a16z founder's Stanford lecture: Whenever Wall Street and Silicon Valley have different ideas, it's Wall Street that ends up being wrong

In the spring semester of 2026, Stanford launched a course called CS 153: Frontier Systems. With 500 seats and a waiting list, each week a global leader from the tech industry was invited to lecture, featuring names like Jensen Huang, Sam Altman, Lisa Su, and Andrej Karpathy. The course was co-taught by Anjney Midha, founder of AMP PBC, and former Apple executive Mike Abbott. Anjney previously worked as a partner at a16z, leading investments in AI companies like Mistral and Black Forest Labs, and managed a16z's GPU cluster project Oxygen before founding AMP PBC, focusing on providing computing power and capital for AI teams.

This session's guest was Ben Horowitz, co-founder of Andreessen Horowitz (a16z). Before founding a16z, Ben was the co-founder and CEO of Opsware, a company that developed data center automation management software, which was sold to HP for $1.6 billion in 2007. In January 2026, a16z completed a new fundraising round of over $15 billion, accounting for more than 18% of the total venture capital in the U.S. in 2025. His two bestselling books, "The Hard Thing About Hard Things" and "What You Do Is Who You Are," have given him rare credibility among entrepreneurs. Recently, he introduced a concept at a16z's Connect/Fintech conference called "founders' AI anxiety": in the past, a strong software product could lead competitors by ten years; then it shrank to five years, and now it might only be five weeks.

This feeling of "never being able to keep up" is consuming the founder community in Silicon Valley, sparking widespread discussion.

Anjney chose to start with a song. The sound system played the 1985 charity single "We Are the World." The producer of this song, Quincy Jones, passed away in November 2024 at the age of 91. The Netflix documentary "The Greatest Night in Pop" recorded a detail from the recording process: before starting the recording at midnight, Jones posted a handwritten note on the studio door frame that read, "Leave your ego at the door."

Jones's most remarkable ability throughout his life was to bring a group of super egotistical geniuses into one room and have them collaborate to create great works. Anjney said that if he had to summarize Ben Horowitz in one sentence, he would call him the Quincy Jones of the tech industry. A quick note for those interested in the reader group: please add: rohanjojo.

1. From $300 million to $15 billion: Venture Capital is a System Design Problem

Ben started from the founding of a16z. When he and Marc Andreessen launched their first fund of $300 million in 2009, there were two widely accepted beliefs in the venture capital industry.

The first was the issue of product definition. At that time, venture capital was essentially a business for making money for investors, where LPs received high returns, but entrepreneurs received almost no practical help beyond a check. Ben felt that the products offered to entrepreneurs were too poor and needed to be redesigned.

The second was the assumption of market capacity. Historical data supported a "15 company hypothesis": in any given year, there are only about 15 tech companies worldwide that can reach $100 million in revenue. The entire venture capital industry operated around competing for these 15 spots. David Swensen of the Yale Endowment Fund famously said: good venture capital is like a basketball team; five or six people are enough.

Ben believed that both of these assumptions were about to become outdated. If software was truly going to "eat the world," then every interesting new company would be a tech company, and the number of companies reaching $100 million in revenue each year should be close to 200. The basketball team-sized venture capital simply could not handle this scale.

The subsequent organizational design decision established a fundamental difference between a16z and all traditional venture capital firms.

Traditional venture capital is partnership-based, where partners share economic returns and control. This works well on a small scale, but once transformation is needed, shared control becomes a deadlock. Any restructuring is a redistribution of power, and there will always be opposition. If everyone has a vote, the organization can never effectively adjust.

a16z's approach is: economic interests can be shared, but control must be concentrated in one person's hands. This allows the company to continuously enter new fields, such as American Dynamism focusing on defense and infrastructure, cryptocurrency, and biotechnology, because it can change its organizational structure at any time without needing everyone's consent.

Ben also imposed a strict rule on the investment decision-making process: the number of people in the room cannot be so large that dialogue becomes impossible. A room of 30 people is not a dialogue; it’s a speech. In his view, truth-seeking requires high-fidelity dialogue, and even at its best, high-fidelity dialogue cannot sustain more than seven people. a16z's solution is to continuously break the company into smaller teams, each responsible for a specific part of the market.

