Navin Chaddha runs Mayfield as managing partner. It is one of the original Silicon Valley venture firms, and he has invested through five technology waves, from cloud to mobile to AI infrastructure. Before he was an investor he was a builder. He founded three companies. One went to IPO. One sold to Microsoft.

Mayfield backs the jockey, not the racetrack. It calls itself a people-first firm, writes early, and stays small on purpose. Around $3 billion under management, roughly 15 investments a year, more than 500 companies and 125 IPOs since inception.

This conversation is about where the value goes as AI rewrites how companies get built. Chaddha thinks the money is crowding into hardware, models, and cloud, while the real opening sits where software never reached. He makes the case for AI sold as a service, for physical AI in hard industries like manufacturing and logistics, and for one discipline founders keep forgetting. Company building is a marathon, not a sprint.

The conversation

Jay: You have invested through so many cycles. What is the market overvaluing today, and where is the still-undiscovered opportunity?

Navin: All the technology waves you mentioned were a 10x disruption. AI is a 100x opportunity. If you go from mainframe to client-server to PC to internet to mobile to cloud, either the front end changes or the back end changes. For the first time in history, machines can understand human language. Until now, humans have been slaves of the machine. You had to learn how to program them. Instead of 30 million developers being able to program machines, more than 7 billion people can do it in the language they already speak. Intelligence is being offered as a service. When that happens, opportunities are abundant. Value has accrued to hardware, to models, and to the clouds that provide this as a service. Knowledge worker spend is $30 trillion. Another $30 trillion is spent on the physical world. Some portion of both becomes available to AI.

Jay: Multi-billion-dollar businesses are still being built at the app layer. Where is the caution you would add?

Navin: On data center spend, six or seven companies are going to spend $600 to $700 billion. That is the entire size of the enterprise software market. Enterprise apps made humans more efficient. For the first time, I can go to AI and say, do the work for me. Where is that showing up? It is showing up where software was never even used. In China, India, and Africa, landlines penetrated maybe one or two percent over 30 years. Then wireless came and leapfrogged by 100x. The same thing is going to happen with AI agents. Smart entrepreneurs will go after markets that had a smaller TAM for software, markets software never penetrated. The opportunity is not to make existing software better. The opportunity is to use AI to imagine what software could not even do and go after greenfield.

Jay: These same companies are building faster with fewer employees. Is this a Linux moment, where opportunity gets built as tooling around an open source protocol?

Navin: [The open-source agent framework the episode discusses, caption-rendered as OpenClaw, verify] is the Linux of agents. It is a framework where agents get built. It is very hard to install and full of bugs, but it opens up an ecosystem. The same thing happened with Linux. Free, open source. First came Red Hat, which owned distribution. Then came the cloud providers. Where is the opportunity? First, security. Once you start using it, it gets access to your apps and your proprietary data, and it can go rogue. You need security to observe what the thing is doing, and then policy and governance. Second, the infrastructure layer. There will be agents talking to humans and agents talking to hundreds of other agents. How do you orchestrate them, meter them, do billing and observability? A new stack of companies gets created. And finally, new agents get created that do not need to own the UI layer. They show up as skills. Skills are the new apps.

Jay: We are already seeing skill-share communities. Sometimes people put them behind a paywall.

Navin: Correct. The massive companies will get created in security and in the plumbing, the picks and shovels. Then in vertical, domain-specific areas, new agents will interface with that co-work layer. There may not be a thousand UIs we interact with. It might be only a couple, and everything sits behind them.

Jay: You have built companies as well. What lessons would younger founders benefit from?

Navin: Company building is a marathon, not a sprint. You do not want to get too far ahead on your valuation and not have a durable business. The difference between now and the late 90s is that public markets are cheap. The Mag 7 trade on earnings, 20 to 25 times earnings. In the dot-com era, hundreds of pre-revenue companies went public with no business model, valued on eyeballs. So balance top-line growth with what your bottom line looks like. Rely less on borrowed money. My second company went public in 18 months, went from zero to $50 million in revenue and zero to $3 billion in market cap. Worst possible timing, May 2000. It crashed from $3 billion to $1 billion, then acquired for a few hundred million. Timing is everything, and you cannot control it. So capitalize your companies right.

Jay: A founder feels like it is an arms race. Being the second player is never as good as being the first.

Navin: I learned that the hard way at Microsoft, from Bill Gates. In any ecosystem, the number one company makes the most money and the most profit. Number two maybe breaks even. Over time there is no number three. But let's cover the myth. I am not saying you need capital efficiency in the early days. When you start with an idea, hire the right team and get product-market fit. Once you have it, raise enough to prove a repeatable go-to-market. At that point you put money in. What I am seeing instead is seed rounds of $100 million, $500 million, when you do not even have a team. Be stage appropriate. How many unicorns are stuck now, lucky to even raise at the same price?

Jay: It is tough for employees too. They take the early risk but the company is priced so high they are not compensated for it.

Navin: It is a huge problem. It is paper money. My point is, it is a chess game. You do not run the fastest at the beginning. You decide when to do it, and that is an art. The most you learn is from failure, so learn from the failures of others rather than your own. Surround yourself with excellence.

Jay: When founders come back a second or third time, what makes you back the jockey again?

Navin: The number one thing is that they feel they have barely scratched the surface of their potential. They are mission-oriented. They have a chip on their shoulder. There is something beyond dollars. Mayfield is a people-first firm. We back the jockey, not the racetrack. We are there at the inception stage when they turn on the switch, and we are the last ones to go, whether it works or not. Poshmark is a good example. For 18 months after the IPO we were the only investor. We did not sell a share. People remember who stood with them on the down.

