4 Oct 2025 · 58m · Nikesh Arora, Sarah Guo, Elad Gil · 22 quoted facts
Nikesh Arora argues that AI will change three things at once: how search and apps make money, how fast cyberattacks happen, and how companies are organized. His view is that value will go to firms that combine strong AI models with proprietary data and an end-to-end platform, not thin wrappers around a model. He is optimistic, but warns that old defenses and old business models will not survive.
The main tension is speed against trust. Attackers can now use AI to get in and out of a company in under an hour, yet most companies still respond in days. Enterprises also will not hand important actions to AI agents that sometimes make mistakes. Arora's answer is to consolidate security into a few platforms, watch for unusual behavior, and keep humans in the loop for now.
Arora says search ads mostly sell leads, while people pay more for a completed transaction. If agents book the flight or buy the item directly, the business model shifts from ads to fees on transactions, and thin app front ends become vulnerable.
A company that only adds guardrails around a general model risks being absorbed as the model improves. Arora says survivors will own proprietary data or a system of record, because even a very smart model must be taught a company's context, like a new PhD hire.
Arora says the average time to break in and steal data was three to four days seven years ago, and the fastest now is 23 minutes. Average defender response is still measured in days, so companies have to respond in under an hour.
Deepfakes and AI make passwords and security questions easier to beat. Arora's answer is to give people temporary, just-in-time access and block activity that is out of character, such as someone downloading everything after years of normal use.
Arora says sales, marketing and support eat 50 to 65 percent of a small enterprise company's costs but about 30 percent at scale. Customers already trust a security vendor, so selling them more products is cheaper than finding new customers, and a security market with 118 vendors in one company is ripe for consolidation.
Arora doubts it will, because Google has the distribution, product skill and AI skill to turn search into an 'ask me anything' product. The open question he raises is how Google will make money once it moves from ads next to links to AI answers and agents.
An agent is software that carries out a task for you, such as booking a ticket, instead of only answering a question. Arora says that could make many apps into background services and change who earns money from the transaction.
Arora says the most exposed jobs are large groups of people doing repetitive, generic tasks like documentation and customer support, where he thinks 80 to 90 percent could be taken out in two to five years. He expects sales to change less and good product people to stay, just working faster.
It speeds attacks up. Arora says attackers can use agents to find weak points in minutes, and the fastest attack his firm has seen took 23 minutes, while most companies still take days to respond.
Arora says he treats acquisitions as product development, buying number one or number two companies that venture investors have already funded. He says Palo Alto has bought 27 companies, often keeps their leaders, and bought Protect AI for AI model scanning and red teaming.
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Nikesh Arora joined Palo Alto Networks in 2018, when it was known as a next-generation firewall company, and the host says it has since grown six to seven times in size.
“He joined Palo Alto in 2018, when it was the next gen firewall player and has since grown it to six to seven times the size.”
Nikesh Arora was a senior vice president and chief business officer at Google from 2004 to 2014.
“previously, he was the SVP and CBO of Google during its massive growth phase from 2004 to 2014.”
Arora says Larry Page talked about Google answering a user's intent rather than the literal words typed, and that he also talked about AI.
“He used to talk about getting to a point where you answer my question, answer my intent, as opposed to answer what I type.”
Arora recalls that for years it was unclear how YouTube would make money while Netflix was seen as the profitable streaming business.
“I remember how for the longest of years, nobody quite knew how YouTube was gonna make money, and everybody was looking at Netflix as the one who are making money in streaming.”
Arora says advertisers pay more for a completed transaction than for a sales lead, which is why agent-driven transactions could change the search business model.
“People pay a lot more for the consummated transaction than for the lead.”
Arora says software developers are the first group to adopt AI at work because they can work with partly right output and fix the rest.
“Developers is the 1st port of call, because the more they're, the most likely to experiment and to be able to work with half baked outcomes.”
Arora says cybersecurity as an industry only appeared once computers were connected to each other.
“because cybersecurity did not exist before connectivity came about.”
Arora says cybersecurity is about 25 years old and became fragmented because each new threat was met with a separate point product.
“It's a 25 year old industry, so every time a bad problem showed up, somebody built a point solution to solve that problem.”
Arora says the average time for an attacker to break in and steal data was three to four days when he started at Palo Alto, and the fastest his firm has seen is 23 minutes.
“the average time to identify a target, get through it and exfiltrate data was in the three to four day timeframe. The fastest we've seen, I think right now is 23 min.”
Arora says about 89 percent of attacks involve credential theft, where an attacker takes over someone's identity.
“89 of the attacks happen because of credential theft. Somebody becomes you or me.”
Arora says Palo Alto Networks is buying an identity security company because traditional identity tools only verify people at login.
“That's why we're buying this identity company, because identity companies today say, Oh, I checked you in the door, you're fine.”
Arora says that at enterprise software companies under $1 billion in revenue, 50 to 65 percent of costs go to sales, marketing and customer support.