2. "Everyone Thought We Were Crazy"

Having a system design is not enough; it must also be proven that this system can win.

For the first fund of $300 million, Ben invested a quarter in the acquisition of Skype. At the time, everyone thought this was a suicidal move. The reason was specific: eBay owned Skype but did not own its core IP. The underlying communication protocol library was controlled by the two founders, Janus Friis and Niklas Zennström; without this protocol library, Skype could not operate. These two could sue eBay at any time and shut down the service. The consensus in the market was that this was an unacquirable asset.

But Ben knew these two founders. He understood that Skype was the most important label in their lives, and they would not actually destroy it. The only question was price and terms. After the deal was made, the LPs who had previously said a16z was crazy subtly softened their stance.

To truly change the views of institutional investors, Ben said there was only one way: "Win. I say one thing, you say another, and then we wait to see who is right."

3. Zero Salary, Donuts, and a HP Phone Number

The Skype deal solved the credibility issue, but a16z faced a bigger problem: how to build a network effect that had never existed in the venture capital industry from scratch?

a16z's experiences investing in companies like Facebook, Twitter, and Skype gave the team firsthand understanding of network effects. Ben repeatedly told the team a simple math: the value of a network is roughly proportional to the square of the number of nodes. Five nodes have a value of 25; six nodes have a value of 36. Once you reach the scale of the internet, no one can build a competitor.

He decided to apply the same logic to a16z itself. If a16z's network is large enough and dense enough, entrepreneurs raising funds here can immediately tap into the resources of the entire Silicon Valley, which is a barrier that other venture capital firms cannot replicate.

The cold start strategy was blunt and direct: partners did not pay themselves salaries and invested all management fees into network building. They hired people to actively build relationships, aiming to meet every engineer, executive, and large company that procures tech products in Silicon Valley.

In terms of execution, Ben leveraged an asymmetric advantage that others did not have. His previous company, Opsware, was sold to HP, so he knew people in HP's corporate briefing center. The briefing center is where large tech companies host client executives, with different companies' CIOs and CTOs visiting each week to see demonstrations of the latest technologies, aimed at closing big deals. Ben's team called every week asking, "Who is coming this week? Can we get their contact information?" After obtaining the list, they invited these large companies to a16z to see startup demonstrations, preparing donuts and various refreshments.

The result was that a16z understood large enterprise clients better than many established venture capital firms that had been around for 50 years. Anjney mentioned that when he was working at another top venture capital firm, Kleiner Perkins, he showed a16z's news coverage to their CMO and suggested they learn from it. The response was, "Oh, that's just marketing."

Ben said that the hatred from competitors became a barrier instead. He wrote a blog post titled "Four Things I Don't Like About Venture Capital," directly criticizing the entire industry. In an interview with Sarah Lacy, he quoted lyrics from Lil Wayne: "When I see another VC throw me a peace sign, I only see the trigger and the middle finger."

The entire venture capital circle hated him. Other VCs shortened Andreessen Horowitz to "Aho," which sounds similar to a vulgar term in English, as a clear derogatory term. Every time a16z's LP met with other venture capital firms, they would speak ill of a16z. Ben's reaction upon hearing this news was, "Great."

Because they hated a16z so much, they were unwilling to replicate what a16z was doing. Hatred itself became a competitive barrier. Ben admitted he was unsure if he would be so aggressive again but acknowledged that the outcome was effective.

4. "You Can't Solve Problems by Throwing Money at Them" — This Rule Has Been Broken by AI

When the conversation shifted to the AI era, Ben's tone changed. He used a term: the physical laws have changed.

He said that from 2009 until the emergence of AI, there was one absolutely reliable rule throughout his career: you can't solve problems by throwing money at them. If a competitor is two years ahead of you, hiring a thousand engineers won't catch you up because some work cannot be parallelized, and communication overhead will drag you down. His favorite joke was: how much is a man year? It's the work done by 700 IBM employees before lunch. The implication is that when the number of people reaches a certain level, actual output approaches zero.

