24:26
I strongly suggest that in the future we will actually need more engineers, not fewer. The goal is not efficiency. The goal is to be able to go out and make sure we benefit from the original ideas that artificial intelligence generates, allowing us to solve problems that we could never have dreamed of before. Do you know how many people in their 50s would pay to go back to the age of 19 if they could? This is a magical time to be alive in technology. Today I'm talking to Jitu Patel, President and Chief Product Officer at Cisco. We'll discuss how Jitu brings a founder's mindset to Cisco by fully immersing himself in the AI happening inside the massive data center they're at the center of, and then
we'll talk about what founders can learn from Jitu's incredible personal story. Jeethu, thank you very much for coming. Thank you for inviting me. Nice to see you. You have an incredible background, from growing up in India, working as a waiter at Sizzler, to being the president of one of the largest companies in the world. But before we talk about all that, I want to talk about Cisco today, your unique role, and what you do there. I think about my role, and that's that I'm just trying to help create the best products in the world that all of our customers can use at scale.
However, what Cisco does that is even more interesting to me is that we think of ourselves as a company that was in the business of finding picks and shovels during the gold rush. We are a company that is doing AI infrastructure for the AI era during this massive movement that is happening now, which is an age-old shift; and we want to make sure that we provide the right level of infrastructure for AI to work the way you want it to work, and for it to work safely and reliably. And that description of Cisco that I love, "looking with picks and shovels for the AI era, " you clearly would n't have described Cisco that way four years ago. So, all of this is recent. This is a transformation of a very large company that was
already operating on a huge scale. What was it like to come in and try to transform an organization of this scale and in such a short time frame? You know, I think that during these massive changes that happen in companies, leaders usually get an undeserved level of recognition, and the people who do the work don't get enough recognition. But the reality is that I didn't write a single line of code while we were doing this. The team did it. But we made sure we set the right direction, hired the right people for the right positions, made sure we had the right mechanisms in place to identify what markets we were going to pursue and do
it in a way that made everyone feel like they could, that they could win. Can you tell the story of how you ended up at Cisco and what you do there? So, I ended up at Cisco. I worked at Box. I was managing the product with, you know, Aaron Levy, who is the founder himself. And so, my mental model was reinforced again when I came to Box, because many years ago I started my own company and then moved to EMC. EMC was a very large company. I thought, "Oh, that's very corporate." And then I went to Box. I thought, "Okay, this could be interesting because you could work the same way I used to work at my own company." And I think there are two types of people. There are people who are
very good functional leaders, and they like to be in their place. Uh, no one tells me what to do in my place. I'm not going to tell you what you do in your place. And there are people who have run a certain business and consider themselves business owners, and they are quite comfortable constantly poking around in other people's business. And I'm prone to that. So personally, I think when you have that kind of founder approach, it's very good for wartime, and it's very good for times when the market is going through change. Things are moving at a very high clock speed, and all you have to do is make sure that you don't completely alienate people who may not be founders, and help them understand the advantage that if you go into
founder mode, we can actually operate at a completely different frequency than we've been operating at so far. And that's what we started doing at Cisco, and I tell my colleagues and everyone in my company: I feel like a I came 36 years later, but I feel like I own the company, and it's my company, and if, um, someone from the outside tries to, um, ruin the company, be careful, because I'm definitely going to fight for it. And we have to make sure that we are very proud of the craftsmanship and what the company stands for. It's all your work. So, if something isn't working well in sales, I'll go talk to the people there. Something is not working right in marketing, I'll go talk to the people from there, and I fully expect them to come looking for
holes in my business one of the reasons you want to bring former founders to Cisco? You know, I also recognize, um, and appreciate the value of experienced operators who think systems. Hmm. Because if everyone at Cisco were founders, it would be terrible. Yes. Uh, so you have to make sure you get a good balance. We did it so that we have, I try to call it the rule of thirds. So, a third of my management team are people who know how to maneuver with Cisco, they are scale operators. The second third are people I invite from outside, from competitors, from the market, who understand the market very well.