“if you look at a enterprise company with less than $1 billion in revenue, 50 to 65 of the cost is cost of sales, marketing and customer support.”
Arora says Palo Alto Networks has grouped its product lines into three platforms.
“We have 44 magic, Quadrant and Stoplight mentions. We've turned that into three platforms now.”
Arora says Palo Alto Networks has bought 27 companies and was about to make its largest acquisition, pending approval.
“I think we bought 27 companies so far, and about to buy our largest one if we get approval to get it done.”
Arora says Palo Alto Networks acquired Protect AI, a company that scans AI models and runs continuous red-team attacks against them.
“So there's a company called Protect AI, which was actually assessing models, they were doing persistent red teaming against the AI models”
Arora says Palo Alto Networks built an AI firewall to keep enterprises' AI systems controlled and data from leaking.
“So we built effectively what we called an AI firewall.”
Arora says he expanded his staff meeting from eight to 25 people after reading about Jensen Huang's wide management structure.
“I actually expanded my staff meeting from eight to 25.”
Elad Gil says Nvidia chief Jensen Huang has told the show he has about 40 direct reports and does not hold one-on-one meetings.
“we've talked to Jensen a few times and he's pointed out, for example, he has like 40 direct reports, he doesn't do one on ones”
Arora says he meets 50 employees every two weeks to check whether the company's strategy has been understood.
“I meet 50 employees every two weeks and I ask them questions.”
Arora says an AI coding agent found a vulnerability in Palo Alto Networks security code that the company would otherwise only have discovered after it was exploited.
“I've seen the vibe coding agent find a vulnerability in security code at Palo Alto, which we wouldn't have found unless it was out in the wild.”
Arora says he has seen an AI coding agent rewrite 500 lines of code as 75 lines that were more efficient.
“I've seen it take 500 lines of code and and come back with 75 lines of code, which would be much more efficient”
Arora says he tells his staff that customer support exists because of bad products, and he moved support closer to product teams at Palo Alto Networks.
“I always joke internally, I tell my people customer support exists because we build bad products”
Arora, a former Google executive, explains why Google is well placed for AI answers but faces a harder question about how to make money from them.
“I call that democratization of intelligence.”
Agents that complete purchases may be worth more than leads, but they threaten apps that are thin interfaces over a back-end system.
“maybe the business model transition is stop giving me leads, give me consummated transactions through agents.”
A smart model still needs company context and proprietary data, so wrapper startups are exposed while systems of record are safer.
“getting the smartest model in the world is like hiring the smartest PhD”
Security sensors stop known threats but most breaches come from unknown ones, which Arora argues need all the data in one place.
“Most cybersecurity breaches happen because of the unknown bad”
Attacks that took days now take under an hour, which forces much faster defense.
“the biggest threat that AI brings is that it continues to compress the timelines”
Arora explains why Palo Alto is buying an identity company and moving toward blocking anomalous behavior.
“stop giving people persistent rights, give them just in time rights”
Selling more products to trusted customers is the main way to cut sales and support costs as a company grows.
“50 to 65 of the cost is cost of sales, marketing and customer support.”
Arora expects big cuts in customer support and documentation, but faster innovation rather than fewer people in product development.
“we should be able to take 80, 90 of customer support out in the next two to five years”
Arora describes his wide staff meetings, direct talks with employees and view of M&A as outsourced R&D.
“I call it product development and research in a highly innovative market”
Nobody agrees on what an agent is, which makes it hard to secure, but Arora stays optimistic about AI overall.
“To agree with the other person's definition of an agent.”
Hi listeners, welcome back to No Priors. Today we're here with Nikesh Arora, the CEO of Palo Alto Networks. He joined Palo Alto in 2018, when it was the next gen firewall player and has since grown it to six to seven times the size. As a leader, as a platform security company, previously, he was the SVP and CBO of Google during its massive growth phase from 2004 to 2014. Welcome Nikesh. Nikesh, thanks so much for being with us.
My pleasure, Uh, I don't know where to start because I wanna talk about AI, I wanna talk about security, I wanna talk about leadership. I do think, given your history, growing Google as chief business officer. Like, we have to ask you, what do? What do you think is the future of search, and how threatened is it?
nothing like a slow little low ball. You know, welcome me to your show and Gotta work we'll talk about
you up a little bit. chat. And
this guy used at Google too at that point, you know, I went to, so
but I talk to him too much.
and what did you think?
I think we should defer the question to you. Is the expert
Oh, look at that, he even learned it for his, for his own mouth. blocking Well done, have
back?
you do all the hard work? Like, I think, um, the idea when when search came about, I still remember going out there and trying to sell search to people. And it was the oh my god, you mean? I can just go to the internet, type something and I can get the answer. And we spent two decades trying to get all the information out there on the internet, so it was easily accessible to people. And I think you saw the benefits, you saw the benefits of, you know, democratization of information.