AI has broken this rule. If you have enough GPUs and data, you can indeed throw money at problems. The capital race has turned into a real dimension of competition.

This brings a direct consequence: in the past, software companies had two natural moats: one was code, which others could not replicate; the other was user interface, which others could not make as user-friendly. AI has erased both of these. Code can be generated, UI can be copied, and the product gap that used to take two years to catch up can now potentially be closed in five weeks. So what is your barrier? This is the question Ben believes every entrepreneur in the AI era must answer.

At the same time, demand is infinite. Ben said that no previous software was usable to this extent. Think about the past when buying a suite of Siebel Systems' customer relationship management software took two years and started at over a million dollars for deployment, which limited demand. In contrast, after using AI products, users' first reaction is, "How can I use this more?" When technology is good to this extent, there is no ceiling on demand.

5. What Does a Real Moat Look Like?

Because of the disappearance of these two moats, a narrative has become popular on Wall Street: AI model companies will wipe out all SaaS companies. Ben thinks this is a typical Wall Street error, with judgments that are too extreme. The old moats have disappeared, but that does not mean software companies have no moats.

"Whenever Wall Street thinks one thing while Silicon Valley thinks another, the arbitrage opportunity is immense, and Wall Street is always wrong."

In Ben's view, the real moat in the AI era is not in the software itself, but outside of it: global supply chain relationships, vertical industry channel capabilities, and deep integration with customer systems. These things cannot be generated or captured by AI.

He cited Navan, a corporate travel management platform where he serves on the board, as an example. Navan went public on NASDAQ in October 2025. To do this business, you need to establish supply chain relationships with every airline and hotel globally; if you scrape their websites, they will sue you directly. You need to integrate with the internal systems of client companies. Your sales target is the extremely vertical role of "travel management manager."

Ben mentioned a detail: the Anthropic that Wall Street claims will "wipe out all SaaS" is currently hiring a travel management manager specifically to interface with Navan. If even they want to be a customer of Navan, how could they possibly create a travel management product? When there are gold bricks everywhere, no one bends down to pick up silver bricks.

He quoted Buffett: in the short term, the stock market is a voting machine; in the long term, it is a weighing machine. Currently, SaaS companies are collectively undervalued because the "SaaS apocalypse" is too good a story, and fund managers holding SaaS stocks have been fired, with newcomers not wanting to touch this field. But when quarterly reports continue to perform well, the weighing machine will eventually correct the voting machine's errors.

Ben also told a story to illustrate not to easily declare a company dead. Stewart Butterfield once created an iPad game called Glitch that ran on Flash, but then Steve Jobs announced that the iPad would ban Flash. The company nearly died, with only $6 million left. Butterfield transformed it into Slack.

"In business, there is only one unforgivable sin: burning through all the money. As long as you haven't burned through all the money and you are a special founder, I won't write you off."

6. Culture Is What You Do, Not What You Believe

Anjney made an observation: many teams that externally seem "certain to succeed," with star founders, ample capital, and good problems, struggle, split, and disband within 6 to 12 months. This issue is especially severe in AI labs.

Ben believes the reason lies in culture and leadership. He quoted the samurai code: "Culture is not a set of beliefs; it is a set of actions." Then he put it more directly: "I don't care how you think, I don't care about your feelings, I don't care what you have in your heart; I only care about what you do."

What he refers to as "actions" are very specific things: Do you come to the office? What time do you leave? When someone asks you a question, do you reply immediately or a week later? Do you believe the best ideas win, or does the founder get the final say? These must be clearly defined and not vague.

With standards in place, it becomes simple to handle someone who does not meet them. Without standards, you can only be angry, and anger leads to infighting and politics. If someone leaves early, you are unhappy but never stated that you wanted to work overtime, the outcome is mutual dislike, and when the first difficulty arises, you split. "Open AI offers high salaries to poach people, fine, I’m done, I’m going home."