Hmm. Uh, who may not know how to maneuver with Cisco, but they think systems, and they've worked at scale. And then the third are the co-founders and CEOs of the companies we acquired. Hmm. But we provide these people with charters that are larger than the company they sold us. You said earlier that this way of working is especially important in wartime. Do you think Cisco is in wartime right now I feel like Cisco is in wartime right now , um, in the sense that we've made one of the most successful turnarounds you've seen from a product perspective. Um, I think we're fully in the AI infrastructure game now. Uh, this is something no one would have praised us for three or four years ago.
But if you look ahead to the next three, four, five, seven years, um, the amount of opportunity that's in front of us is so vast compared to anything we've experienced before that we have to make sure that we can take advantage of all of it. And that means we have to operate like I like to tell my team internally, that we have to operate like the biggest startup in the world. This means you have to work quickly, but with scale. If you're just working at speed, then you're just a startup. If you operate simply on a large scale, then you are simply the largest company. If you want to operate like the biggest startup in the world, it has to be fast and scalable.
What are the key things you need to do right over the next few years to seize this opportunity? Let's assume for now that less than 2% of people use agents in any capacity as power users. And think about the fact that all the infrastructure that is being created today is being consumed. We lack infrastructure. We are limited in supply because only less than 2% of people use agents. Imagine when 25, 30, 40, 50% of people start using agents. The amount of infrastructure you will need will be on a completely different scale than what we have now. And we want to make sure we can help solve this problem. This is problem number one. Problem number two is these agents. Actually, this week was very interesting because these agents will be
scamming. Sometimes they may have their own opinions. It's actually getting harder and harder to distinguish when an agent is acting with malicious intent because the enemy is influencing or poisoning the agent, or when the agent just thinks it's the right thing to do and is just, you know, dutifully following your slash/go command. And that's why we need some means of security, protection, and surveillance. And that's the second big problem that we want to solve, is that we want to make sure that AI is safe to use for everyone, and that it doesn't turn into a constant fear that keeps people from using AI. In fact, I think it's the opposite. If you don't use AI, you will be less protected. And that's true, because most attacks happen at machine scale, so you
need machine scale protection as well. So, we are one of the largest network companies in the world that provides infrastructure. We are also one of the largest security companies in the world. We want to make sure we do that, and we're also one of the most important And so the combination of these three should give us all the basic machinery we need to become critical infrastructure for critical infrastructure companies in the age of artificial intelligence. You're fully invested in AI, both in terms of investing in the AI and data center market, and in terms of implementing AI at Cisco to make everything more efficient.
I'm actually really curious to see what implementing artificial intelligence looks like at a company the size of Cisco. This cannot be related to efficiency. The goal is not efficiency. The goal is to be able to go out and make sure that we benefit from the original knowledge that artificial intelligence generates, allowing us to solve problems that we never dreamed of solving before. Which is very different from simply speeding up the work by 20% or providing fewer people. In fact, I strongly suggest that we will need more engineers in the future, not fewer. Because every time I've seen an improvement in step-by-step functionality in AI, I've seen a bottleneck that's emerging, which is a human bottleneck. You know, coding is automated, code review becomes a
bottleneck. Code review is automated, deciding what to build and what not to build becomes a bottleneck. So I feel like we're going to need more people, but the most interesting thing was that 32,000 engineers and the rest of the company: first, tasted honey ; and secondly, were not afraid of the fact that if you use it, you are actually creating a trap for yourself where you could lose your job. Instead, we did something completely opposite. We guaranteed our employees that if they didn't use AI, they would lose their jobs. And so it was a completely different way of approaching the problem. This wasn't a very controversial statement, because think about it today: if you had someone who couldn't use a PC or the Internet, would you hire them to be a