Farmers in India could get stuff and people get information. I think now, we're in age, people are saying, great, you know, don't give me all this stuff to sift through myself, try and make sense of all of it for me because it's too much. And that's what you're seeing in today's generative AI models. So, and I sort of, in my own words, I call that democratization of intelligence. All of us will have the basic intelligence which every other person next to us has, because we can kind of go figure it out.
I don't have to hire the same people to solve the same problem for me the ten thousandth time and pay them money because it's already been solved nine, 999 time. And the outcome's on the internet somewhere. So I think to the extent that Google has sharpened its sort of scales on putting all that information together. Being able to synthesize it, understand it, being able to interpret my intention as an end user and try and present me the most likely outcome.
I think that should translate well to the notion of generative AI, being able to summarize the same thing in a much more enhanced, ordered way for them. So I think from that perspective, will they have the ability to transition the current search product into a future product? Which is basically, you know, call it what you want to call it, you know, ask me anything, or and I? It's so funny. Like, you know, when you worked at Google 15 years ago, Larry had that vision.
He used to talk about getting to a point where you answer my question, answer my intent, as opposed to answer what I type. So he, he had the foresight to talk about that, he used to talk about AI. So I think from a product perspective, they are in a good position to be able to transition the product to what the end users need. And you've seen that with Gemini, you see that with ChatGpt, you see that with other models which are getting to the same place.
Let's not underestimate the distribution power they have. There are two or three companies in the world which have distribution of the billions. And whether it's Facebook with all their properties, or with Apple with their properties, or Google with their properties. So they have the distribution, they have the product chops, they have the AI chops. I think the question, bigger question is, how does the business model transform from what it has been? With? Ten blue links and ads being representative against them?
And what's the new monetization sort of that will come as a result of do you think that? Now?
it's likely to be agents or something else? In terms of starting to take action yeah, we can, we on your
can come there too. But behalf? Is that I think
fair?
because, see, how the search question, search is an advertising revenue question. So yes, the question is, how does the advertising revenue morph into some version of consumption or transaction metric? Which cause? That's what is kind of that's the intent. When I go say, what are the best blue pants in the world? It's not. I'm just not doing it for academic interest, I'm actually going to transact. So, yes, maybe an agent could do this, or the fastest flight to get to Rome, so the agent could do that.
And we'll talk about agents in a 2nd. But I think that's much more disruptive than generative AI. And to that extent, I think how the transition, the business model is going to be interesting. I don't think anyone knows what the business model is. But all I can say is that having been there and admired what they do, they do spend time getting the product adopted. 1st, and eventually, when there is tremendous amount of distribution for the product, you find a model emerges.
I remember how for the longest of years, nobody quite knew how YouTube was gonna make money, and everybody was looking at Netflix as the one who are making money in streaming. And YouTube wasn't. I think YouTube's a big ass business now compared to most of the streaming players in the world. So I think they'll figure out how that model transforms. I do think the agentic challenge is a much bigger challenge than the generative AI challenge.
What sort of challenge do you think that is, or how do you think that's gonna substantiate?
Like the the idea that you know, if if you step back, you know, we've spent 30 or 40 years building products where we focused on UI, right? Product managers are effectively glorified UI managers. So we're trying to make sure that us common folks can interact with data and some sort of engineering algorithm behind it, because we're not smart enough to talk to the engineering algorithm ourselves. So you go to Expedia, say bunch of boxes you're supposed to fill in as humans, and it gives the answer to us.
Because, without having us to write any search query, SQL, XQL, whatever you want to go find the information. Now, I think generative AI has made that easier. We can talk to the UI in natural language to some degree, it can generate outcomes on the fly. So I think to that extent, I think all of product development is going to change with generative AI and this natural language capability. Now, if you take that next step further and say, actually don't need to come interact with the UI, my agent can go do the task for you.
Now, step back and think of that. Like, you know, 50 of the applications in the world are some sort of transaction fulfillment applications. If there are, fi, I am picking a number, I think it's more than 5 million, but if there are 5 million apps on the iPhone or the Android phone, half of them are trying to get you and I to interact with the engineering algorithm and give some data. They've all those become agentic actions of an Uber agent.
And the question becomes, who sits atop, who's the who?
yeah. And that's also a lot of Google's traditional revenue, at least on the advertising side, is direct response ads. It's not the branding ads, so it, it, it's part of that business model in some sense.
yes, it is. And look, most direct response is lead gen, right? In the in marketing speak, it's lead gen, which eventually results in a transaction or fulfillment of information request. So at some level, it is a precursor to a transaction. People pay a lot more for the consummated transaction than for the lead. So maybe the business model transition is stop giving me leads, give me consummated transactions through agents. Maybe I'll get paid more to buy you the airline ticket directly than have you be able to find an airline ticket provider, which is advertising versus transactions.
I think the opportunity is there from a business model perspective. But I think before we go back to that stable world where these business models have transformed, we're gonna go through a very disruptive phase. Where a lot of these apps will be rewritten and some of the some of these apps will have to question, are they direct consumer apps? Or are they a APis that AP or perhaps run CP client server interactions? We don't call them APIs anymore, which will actually consummate that transaction.