Ben opposes the co-CEO model, the principle of "we are all equal," and completely flat management. A company needs one person to break the deadlock: you want to go left, you want to go right, we are going this way; if you don't accept it, leave. In a competitive environment, one person making the call is always faster than a collective vote. He said Silicon Valley deviated from this principle during the "plentiful era" of network effects, where CEOs compromised with employees and let everyone vote on the company's values. The results were not good.

He also distinguished between companies and nations. If a leader does not seek personal gain for themselves and their cronies, purely pursuing the public good, the efficiency of centralized decision-making may be higher than that of decentralized decision-making. The problem is that once a bad leader takes over, concentrated power becomes a disaster. Therefore, nations must resist bad leaders through decentralized systems because nations need to endure for hundreds of years. Companies do not need to be designed this way. HP declined after its founder passed away because the company had completed its mission. But a nation cannot collapse just because of a change in leadership.

7. "Invest in People, Not PPTs": The Terrible Pitch from Databricks

In the last half hour of the conversation, students voted on Discord to ask about the most memorable fundraising pitch. Ben laughed and said the most unforgettable was Databricks.

Berkeley professor Scott Shenker called him: "I have the best distributed systems expert in academia from the past decade, named Matei Zaharia; do you want to meet?" Ben said that when he heard this, he knew he was going to invest.

However, Zaharia's co-founder, also a Berkeley professor, Ion Stoica, came to do the pitch, and the slides were like a completely incomprehensible computer science lecture. "It felt like you were in a university class that you couldn't understand at all." Ben's partners were scared. But because of that phone call and his judgment of the founders, he still invested.

This was the starting point for Databricks. Ben said this story "scarred my partners," literally leaving a psychological shadow on them. But the outcome proved the judgment was correct.

8. Not Doing Leveraged Buyouts: Saying No Even When There’s Money to Be Made

As a16z managed more and more money, AI made the path of "acquiring traditional companies and improving efficiency with technology" very attractive. Ben received suggestions to open a new business line specifically for AI-driven leveraged buyouts at least a dozen times. Leveraged buyouts, known as LBOs in English, are a classic private equity strategy: buy an old company, use new technology to make it more efficient, and then sell it for a profit. The logic is clear: AI can significantly improve traditional companies, just as spreadsheets gave rise to the entire private equity industry. Many VCs are already doing this.

But Ben refused for two reasons.

First, cultural conflict. The DNA of venture capital is to invest in entrepreneurs and help them grow rapidly. The DNA of leveraged buyouts is to drive down purchase prices, lay off workers, and extract profits. This is a movement in the opposite direction. If you put these two cultures in one organization, the venture capital team thinks about "who has breakthrough ideas" when looking at entrepreneurs, while the leveraged buyout team thinks about "where can we cut people" when looking at companies. Mixing them will tear the organization apart.

Second, life choices. "I have the opportunity to fund great new ideas that push humanity forward. LBOs might be a good business, but it's not what I want to do." He said that great companies do not exist to make money but to do something greater than themselves. If you achieve that, the money will come naturally. You don't need to rush to every place that seems to have money.

9. Six Pieces of Advice for 20-Year-Olds

In the last half hour of the conversation, students voted on Discord to ask questions. Ben's answers covered topics like career choices, AI and employment, and starting a business.

First, understand AI as electricity. Imagine a world before electricity, where you had to go home as soon as it got dark, and your entire life was constrained by physical limitations. AI is the next level of tool revolution like electricity. Master it, and then apply it to the fields you are truly interested in, whether it's biology, materials science, rocket technology, or even creative fields. He specifically mentioned: someone who could only be considered "fairly good" at guitar in his time can now independently produce a sci-fi movie with a soundtrack. The world has really changed.

Second, solve problems, not build companies. The best startup ideas never start with "I want to start a company." Meta was a larger opportunity that Mark Zuckerberg stumbled upon while working on Hot or Not. Dropbox came from Drew Houston's frustration with transferring files via USB. Elon Musk's first project was closer to a Yellow Pages competitor, then it became PayPal, and only later did it lead to Tesla and SpaceX. First, solve a problem that you genuinely have. If you have that problem, it likely means it is real. In the process of solving it, something bigger will emerge on its own. This is similar to serendipity in scientific discovery; penicillin was discovered while doing other experiments.