knowledge worker? You would never. So it's not that controversial to say that if you're seeing age-old changes, and if there's this huge platform that's going to be reorganized by artificial intelligence, then anyone who's not using it is just not going to be as relevant. And that's why we always believe that there will only be two types of companies and two types of people. Some who are very fluent and skilled in artificial intelligence, and others who will really fight for relevance. And so we just made it very clear to people that if you happen to fall into that second camp because you're not using AI, chances are Cisco is not the place for you. And so what you need to do is fully commit, and we will give you all the tools and all the support to
fully commit. At first, we actually gave everyone unlimited tokens to say, "Just start using this." Because, in my opinion, you need to…It's like riding a bike. First you need to become familiar with something, then learn to do it well, and then you can become effective. Yes. But if you try to focus on efficiency first, you'll never get familiar or you'll never get good. And that's why we started doing it. Initially, many people faced great resistance. The first few months were difficult. And the deciding factor for me was the fact that I have one engineer on my team who is an engineering leader, one of the most amazing people. And he is always a
skeptic. He always tells me, "Dude, these ideas are crazy. I don't think that will ever happen." And one night he called me, and he was as sincere as possible, and he said, "Hey, I just want to let you know that I was wrong. You were right. Models are now evolving at a rapid pace. And I got some benefit because, you know, Kevin Wills was on our board of directors, and I had some, you know, kind of. I knew some people in the ecosystem that every time I talked to someone from the startup world, I got a completely different set of data than what I was getting internally. And so, Will says to me, "It finally got to the point where we took and refactored, you know, half a million lines of code, and we were able to refactor it down to
about 100,000 to 120,000 lines of code or something like that. And so, a 5x improvement, and we were able to do it in about 2 weeks. 2 weeks, yes. Do you understand? And so, first of all, you're right that this is actually going to move very quickly." So, I can't just think about where we are today. I have to think about where it will be in 3 months? Not in a year, but in 3 months. And assume this and start making changes today. But secondly, I'm also very worried that this could mean a lot of bad things for people, and we might have to make a lot of changes to our staff.
Hmm. And I'm like, "Or we can actually motivate everyone, Hmm." to make sure that we can, um, we can all become better as a result of using this." Then, you know, it just got worse. We actually became the first design partner with Open AI and Codex. They gave us a few full-time employees. And to this day, I, um, every, um, it seems like every two or three weeks, um, I sit down with my entire engineering leadership team and their staff, and we discuss what went well and what went wrong in the last two weeks. And, um, it was very clear to people that I was just not going to let it go. And I think in big companies there are a few things where you have to go from the top down.
Most things are going to be pretty hard to go top- down, but some things that you consider religion, you have to go top- down, and you have to just make sure you're relentless about them, and AI was one of them. We've all seen how AI is transforming software development, but Cisco is doing a lot of hardware development Yes. Do you see the same impact on hardware development, or is that still a matter of the future? So Cisco deals in software, hardware, and silicon. And each of them has a different implementation cycle. So, on the hardware side, you're starting to see, um, good design patterns starting to emerge now. Uh, but it's not as strong as Software, clock speed is very high. Silicon. the team is somewhat
intentionally not fully committed, because they're very concerned that the data and intellectual property won't disappear, and then, before you know it, someone else will have recreated it. Right. You know, so on the silicon side, we were moving a little slower. But now, hmm, I think that with the improvement of the models, with the emergence of models with open weights, with their own infrastructure, etc., I think that will change. But we need to compress the cycle time for silicon. The interesting thing is that we are a full stack, right? We create silicon, we create photonics, we create system hardware, software, the whole stack. Each of the elements and layers of the stack has a different
planning cycle. So, silicon has a 5-year planning cycle. What I'm doing today is making sure I know what will happen to large-scale production in 5 years, and then making the appropriate investments today. This is nonsense. How can you even do that with artificial intelligence? Artificial intelligence is only three or four years old. Think about the entire, um, production cycle, not just the silicon, but from silicon to systems, SDK, and getting it to market. You have to make sure you start because in order to create silicon, you have to create intellectual property for the silicon, like