Who do you think is most vulnerable?
Wow, you guys don't ask like simple questions. You guys, like, go for the jugular on every one of them. Kudos. That's why you're such a good investor.
yeah, it's, uh, it's, uh, we're. We're looking forward to your insights on this topic. So
No, I don't look. I think the most vulnerable people are where there is poor loyalty to the UI. If it's just a s sort of thin front end to effectively a transaction processing system in the back. Do you care who you get your ticket to India? My wife is coming back, who do you care, who get your ticket to India from? Which airline? Or which? Which travel entity or which travel UI? Perhaps you don't, you just want a ticket that allows you to get on a plane and get to the other side.
Do you care who books your reservation for a restaurant? So at some level, where you're at the end, you're effectively a transaction processing interface. Where you build some degree of brand loyalty through a whole bunch of sort of non product experiences, I think those become vulnerable because
There there are, um, like, several more controversial ideas that OpenAI is trying to prove out. One now is the scalability of consumer subscription, and another, I think, is
how did that work?
the it it. Well, it's just like, can you? I? I think it's actually quite surprising how many people are paying subs for this intelligence today, right? I think the other is actually, and I wanna talk about the B2B side, is that you should get paid for thinking harder and solving harder tasks in like a a scalable way. This is what they want, they wanna sell, like work right to businesses. What, what?
What? How do you What do you mean by
react to?
work to businesses?
I? I think there's a view that the traditional way that you sell most enterprise software or products is like, it's a seat unit, or it's some sort of like volume unit, like an appliance or something, or a coverage yes, throughput unit, right? based, yes, And here, um, throughput, traffic, et cetera, um, here. The view would be like, well, if I solve a really hard problem for you, I have a unit of work, that's essentially translating to a unit of compute or sort of charging for value.
And and so I think there's a strong belief in some of the labs that they should be able to charge for that. How do you react to either of those business model ideas? Cause you're now at Palo Alto in the business of selling direct value versus ads.
yeah, okay. So are we pivoting from consumer to business now? Are we? Are we
let's talk about that.
because we? We go away from the subscription, because we went off on the whole subscription idea that people should pay for subscription. Look, I think that is a bigger leap. The bigger leap is in the consumer world. We are much more tolerant of inaccurate answers, sometimes, or not perfect answers. I mean, how many times you go to a search even today, where you're looking for something, you don't find the right answer. And you say, Well, let me look again.
Oh, I, I must have asked the question wrong, let me ask the question. You probably do that in your prompts, in in sort of ChatGpt or Gemini, pick your favorite, you know, LLM. But in the enterprise world, there is not that tolerance for an inaccurate outcome, especially if you get on the agentic world. Say, oops, sorry, I made a mistake, I meant you to turn off that server, not this one. Oops, shit, you know, I blew up a whole bunch of my enterprise because you made the wrong, gave me the wrong answer.
Dear, you know, friend, LLM. So I don't. I don't think we're gonna. I don't think we're there yet, to be fair, where there's none of us are giving autonomy to any form of LLMs to create any agentic task or do any work for me. We're all using them with human in the loop for suggestions, and we're still sort of using the use cases where we are okay with multiple answers, where we're actually having the humans be.
It's almost like a glorified assistant or a better assistant, or somebody who's sort of a knowledge worker who's summarizing a whole bunch of information that I may not be able to have from the outside. I don't think we're getting into precision tasks and we're getting into precision actions yet. So look, if you can generate precision tasks with accuracy and precision actions with accuracy, yeah, sure, maybe you can charge me for, you know, the work. But I almost feel in the enterprise, we're gonna go back to some version of redefined, for lack of a better word.
Let's call them AI as a service, instead of SaaS, perhaps AI, as uh, right, sounds not It's a as good as racy term. Yeah, yeah, SaaS. Wasn't a great term either, but sure, AI as sounds even worse. But anyway, it's like, well, imagine that, because you literally have to design the enterprise workflow end to end and see how I can do that from an AI 1st perspective, right? Perhaps like the likes of cursor or vibe coding apps today.
They're trying to at least look at a w part of the developer's workflow and say, here's what you do. Here's a bunch of tools that are can help you through that journey. You're human, and I'm gonna see, based on how you use it over time, I'm gonna get smarter and smarter and be able to take over more and more of what you need to get done. So I think it's kind of like, it's kind of these are AI apps under training.
When they grow up, they're gonna take over more and more of our tasks and allow the repetitive tasks to go away. So you can apply yourself to new, unique problems. But I think there's a long time coming.