Third, what does a good idea look like? The world needs it to exist, but if you don't do it, no one will. a16z itself is an example of this standard: the world does not need another ordinary venture capital firm, but it needs a different kind of venture capital. OpenAI is also an example; the AI field does not lack players, and Google is thought to dominate everything, but the world needs an alternative to Google.

Fourth, good ideas are not lacking in funding, but don't think too big in your dorm room. Ben said that currently, capital is unlimited for good ideas. But he also warned about the "dorm problem": students have limited perspectives, and ideas that sound good when chatting with friends late at night should be revisited after at least a night's sleep. Trying to swallow the whole world without experience from the start may help with the pitch deck but won't help the company.

Fifth, don't listen to others' career advice, including mine. "No one can give you good career advice. They can only give themselves good career advice. I can give myself good advice, but I can't give it to you. Especially your friends; they give advice that is useful to them, not to you."

Sixth, software engineering jobs have not disappeared. In response to the popular narrative that "AI will eliminate programmers," Ben directly countered: all data points in the opposite direction. Software engineering positions are still growing rapidly. He specifically pointed out that Anthropic's own engineering recruitment is also growing fast, and Dario Amodei's comments are often taken out of context on social media. What Dario actually said was that during the transition period, some low-skill jobs would be replaced, and workers would need to transition. But the narrative on Twitter amplified this viewpoint into "everyone will be unemployed."

Ben also discussed that what worries him most about AI is not AI itself, but the excessive regulation in the U.S. due to fear, such as pausing data center construction. If either the U.S. or China monopolizes superintelligence, that would be more dangerous than a balance of power between both. Concentrated power has never been a good thing in history. The problems created by fear itself may be more severe than the thing being feared.

10. When Zuckerberg Was 20, "He Really Wasn't That Great"

When discussing team culture and founder growth, Ben shared an unexpected observation. He knew Zuckerberg when he was 20 and frankly said that if Facebook were not a business with network effects, it could not have sustained itself solely based on his abilities at that time. "He really wasn't that great back then."

But Facebook's vertical takeoff gave him the time to grow. The current Zuckerberg is a completely different person from the one at 20; he evolved step by step while operating the company.

The insight from this observation is that different people reach their most effective state at different ages. Some need a business with built-in growth momentum to buy themselves time to grow, while others need to accumulate experience before taking action. Regarding whether one should drop out of school to start a business, Ben's answer is: it depends on who you are, not on others. He himself finished college and felt it was right for him. Zuck dropping out was right for Zuck. "Your friends will give you advice that is useful to them. Don't listen."

Core Q&A

Q1: What is the most critical system design decision that a16z got right? Economic interests are shared, but control is centralized. The partnership model of traditional venture capital gives everyone a vote, leading to an inability to transform the organization. a16z achieves the ability to continuously restructure and enter new fields through centralized control. The "basketball team" model of traditional venture capital is sufficient in a world with only 15 companies worth $10 billion, but in a new world where 200 companies are vying for $100 million in revenue each year, a different organizational structure is necessary.

Q2: What constitutes a real moat in the AI era? Code and user interfaces are no longer moats. The real barriers lie in global supply chain relationships, vertical channel capabilities, and deep integration with customer systems. Ben's judgment standard is: will Anthropic do this? If they, with gold bricks everywhere, won't bend down to pick up your silver bricks, you have a moat. Navan needs to establish supply relationships with every airline and hotel globally, targeting the extremely vertical role of travel management manager. These AI companies won't touch that.

Q3: What is the correct way to initialize team culture? Culture must be specific behavioral standards, not abstract value statements. "The best ideas win" is a standard; "we have integrity" is not. There must be one person with final decision-making authority to break the deadlock; co-CEOs and collective voting will fail in a competitive environment. Once standards are established, they can evolve, but the evolution must be led by decision-makers and completed by the team, not everyone going their own way.

Original link