series and all these components. And it actually takes, um, you have to create them in advance. So there's a sequence that has to happen in silicon that can't happen in, you know, you can't just say, " Make a chip in 12 months." Uh, you could build a chip, but it just wouldn't be that good. Hardware typically has, um, an 18 to 24 month planning cycle, maybe 30 months in some cases because of the supply chain. Because we think about it. We do this on a large scale for very large volumes of units. You move to software, operating system, it's a 12-month planning cycle, maybe an 18-month planning cycle. Then you move on to models, agents, and applications. Now you have a 3-month
planning cycle, maybe a weekly planning cycle. And we have to make sure that this co- developed full stack actually has all of these things working harmoniously. And if you can figure that out as a company, you have a very strategic moat in the sense that you've been able to figure out all these different layers that work well together from an intellectual property perspective. This is something that is very difficult for other people to do and replicate. Is it possible to speed up cycles now that artificial intelligence is here? Absolutely. They will all be compressed. But when I say planning cycle, I don't just mean how long it takes to create. I mean the process from idea to large-scale production with
billions of dollars in revenue. Speaking of chips, supply chain, I want to talk about data center construction because this incredible historical event is happening. We are building more things than we have built in a generation, and you are at the center of it. Can you talk about the role that Cisco is playing in this huge everyone knows the role of the graphics processor in the development of data centers. And what happened was that when you first started seeing these Transformers come out, they were housed on a single GPU. So the training launches were pretty simple. And then, uh, you had more options, and these GPUs needed scaling. So you had eight GPUs on the server. And so, you went to the server, but all these GPUs were connected together in
one server. Then you had the server racks. The graphics processors were interconnected in a server rack, working in unison. Like one, you know, giant machine with, um, different things. And then the racks started to scale into rows. Because now you had multiple rows of racks that needed to be networked together in a large data center. they call a scalable network, not just a scalable network for, um, for rack-level scaling. And now, due to huge capacity constraints and electricity shortages, you can't always deliver all the power to where the data center is. So you turn to where there is available energy. And so you can have data centers located hundreds of kilometers apart that have to operate as one complete, ahem,
logical unit. Oh, two different physical data centers separated by hundreds of miles that need to be networked as if they were in the same building. It's like they're in the same building. It's as if it's the same This requires a completely different family of networks, called scaling networks. For a data center, you can have a scalable network. And for the rack, you have a scalable network. We are mainly focused on scaling and globally. And we make not only the silicon and the interconnect trays to connect GPUs together, but also the photonics, which is the optics. We started our business selling to hyperscalers 2 years ago for AI workloads
. And we thought we would give the market, uh, this street, a certain landmark, and said we would place an order for a billion dollars. In the first year, we received 2.3 billion. Wow. So the next year we said, "Well, let's double it." Instead of 2.3, let's make it about 5 billion. Last year, which was the fourth quarter, which just ended a month ago, we received 9 billion. Wow. In both years, you doubled the very optimistic forecasts you made. Yes. And so we actually went from zero to 9.3 billion in 2 years.
Wow. Right? And this is the speed at which everything moves. We still need to capture so many shares. This is a complex business, $63 billion in revenue. Hundreds of products, ahem, that we have to make sure we're maneuvering in the right direction But 9 billion out of the total amount of 63 billion is enough material for a completely new business direction. 9 billion orders. It takes some time to get income, but yes. Yes. Of course. Yes. But you're right. This is starting to make sense. And it's exciting because you're right in the thick of things. And the market is so huge. A gigawatt of capacity would cost somewhere between.
It used to be 35 billion. Now it is almost 50 billion dollars. Uh, that's how much it would cost to build a gigawatt of data center capacity. And about 10 to 15% of that typically goes to networks. It's mostly in the GPU, but there's a huge complexity, you know, you need to build a shell and get power permissions for it. When you go into a data center, you're amazed at how complex it is, when you make a request in your query, like give me the pizza recipe, what's going on behind the scenes to get it in a second or two? Because the buildup is so extreme, some people have wondered if we are rebuilding. Is that so?