Jay: Are there things a first-time founder can learn to stay even-keeled?

Navin: Surround yourself with experience. Good advisors, stable investors, good board members. As Einstein said, if failure is not ahead of you, you are not trying hard enough. I tell founders, do not shoot for the roof, shoot for the moon. And in doing that, you will fall. What matters is how you regroup, take a timeout, and figure out the next move. You are not going to zero. It is a speed bump.

Jay: Mayfield has been around 56 years. How do you make the case to this new generation of young founders?

Navin: We do not do 100 investments a year. We are in the rifle-shot business. Managing $3 billion, we do only 15 or 16 investments a year. We are not going after market share. Apple never did that. Mayfield has funded over 500 companies, 125 IPOs, 225 acquisitions. The vast majority were first-time founders. Serial entrepreneurs are only 5 to 10 percent of our success. First-time founders used to be in their mid-30s. Now it is mid-20s, even 18, 19, 20. So we look for founder-Mayfield fit. I call it FMF, not PMF.

Jay: Jagdeep at Rhoda AI is a good example. What got you excited?

Navin: Company building is a team sport. It is Jagdeep plus his co-founders. Second is the market. Physical AI is a $30 trillion opportunity. They are building a new model, but they are not depending on OEMing it to others. They package it for domain-specific problems and provide robotics as a service. In manufacturing, auto logistics, returns handling, they go after work where there is a shortage of talent or jobs humans do not want to do. They become part of the physical workforce where there is a gap.

Jay: These industries have been promised a lot and delivered very little. What advice do you have on balancing the narrative with what the customer actually gets?

Navin: Under promise and over deliver. Reality cannot run ahead of a customer's experience relative to your vision. In the old model, the robotics industry got paid up front and you figured out how to use it. The new model that companies like Rhoda are pioneering is physical intelligence as a service. You pay for outcomes. I am not spending hundreds of millions of capex up front on something that might not work. The vendor takes the risk. All the risk transfers to the vendor. So you need conviction that your product is good enough to move to that model.

Jay: That changes the economics for the founder.

Navin: Dilution is an EQ test. As a founder I can own 100 percent of a company worth $10 million, or own 20 percent of a company worth $100 billion. Which is better? But again, raise the money and do not spend it. Build the product, get early design partners, get product-market fit, and scale. Once you have proven ROI, the money you need is cheap and available. And for robotics as a service, there are plenty of debt vendors who will fund it.

Jay: More seed-stage companies tell me their go-to-market is a forward-deployed engineer. Are we comfortable with founders being at services margin, not software margin?

Navin: Even enterprise software taught us this. For every dollar spent on perpetual licenses, $8 to $10 was spent on services. AI is not standalone in an enterprise. You have to integrate with the data and the workflows. But here is the difference that keeps margins closer to software. For every one FTE, there might be 20, 30, 40 agents behind them. You do not want 100 people in seats. You want one, two, three, with 10 to 20x agents behind them.

Jay: So would you be long Accenture?

Navin: Accenture is going to shrink. The big services industry is going to shrink. It becomes a hybrid model, an AI-native services company. Very few humans, most of the work done by agents, and the business model is outcome-based. Most legacy services companies will not make the transition. That always happens, and that is the opportunity. It is a multi-trillion-dollar opportunity.

Jay: Exit value is becoming more and more power law. How does that factor into the swings you back?

Navin: History shows it has always been a power law. Over 50 years, 10 to 15 percent of the companies in every fund drive 80 to 90 percent of the returns. The problem is that fund sizes have gotten so big that for the power law to work, the exits have to be 8 to 10 times bigger and you have to own enough. Maybe it gets worse. Maybe 1 percent of companies drive 80 percent of returns.

Jay: Do the chickens come home to roost at some point?

Navin: It depends on your product. We are an early-stage venture fund that leads the first institutional round. We keep our fund sizes small because we are driven by outperforming the top quartile, even the top decile, again and again. Then there is late-stage venture, an IRR game. And now there is a third category, the capital platform, competing with private equity and buyout. The venture industry will go from 1,000 firms to maybe 100. It will not go to five. It is like choosing between a single name, Apple or Nvidia, and QQQ.

Jay: Say we are having this conversation in 2030. What does a company that shows up to Mayfield look like?

Navin: Anybody who thinks they know where the world will be four or five years out, I do not know what they are smoking. But some principles hold. You need your own north star, grounded in mission, vision, and values. Now AI is part of your workforce, so define the role your AI teammates play. Startups die of indigestion, not starvation. Have a culture of nimbleness, because dinosaurs do not survive. Be people-first. And there is no overnight success. Company building is a marathon, not a sprint. It will be AI teammates and humans working together, and the people who figure that out are the big winners.

Jay: Thank you so much for sharing your wisdom, and for all the support and guidance you have given us.

Navin: Wisdom can be overrated. In my case, I am still learning. The VCs and entrepreneurs who succeed will have a prepared mind but also keep an open mind to listen, pivot, and make the right decision. The people who do that are the next winners.

Pull quotes

1. "For the first time in history, machines can understand human language. Until now, humans have been slaves of the machine."

2. "The opportunity is not to make existing software better. The opportunity is to use AI to imagine what software couldn't even do."

3. "We back the jockey, not the racetrack. We're there at the inception stage when they turn on the switch, and we're the last ones to go."

4. "Startups die of indigestion, not starvation."

5. "Physical intelligence as a service. You pay for outcomes. All the risk transfers to the vendor."

Source

From CLIMB Episode 096 with Navin Chaddha (Mayfield). Transcript cleaned from the published episode. Watch the full episode: https://youtu.be/ypUc0V6J9WI