If you if you look at a lot of the platform shifts that have happened in the past, so for example with Microsoft OS. They eventually bundled on top the office suite, right? Those were applications that were running on top, they ended up rebuilding or buying them, bundling and cross selling. If you look at Google as a platform, it forked, integrated into the biggest areas of vertical search, right, travel and local, and a variety of other areas.
hm. But Um,
they've been done in workspace too, right? You know, I was beginning to see Gemini show up
doing a workspace.
all over the place, Gemini show up. And I thought, Yeah, sure, and And so actually, right, so
coding is a great example, right? OpenAI tried to buy Windsurf, Anthropic has Claude, people are trying to fork and integrate there, Claude also, or Anthropic also mentioned they want to fork and integrate into financial tooling right now. Do you view this world as basically the platform? The big foundational app companies are likely to try and move into the biggest business verticals directly over time?
that's an interesting debate. So let's let's go back. Z. Why? Look, the reason you're seeing that? Developers is the 1st port of call, because the more they're, the most likely to experiment and to be able to work with half baked outcomes. Like, you give me 75 of the code that I need, I can parse through it and figure out the remaining 25, because that's what I do for a living. It's much harder to do in other professions, where we're not fully sort of in tune with all the guts of what we're doing.
So I think that's an interesting place to start. And over time, that workflow will go into testing and other parts of the software development life cycle. So I think that's kind of interesting. I've had this conversation specifically with some of the people who are driving some of these larger LLM businesses. And you know, I remember the very 1st phone call I made to to Thomas and we talked about small models and large models. And you know, he gave me good advice, like, don't, don't, don't chase a cybersecurity model or don't chase a small model, because these large models become so much smarter that the smaller models will not be able to as smart, be as smart as these things.
So so I do believe, Have a sage. Yeah, Yeah, yeah, that was very good. And then so we decided not to, you know, there was a member like the very early days. One of our competitors announced, say, they're gonna work on a cybersecurity. LLM. Now, if I go back and look, I'm pretty sure that sort of that was a great announcement. I don't think anything's come out of it, so we decided not to chase that because it didn't seem to make sense.
That was pretty sage. I do believe that after a point in time, and I think you and I talked about this right before doing this. You were giving me great insight that the models are converging, their reasoning capabilities are converging, so if you believe all these models are gonna be extremely smart, but somewhat similar in capability. The question I and I always say this to people like, look, in the enterprise world, you know? Getting the smartest model in the world is like hiring the smartest PhD from the best university you can find in the world.
For that PhD to be useful at Palo Alto, we still have to teach them our ways, right? Because they're not gonna be useful. When they walk in the door, they have to understand our context, they have to understand how we do things, they have to understand how to take our problem and solve the problem. Now, whether you call that, you know, pharmaceuticals, you call that genetics, you have to make the model get smart about genetics, very smart model.
But you have to give a lot more training data in a particular domain for it to become really smart in that domain. So I think the the interesting opportunity will be, how can we take these models and apply them to domains where we have proprietary data? Now? Developer use case is a it's a generic use case. There's enough code out there in the public domain that you can actually get smart 90. If you're smart at coding, there's not enough genetic data out there in public domain, or pharmaceutical drug discovery data out there in public domain, or cybersecurity proprietary data out of the public domain.
So the question becomes, how do you take these models, how do you take that brain, apply that to a domain, make it really smart? I think the challenge right now is that if you're building a wrapper effectively as an AI as a service company, and all your wrapper does is enhance the capabilities of a model to put some guardrails around it. Then your biggest risk is the model slowly expands into those capabilities and you're no longer in business.
Now. The difference that those companies are going to have from others who might survive are, if you look at every SaaS company, it's actually the packaging of a workflow, the system of record, right? Well, I mean, my HR system is connected to the workflow, every employee knows how to use that workflow. And then it every time it locks down a certain table and say, this is your system of record. This is what Nikesh gets paid, this is when he took holidays, this is what his equity looks like, this is when he vests.
It's less about the app, it's more about that data and that system of record. So eventually, if these apps have to live in the long term, they have to marry the capabilities of AI. With the fact that it becomes an enterprise system of record, right? A model is not going to become my system of record. It will still be some proprietary log database somewhere, or some data tables, pick your pick. And maybe the interaction mechanism is no longer a workflow, is some sort of agent or some sort of AI interface, which allows that system of record to be maintained and created.
But there still has to be some rules. So I think that's kind of where the opportunity is, as compared to just putting a wrapper around a generic process. Saying today, I'm going to help you analyze legal contracts. Well, guess what, you know, I ran my legal contracts through ChatGPt works just fine.
Where are we actually given, um, your visibility into enterprises and actual adoption or value in use cases?
So I think the the use cases where there are two, two current major use cases, right? One, let's call it what do you call it, generalized, or perhaps cross enterprise, consistent activities, right? generalized. So do I have a legal team? Yes, I do have legal teams. Every enterprise has a legal team. Yes, do they have any particular proprietary knowledge compared to, you know, particular Palo? Unlikely. It's more I need them for legal advice, not for Palo Alto advice.