Absolutely not. Let me explain why. And people try to compare it. First, there are no previous events in artificial intelligence. So I fundamentally believe that, and this is one of the big mistakes that people make, they keep going back to the experiences they already have. I think one of the most important skills to master in artificial intelligence is unlearning, because your past experience can be a huge burden. But people usually try to compare it to the dot-com boom. Oh my God, so much infrastructure has been built. What happened at the time was this: you created an infrastructure structure and hoped that demand would catch up with supply. Now, no matter how much infrastructure you build, it is consumed almost instantly. So there is no
time to catch up as your supply is limited despite exponential growth that looks like a vertical line. So I feel like from that perspective, I don't think there's, um, a huge, um, overgrowth. In fact, I think it's the opposite. If less than 2% of people use agents, and those agents consume this much resources on average, then an agent consumes 450% more network bandwidth than a human doing the same task. Now think about trillions of agents. Right. Um, now think about most of your tasks that agents do, where they will need to make sure they are constantly enriching the context. You have a skills file, you have a markup file. You know,
you have a G-stack from Gary, and it's going to be constantly moving back and forth. It seems like there will be a lot of data exchange back and forth. That will be a lot of tokens. Yes. Speaking of agents, I heard you've recently become a fan of using your own personal agents. I'm curious how you use agents? I've been using it for a while now, but Instinct has been a force multiplier for me, and that was only 4 days ago. I'm not good at email. I'm really bad at email. For example, 5 days ago I had 171,762 emails or something like that. 146,000 of them were unread.
OK. I used Instinct. I took a risk and said, " You know what? I'm going to give him access to my personal email." I have 200 letters. Wow. It swept through your entire mailbox. He took care of everything. He told me what to do. He was actually moving things. He archived them properly. If I needed something, he would bring it back. It actually summarized my news from all the sources I was getting. I felt like even though I didn't really use email that much, there was a part of me subconsciously that felt overwhelmed. And I just felt better, like, wow, I got through this. Now suddenly I'm checking my email. I'm like, "Why am I checking my
email?" Hmm. Because it's fun to check your email when you have three emails. This is really due to the fact that building a data center is never enough, because, for example, tokens to look at 140,000 emails and figure out what to do with them. Absolutely. This is where I feel like I just don't think it's somehow unfortunate that there's this level of exaggeration going on, oh my God, this is a bubble. Everything will be as you can imagine, but two conditions may still be met. You're going to have an age shift that's going to completely fundamentally change the way people live, and that's going to require a lot of infrastructure, and you're also going to have some companies that are very volatile in valuation. Because every now and then, any large-scale
platform shift like this one will see a lot of experiments, and only a few will win. Not all of them will win. So, there will be some volatility in the valuation, but in my opinion, that doesn't mean it's a bubble. The bubble is when this age shift itself is questioned. I think it's very difficult to come out and argue about this. So, recently there's been news in the news about, um, open AI agents going rogue, compromising themselves, hugging faces, etc. What do you think about what's happening with AI safety right now? And also, how does this affect the deployment of AI in a large, mission- critical enterprise?
Yes, I think that's a great question. One of the things I would say right now is that this is going to be a defining, um, you know, theme of our time over the next, um, next few years, because the constraints and the infrastructure have been a really big problem. And we talked a lot about whether there is accessible computing? This is still a problem. But the much, much bigger issue is whether I can trust these systems to which I am going to delegate all my work. And my entire personal life. The difference between trusted delegation and untrusted delegation is the difference between market leadership and complete bankruptcy, in my opinion. So, I think we have to make sure that we actually get through this. And for this, a
whole series of things are needed. And we, Cisco, are actually working hard to provide a holistic agent security platform that we, And if you jailbreak it and you ask a question that you don't want it to do, will we have an answer, or if you force the agent to do an action that you do n't want it to do? How can you intercept that and make sure that dynamically at runtime you can set, you can apply the safety fences? All of this apparatus that we actually make available, including the security side as well as the We have this as a product, and thank you for bringing it to market. In fact, we have a product called AI-Based Protection, which is the foundation of something we launched a year and a half ago.
