-called “full-stack companies” have begun to emerge. These companies not only provide the underlying technology, but also penetrate into the end market, trying to completely rewrite the way an industry operates. Uber, Airbnb, Tesla, SpaceX are typical examples – no longer just tools, but reinventing the entire value chain.
Marc pointed out that the generalist system was established in the 90s because most of the technology companies at that time focused on tool-based products, such as databases, operating systems, productivity software, etc. These companies have similar organizational forms and similar market models, and the complexity of investment judgment is relatively controllable.
But the rise of full-stack enterprises has changed everything. They are deeply rooted in the real economy, the market pattern is changeable, and factors such as technical architecture, regulatory compliance, user acquisition, and capital expenditure are intertwined, making it difficult for generalist teams to accurately judge “who will win”.
More importantly, the investment structure of venture capital does not allow for “all-in and restart” – if you bet on the wrong company in a vertical industry, even if you see the industry trend right, you will not be able to invest in other competitors. Therefore, judging “who wins and who loses” is far more critical than judging “which direction is fire”.
This prompted a16z to start implementing “vertical operations” in 2013 and basically complete the full transformation in 2017. Today, they divide funds according to industries and set up special teams to gradually create a compound investment platform covering enterprise software, encryption, games, biotechnology, financial technology, defense technology and other directions.
Talking about the change in the founder’s background, Marc mentioned that one of the reasons for the success of Y Combinator more than a decade ago was the information advantage of the “technology founders”. At that stage, as long as you master engineering capabilities, you can quickly complete prototypes and launch products, and the business model can be supplemented later.
The dividends of this “engineering arbitrage” reached their peak after the maturity of mobile Internet and basic cloud computing. Now, with the sinking of AI technology, the “cognitive threshold” that founders rely on is changing.
Marc believes that AI is lowering the technical threshold, code generation and model calls are becoming more and more accessible, and in some scenarios, industry knowledge and distribution channels have become more scarce resources. This means that the next “cognitive arbitrage” opportunity may come from expert founders who are well versed in the logic of a certain industry and have real upstream and downstream networks.
However, he also emphasized that at this stage, technical capabilities are still the core of success or failure. Even companies that emphasize design-driven are almost all deeply technocratic behind their founders – but they don’t take the initiative to promote this side. For example, Ben Silbermann, the founder of Pinterest, is considered to be focused on product and visual design, but he actually has a strong technical background.
Marc is open to the possibility of AI changing the founder’s persona. He proposed a thinking experiment: If a founder with a non-technical background could call 1,000 AI engineering agents in the future, would he be able to beat a company led by a 100-person technical team?
At present, the capabilities of low-code platforms are still limited to rapid prototyping and lightweight service development, and truly building a world-class software platform still requires the deep involvement of technical teams. However, if AI tools are intelligent enough and can be combined in agent mode, the organizational structure may usher in a fundamental restructuring.
At that time, the criterion for judging “who can win” may no longer be “who knows the most technology” but “who can mobilize the most AI and complete tasks efficiently”.
In Andreessen’s view, this is not a fantasy, but an inevitable trend as AI enters the middle engineering and organizational intelligence fields.
Andreessen concludes that almost every company that is now called a great company has been at a fork in the road in its early days. He mentioned that Netscape had considered buying Yahoo, Yahoo almost bought Google, Netflix had tried to sell to Blockbuster, and Uber and Lyft were close to success in merger negotiations.
In his view, certain technological trends are destined to happen, such as the rise of smartphones and the emergence of killer applications on them, but which company really dominates it and through what path it takes depends heavily on individual decision-making, market conditions and the contingency of time nodes. These paths are extremely sensitive to initial conditions, and if they differ slightly, history will be completely rewritten.
The formation of the “small technology” position
Marc mentioned that a16z initially had no plans to get involved in the policy area. In initial contact with Washington, D.C., they originally only wanted to speak on behalf of startups, but soon discovered that the government often confused them with big tech companies like Google, Meta, and others.
“As soon as we opened our mouths, they started complaining about Google,” Marc recalls, “and we had to constantly explain that we were not Google, that we were voting for companies that challenged Google. This confusion has led them to gradually develop a new identity framework – to distinguish between “little tech” and “big tech”.
This framework is not only a formulation strategy, but also gradually evolves into a policy initiative. “We are increasingly discovering that there is actually a fundamental conflict of interest between venture capital and big technology.” He said that big tech tends to build barriers, build moats, and try to exclude new entrants, and the essence of venture capital is to support innovators who try to break these monopolies.
The conflict also prompted a16z to win more policy sympathy in Washington. A rare cross-party understanding was gained when they conveyed to both parties that “we are not speaking for big corporations, but on behalf of challengers, disruptors.” “At that moment, they breathed a sigh of relief and showed the expression of ‘so you are good people’.” “They don’t trust big tech, but they believe in startups. ”
“Small technology” is the identity, and the “agenda” is the program of action. The “Little Tech Agenda” launched by a16z is not an anti-regulatory manifesto, but an advocacy for a clear, transparent policy framework.
Marc emphasized that they are pursuing “freedom to innovate”, especially in emerging fields such as AI, encryption, and network communications, and hopes that policymakers will allow startups to explore boundaries rather than blindly block them when the law is not yet clear.
He made it clear that a16z is not against regulation, let alone fraud. “We don’t want our company to compete with players who are not compliant but have no one to regulate.” Instead, they have been more active in calling for rules to ensure that compliers are not kicked out of the market by bad money.
They are not advocating “unlimited freedom” but “clear boundaries”. Marc concluded: “We speak out for regulation as often as anyone else. ”
In his review of the relationship between politics and technology in the United States over the past decade, Marc mentions a series of key turning points:
- In 2012~2013, the political environment tended to be polarized, and anti-technology sentiment began to be intertwined with the wave of anti-elite and anti-capitalism.
- In 2015~2016, Trump’s election sparked different bipartisan attacks on the tech industry: the left was angry about income inequality and platform influence, and the right was hostile to Silicon Valley’s political tendencies.
- In 2017~2020, social media became the main focus of contradictions, followed by a full-scale crackdown on the crypto industry.
- Since 2021, AI technology has been on the cusp, and the government’s desire to control it has increased.
Marc admits that they are not naturally willing to get involved in these fields, but reality forces them to intervene. “If we don’t speak up, those who hate us will create an image for us.”