So in that use case, yes, could we use a, you know, Harvey equivalent or whatever? Those are sure, it enhances their productivity, they get their 50 faster. Could I possibly in the future use some sort of AI based interpretative app or interpretative, uh, yeah, application, which helps me process my accounts, receivable accounts, payable faster, codify them? Sure, so I could. So there's a whole bunch of repetitive, generic tasks across enterprises, which I'm pretty sure could be done by some version of an AI wrapper around LLM with some particular context or my data.
Sure, so, to that extent, I think we're all experimenting with those things. But my caution to my team is don't try and build them, somebody's going to build them for all of us. It'll be much cheaper to rent them by some, perhaps metric AI of work. Yes, ass. Yes, yeah, as a metric of work, or or per seat, maybe agentic seat, I don't know, but there'll be some mechanism that they'll charge us on. But we don't have to build it because it's going to cost me a lot more to build my own.
You know, accounts payable, accounts receivable, smart AI system compared to what I can buy off the shelf.
Is that what your customers believe now? Like all the enterprises, largest ones I think that,
many of them do, because this is not an easy problem to solve. Like, 1st of all, finding the skill set, finding people who understand this, you know, living in this world of constantly evolving models, where, if you keep an, you know this better than me. Like, there's no, you can't take one model out and take the next version and stick it in. It works just the same way. This is like getting a new PhD and training all over again, saying, let me explain how we work here.
So from that perspective, I think most rational players in the enterprise space, as in customers, would want somebody who's the expert to build it. And for us to have some version of adaptability or adaptation to it, make sure it's secure. Like none of us, no enterprise customer wants their data to be floating in a multi tenant environment saying, Oh my god, my data is training other people's data. Now, to the extent it's the cons, available, have a good time, right?
You know, you can understand how I codify stuff, have a good time, but to the extent it's proprietary data. When I'm doing FDA trials, I don't want my FDA trial data training somebody else's data as a customer. But I think they will err on the side of caution and say, I want my instance to be secured. So I think we spend half our time before we look at any of these packaged AIS apps, talking to them, understanding the security, like, I don't want my source code to be training somebody else's coding app.
So to that extent, there's a whole bunch of conversation. Is it ring fenced? Is my data mine? Are you using it to train your model? Are you using the trainer system? Now? Yeah, and then we spend a lot of time testing it to make sure that they're not doing it. So a lot of time and effort is spent there. Not not every enterprise is as discerning. We have to be because we're in the security business. But I think you'll see some version of stability and an acceptance there that people will take.
These generic sort of AI as a service, system of record, or sorry, AI as a service apps, which are helping humans get better.
yeah. How do you think about that? In the context of UM applications, you think the may make the most sense on the for cybersecurity? So if I look at founder activity, there's more and more activity around SOC, there's a lot of activity around pen testing, there's, there's activity around a lot of areas that are very human intensive, in some cases repetitive tasks, which make a lot of sense for this form of generative AI to take over.
And then there's people incorporating AI into existing products like Socket for sort of like a Snyk like competitor. Other aspects of UM code security. I'm just sort of curious from your vantage point, what do you think are the most interesting areas of cyber AI?
So I think if you, if you step back and think about cybersecurity, right? There's a world of cybersecurity which operates and says, that's the known bad. I found it, let me stop it. Sure, that's a good thing. I found a bad actor, let me stop it, I found malware, let me stop it. Now, to be able to stop bad things from that that are getting into your network, you have to be deployed at every sensor. So the 1st thing a cybersecurity company says, Look, I can't stop what I don't see, so I have to be present at every edge, every endpoint, every sensor of your big organizations.
Five years ago, we made a conscious choice. Our strategy should be to get to be in as many sensor places or control points as we can. So we did that. You know, we have, you know, a Sasi product, we have an endpoint product, so that's good. I think sensor business will have to stay, because if you don't, if you're not there, you can't find anything. It doesn't matter, AI or an AI, I got to be able to be there to find it.
And then sensors are pretty good at stopping the known bad. If the known bad, I stop it well. Most cybersecurity breaches happen because of the unknown bad, because we stop all the known bads. And every time you build a new company, it's like, let me go. So everything you're talking about is trying to find the unknown bad or a vulnerability which has left the door open so bad guys can get it. Whether it's a socket or a Sneak like competitor, that's what you're trying to do now.
In cybersecurity, there are companies which are the sensor now. The sensor allows you to do two things one, stop the known bad, but also collect valuable data to analyze the data to understand what the unknown bad may be. So you get benefit from both. We are at sensor stopping the known bad, and we also collect a lot of data. Traditionally, in cybersecurity, people have been, I'll call it feature integrated from end to end. Oh, let me sit on the endpoint, I'm gonna trap all the data around a particular topic, I'm gonna take that data into the cloud, I'm gonna analyze the data and then tell you, Oh my god, I found something suspicious.
Here are five suspicious things you should investigate. But the problem is because I don't have context, when the data passed my sensor, it's gone. So take a simple example I send you an email, a phishing link, where you click and you know, when you go to the website, I steal your credentials. Now I'm an email security company, so I stopped. I, I'm at the bed, I see a bad email, I know how to stop it, I don't know, it's a bad email.