It works extremely well and, um, we just need to make sure we get more and more of it and it was only made for the chat era. So we did it for the model first and then extended it to agents. And now we have dynamic monitoring of agents at all times, and you know, if you ask an agent to refund you, they shouldn't refund you three times what you paid for a pair of shoes, you know, and that's something that happens all the time in retail, so we acquired a company called Galileo. Hmm, there were people from DeepMind and Uber who came, and these people did a great job. They are part of the Splunk team, and they have done a great job on this Can you take us back to your early career and help us understand how your childhood and the early jobs you worked at led
to this unique position that has allowed you to make such a transformative impact now? When I look back, I was very lucky. At the time, it seemed like I wasn't very lucky, but my dad was a He looked like Bernie Madoff. Uh, he cheated people to get money. Uh , at some point it became dangerous for me to live in India, and there were people who wanted to kidnap me. He had a mistress, this mistress committed suicide. I did n't want to stay. Oh, my God. You know, it was exactly what a movie could be made about, that part of my life. And that's what I decided to do, and my father was very cruel to my mother, and I was very close to her. And so we both ran away, and she decided to, you know,
live in, well, an undisclosed location, and I said, "Let me just go and check out America, maybe I can check out some schools and stuff." They came here. Were you there? I was 19 years old then, in 1991. Right? And so, well, I came here, my uncle took me in for a while, I was a waiter at a fancy restaurant called the Sizzler. dollar. And, well, you know, during that time, there were a few seminal events that really made me rethink my mentality, because I was a terrible student in school. Um, I wasn't like that, you know, I was the black sheep of the family in the sense that a lot of my cousins were very smart academically,
and I wasn't. And that's why I always had this inferiority complex. I stuttered, so I was always silent. Um, I couldn't say hello when I picked up the phone, and so all this stuff happened, and then I started waiting tables, and I'm like: boom. You have to find a way to entertain people, otherwise you won't make money. And then I came, you know, when I was here, I was a student. I came across a company called DocuLabs, and they wanted to give me an internship because my professor said to me, "Hey, maybe you'd like to do an internship here." She was my mentor in college, and she was great, so she, um, introduced me, and I got the job as an intern. A few months later, the guy, the founder, comes up to me and says, "Well, we're
going to leave. We are going to buy out his share." I'm like, "Oh, I want to." And they're like, "Well, each of us is investing, um, a quarter of a million dollars." You can invest a quarter of a million dollars, and you can keep a share for yourself ." I went to one of my friends, and, uh, he was a banker, and I asked him, "I need a quarter of a million dollars." And he said, "Why do you need a quarter of a million dollars?" I replied: "I will buy this company." He said, "What are you going to do when you buy the company? "I'm going to manage it." Remember, I was not the founder of this company. I was an intern. You were an intern. You were still in college.