And then Elad clicks on it. Now you've gone from an email product somewhere else in the company, gone to the internet through a firewall. I have no idea what he did. So all I can say is that link looked suspicious. Elad clicked on it. Maybe you want to investigate now. No customer wants a list of five zero things they have to investigate. So people are busy building these agents, saying, well, let me build an agent to help you investigate this, investigate that.
But I think the better answer is, if I had all the context of the enterprise, I can go mine that data to find what actually happened. So we're taking the different tack, we're saying, let's see if we can consolidate all the data in the enterprise. If we can, we can run a whole bunch of machine learning algorithms and do all these activities on top of that. The the current startups you mentioned are like startups that are trying to do what AI wrappers are trying to do around LLMS.
And over time, we're gonna get better and better, where we're gonna squeeze their capability over time that you wouldn't need.
Yeah, yeah. And there's a few different sets of them because there's also the ones that are just trying to automate human services that are associated with the security world today. And so pen testing would be one example where you often hire external consultants to help with that. Um, SOC may be another one where effectively you have this, you know, you're. You're basically looking at at the operating center across all the different events that are happening and trying to automate that.
So it's a little bit more of a So the reason human in the
you're trying to automate the SOC from the outside in is because you're not running 10,000 machines yeah, and you're saying running your models
keep the business side out. Yeah, at the ingestion point. Why hm, hm,
should I 1st collect all data, analyze it after? I should analyze the data and cross correlate it? Ingestion? To make sure I understand the bad things and I put them up there and then say I am gonna, and then build agents to investigate the bad things.
well, this is. This is my my question of, uh, uh, no. The the environment in a large enterprise today is still very fragmented, right? They're just doing this, you know, past the hat trick that you're describing with all these different tools. Hopefully they're very dominated by yours. But as a lot
You mean, chosen by customers, mentioned, yeah, I'm. I'm using the bad word. chosen, If you've worked once in a large company, you understand, gently chosen, gently chosen, gently yes, gently chosen. Yes,
chosen in a highly competitive environment,
that's right. Yes.
of course. But uh, but like, you know, I was talking to a friend who runs security at a large financial. And he's like, we have 100
So do you want some and he, cybersecurity?
he's definitely got some Palo Alto All right, go, all he wants more,
right, appreciate it, we'll talk.
but he's got 118 identity tools, right? Like, I didn't think there were 118
he has an, uh, identity crisis. tools. There well we go, there done, you Yeah, you
we go.
know? gotta try. But So you bought an identity company, you just told that to, you know,
um, maybe the answer is, uh, just make it all cyberark. But I I think his view is like, it's like, just untenable for me to consolidate all of this in the near term. And someone attached something after, which does the human consolidation, like the Slack automation thing, right?
The interesting part is we're the youngest industry in technology because cybersecurity did not exist before connectivity came about. Nobody had to like, you know, you were on a mainframe, you sat in the office, you connected the mainframe to a pipe into the backend and you're off to the races. So nobody actually was going to come and intercept your traffic. Now, it's not scary suddenly, before the web. Yeah, that's right. So now the web and applications out there made it serious.
It's a 25 year old industry, so every time a bad problem showed up, somebody built a point solution to solve that problem. So it is a fragmented architecture, it's a fragmented environment. There's a lot of data being used and analyzed multiple times by multiple vendors because they were needed for that point in time. But like any industry, that capability commoditizes over time, like there's no genius in building a firewall, every firewall within plus or minus. Ten does the same thing, right?
Some things do it better. The question is, what do you add on to it? What all capabilities can you do, build beyond that? So if you believe that, sure, Nikesh, 80 of the capabilities are going to be capabilities that even most companies still don't have. And they can be delivered in one platform. On the other 20, we can put add ons, but over time, you know, we will eventually come and gobble those up because there'll be a new 20.
Until one year ago, nobody was talking about AI security, right now, there are probably more companies than you can count who will show up at Elad's house. Say, please give me a you, so please give me some money. I'm starting a cybersecurity company for AI. I do prompt injection, I do malware checking on on models, I do model data poisoning, so agentic uh attacks. None of these things existed. So, of course, you'll find people building features to protect against those.
Because we were too busy still fixing the problems of last five years, where the mass market problems are. So I think over time, what you'll see is the the platform approach is gonna sort of eat into these feature approach that's happening. And you know, there are companies with cobol code out there as well. So 118 I vendor sounds like a lot, I didn't know there were 118 vendors, to be honest, I didn't. He might have meant cybersecurity vendors, which I can believe, which is also a lot.
Uh, what new problems from AI do you actually pay attention to? You say like, this is gonna be a mass market problem.
Look, I think, if you, if you back up, and you know, we eat our own dog food, believe our own rhetoric. If you believe your own rhetoric, then I, as a bad actor, should be able to unleash agents against an enterprise, against every aspect of it. And figure out where the breachable parts are, or where the holes are. In a quick, in a matter of minutes, or, you know, less than an hour. And I should be able to point my attack towards that vector.