I had the first loan I took out was 10,000 and the second was 250,000, but the first loan I took out was 250,000, but the first loan I took out was in college. I was earning minimum wage. Yes. Right? And I said, "Give me some money and I'll give it back to you." He said: "10% of the year." I ran this business for 17 years. And now it's a company with a turnover of less than $5 million. You stayed there for 17 years. In a way, it was the biggest mistake of my career. Hmm, because it wasn't because the company was bad, but because the business model didn't fit what I wanted to do in life. And what. And sometimes, you know, for example, our
CEO Chuck says to me, "Jit, once you go public, um, you can't model is your business model. It takes a lot of effort to change your business model. But when you work in the service industry (it was a kind of marketing research company in the service industry), and I loved software, and I loved Silicon Valley, and I used to come here, and I had an apartment on Russian Hill for a while, and I would come here, and I just fell in love with the people in Silicon Valley, and I'm like, "I want to work with these people because the decision they make in this Windows Start button is going to affect billions of people, and I want to be that person, you know?" "Um, but I never got the same level of, uh, enjoyment out of it, but I stayed there because of
ego. And that's another lesson I'll give to founders. I stayed there because I had this ego that I would never work for anyone else. This is my own company. I am my own boss. I do n't want to work for anyone else. The biggest mistake I made. Because ego guided my decision, not learning that guided my decision. So, in the end, I decided to go work at EMC. And the guy who hired me at EMC is still my coach. Um, and we had dinner. EMC, a huge company. 60,000 people. Right? My headcount grows from 25 to 60,000. He hires me as CTO. Um, and we're having dinner, uh, like this across the table, and I say, "His name is Rick Davenuty, and I say, 'Rick, uh, you should pay me
more because, you know, I have 17 years of experience.'" "He said, 'G3, you don't have 17 years of experience. You have 1 year of experience, 17 times exceeded. "Come work for me, and in your first year I'll give you 17 years of experience." I learned more from him in the first year than I did in 17 years of running my own business. It was absolutely magical. I hated working for him at the time. But after we left and I moved on to Box, there's a person who is my co-founder in life, and I think of her like this: you know, she's, um, she's my COO. I don't take on work without it. She and I always thought, " What would Rick do at this point?" And we kept asking ourselves that, and it was a great mentor.
Um, then I moved on to Box. Box helped me grow the business from $200 million to $800 million during my tenure, where I managed all products for the rest of its existence. Then I was going to move to advertising at Fang. Not from Cisco at all. Yes., but the offer they were going to make to Jessie, who was with me, didn't suit her. And she was my co-founder during my lifetime. I wasn't going to go anywhere without her because she was with me the whole time. Um, and I'm like, "Okay, I need to figure out what to do." Um, and literally by chance, Chuck calls me the next day. Hmm. And he says, "Hey, you should probably think about coming here." And I told him, "Hey, listen, I actually accepted this job." And he told me: "This company
will work well with or without you. But if you come to Cisco, the trajectory of development will be shaped by what you do." And then we were only given two offers within 9 days. So you knew the company was ready to run at a very high clock speed. The question I asked him was, "How much freedom will you give me to just be obsessed with the product?" Uh, and he said, "Yeah, I'd like you to be obsessed with products." And we did that, and before you know it, you know, Cisco, um, became the place where I think I got some of my career, the biggest level of growth that I've had, probably in the last 6 years. What lessons would you share with someone who, like you, is around 19 years old, maybe even grew up in India or somewhere outside the valley where
they are not where they are from the outside looking in? I think the first lesson I would say is that your upper limit is usually limited by your own ambitions, not, um, any other factor. Usually, it's not your ability. I am a very averagely intelligent person. Hmm, I didn't have a high school diploma. Uh, but I was hungrier than everyone else. And so I'm still, um, I'm hungrier than most of the people I work with. Not because they aren't hungry, but because I just feel like I don't want to let them down. So I'll just continue in the same vein. And no one will tell me that I'm not wasting my time on it. And I think the first thing to do is take your time.
Think big and have very bold, uh, ambitions. Because if your imagination, as imagination, can limit us much more than we think it can. You know, if you have a goal that big, chances are you'll never reach it, but if you have a goal that's right here, and if you reach it, that's pretty good, you know? And that's a pretty important, uh, you know, dimension to keep in mind. The second aspect- business-is a team sport. Take pride in the longevity of the relationships you can build where you have actually made other people successful because of you. And this, I think, will be of great importance. And if you
think like this from the age of 19, then your advantage will be incredibly large, you know? Um, and then the third area that I usually talk to founders about is the concept of the power of compound interest. For example, you just need to become 1.27% better today than you were yesterday. And within a year you will become 100 times better. And if you do it for 10 years, that's pretty good. 10 years generally seems like a long time when you're 19. For example, Sam Altman had a blog that he wrote once a day-that's a long time, but decades are short. I read this every year. This is an amazing blog post. This is one of the best blog posts he has written. For example, Jeff Bezos would say that a founder's ability to think in a 20-year window versus a 3-
month window significantly changes the quality of the decisions you make. And if you think about the really great entrepreneurs that we all respect, think about Jensen, think about Elon Musk, um, none of them have been accused of thinking in a very short-term perspective. I mean, they gave it their all for a long time. The only difference is to go all out when the business model actually makes sense, but don't do it like I did and go all out in a business that doesn't really match your ambitions, because my business, the one I ran, was great for business. It was terrifying to change the world. I wanted to change the world. You know, this is the wrong business to be in to change the world. But if you wanted it to be a good lifestyle business and
provide well for your family, great. Continue in the same spirit. This is great. And by the way, none of these answers are wrong. It just so happened that my ambitions didn't align with the business model, and that's another thing you have to remember, is do your ambitions align with what you're actually doing right now? Where will you feel satisfaction when you look back in 80 years and say, " You know what? I'm very pleased that this is how I spent my time." I think this is a question we should constantly ask ourselves, and I always tell younger people: "Time flies so fast. I never thought I'd be 55. You know, I never thought I'd be called uncle. I never thought I would have a child. And I never thought she would say, "Dad, you're old." Or bald.
Uh, and so things like that sometimes give you a lot of humility to make sure you don't take time for granted and spend your time where it really matters. What advice would you give to entrepreneurs or people coming into an organization like Cisco and trying to change it the way you did? I'll tell you a story. So, when I.I haven't been back to India since I left India. In 1991 I left. My dad died in 2004. I went to see him once for 48 hours before he died. But then I only came back in 2017. So, for whatever practical reason, from 1991 to
2017 I didn't spend much time in India. I took my daughter there. We went to the Taj Mahal. And there was this guide. His name is Raj. He shows us the Taj Mahal. This guy seemed like a really good project. He knew every detail about his product, that is, the Taj Mahal experience he was selling. But then people would walk by and he would start blasting out different languages: Chinese, German, French, and Spanish, and at some point I said, "Dude, how many languages do you speak?" He gave me some incredibly large number, like 12 or 14 languages or something. And I said, "This is crazy.
" He says, "Well, you know, I want to honor the people who come here because, um, they come here, and I shouldn't be so presumptuous as to think they're going to speak my language. I have to speak their language." I mean, that's very noble. He's obviously smart, dude. Eh, I think to myself, "This guy is smarter than anyone I worked with at Boxed at the time, anyone I work with." He probably earns $10 a day. I'm sitting here in Silicon Valley. Well, why is that? Because I have a platform and he doesn't. Right? I have an education platform. I have an American platform. I have a technology platform. I have a platform that simply gives me opportunities that he didn't have. And so, no matter how smart he is, no matter how hard he
works, he finds it difficult to stand out from the crowd. If you happen to be given a platform, my only request to people is not to waste it. And as a founder, you know, we can all look at everything as a glass half empty, and that's what we do n't have. But, my friend, we are all living in one of the most special times. And if you're 19 or 20 years old, or you're early in your career at this point in the AI movement, do you know how many people in their 50s, how much money would they pay to go back to being 19 if they could? This is a magical time to be alive in technology. This is a great time to make a real impact on the world. And if you spend your time associated with YC, which I think is one of the best organizations (which is why I love
working with you), you'd be crazy to just work your tail off and do everything you can't do for any other reason than to make money, but because your impact on society as a whole will be so nice when you're older that it'll be something worth doing. And if you happen to sell your company to a large company, don't be cynical about going to a large company. Go learn what a big company teaches you, because all of these big companies were startups at some point in time. And if you succeed, you will eventually become a big company. And don't disrespect big companies. Actually learn how to get big companies to operate with a level of, you know, kind of, accuracy. And then you can benefit from it too. And go with extreme curiosity, no matter what path
you take, but remember that you are being given a platform, and very few people in the world look up to 8 billion people, very few people have a platform like we do. Gigi, that's such an inspiring note to end on. Thank you very much for joining us. Thank you for inviting me. It was wonderful.