I could should be able to run simulations on how should I attack this thing, and I should be getting exfiltrated data. Now, you know, when I started seven years ago, the average time to identify a target, get through it and exfiltrate data was in the three to four day timeframe. The fastest we've seen, I think right now is 23 min. So if you're if the bad actor can get in and out and exfiltrate data, or shut down your endpoints with ransomware in under an hour.
Then, by physics, your response time has to be less than an hour. The average response time is still in days. So from that perspective, the biggest threat that AI brings is that it continues to compress the timelines. To be able to come, you know, either shut down your business, cause a compromise, cause ransomware, cause economic disruption. If that's what it is, I think the pressure just went up higher on our customers to get their their infrastructure in order.
So that's the that's the risk and the opportunity.
yeah. I I think a lot mentioned pen testing, which hasn't traditionally been like a very strategic part of the security Pen testing, Yeah, landscape.
pen testing is just to knock down on every part of your defense. I think half the companies don't do pen testing because they're scared of what they'll find.
they're doing some minimum compliance level, but I think from a technology perspective, to your point, like, what is pen testing? is trying to attack the area. There are companies like Run, Sybil now that do this. They can do it continuously in the 23 min We, you described, and I'm we, we like, that's exactly what run, we an attacker would do. run seven by Yeah,
24 by 365 at Palo Alto. We don't have, we don't hire 3rd party people. So what you're talking about as a company, we've done that as a default, because that's our existence. If we get compromised, we get breached, we have a problem. you mentioned email. Um, Yes,
I think it's like a well known issue that a lot of the breaches they happen because of social engineering, because of email. Because, you credential know,
take over 89 of the attacks happen because of credential theft. Somebody becomes you or me. okay, so
people like us are not getting any smarter. And you know, you have models, Oh, come on, ah, like you try one, one a day, yeah, that's a good right? Atomic habits, point.
appreciate it.
but but you, you know, now, how concerned are you about, like, deepfakes and generated spear phishing, and, you know, voice attacks and all that stuff?
So, to the extent they enable the act of social engineering, yes, those are concerning, because I think most forms of two factor authentication are going to be, you know, out of the window. I still once they reach banking, it's going to go to a call center. But I call them and they say, Oh, can you please confirm your identity? And they ask me three arcane questions, which I'm pretty sure Chatgpt or Gemini will be able to answer in sub seconds.
Because they're they're only s scouring the web to find public information about me and ask me questions. So I think all those forms of authenticating who you are are getting easier and easier to compromise. So the question becomes, so, let's so the. The problem you have to figure out is, you know, you can solve it their way or our way, as in the way they're looking at it or the way we're looking at it. At the end of the day, every one of these social engineering attacks, credential takeovers, eventually initiates some bad activity in the enterprise, and the bad activity in the enterprise often takes the form, takes on the form of what I will call anomalous behavior, right?
Suddenly, Sarah decided to exfiltrate all the data in Elad's company, even though she used to do email with him every day. Today, suddenly she's logged in and she's downloading everything onto a laptop.
This actually happened last week. Thank you. Yes, Yeah, It is, great deal flow. Yeah, is there nothing yeah, just trying to steal everything.
suspicious in that? That sounds pretty suspicious. So I can spend my time trying to make sure that nobody can take over Sarah's identity. Or I can make sure that if s, you know, Sarah doesn't act anomalously, and if she does, then I throw in a block at that point in time. So I'm looking at life the different way. That's why we're buying this identity company, because identity companies today say, Oh, I checked you in the door, you're fine.
And now you can roam anywhere in my entire enterprise and do whatever you want. Because you were, you were checked in at the door. Well, that kind of, that doesn't work anymore. I have to make sure that I like now today. Compute and data and AI is going to allow us to analyze all the anomalous, anomalous patterns. And if I could say that, that's pretty weird. She would. She's never done this in last seven years. Why is she doing this now?
Does she have the rights to do it? And I could do it just in time, rights, I can stop you from accessing this stuff. So you have to change the name of the game, you can't do it the same way. So the whole idea of us buying an identity company is to think about life. Saying, stop giving people persistent rights, give them just in time rights, give them rights, which are analyzing for non anomalous behavior. In fact, what I want to do is I'm like, I actually don't need you to give me a 2nd factor authentication.
I can see the way you type, and if your identity starts typing differently, I'll block you. hm hm. One So the
of the things you've done incredibly well at Palo Alto over time is really starting with the core set of products and then expanding and really providing both the platform and the add ons that you mentioned. Was that something that you came into the company knowing that you wanted to do, was that something that you ended up adopting over time? I'm a little bit curious how you thought about that puzzle.
only that you and I have worked together before, and I remember, you know, you had Google, a smart young man, and his, that hasn't changed, you know? I came to Palo Alto and I just analyzed the business problem, you know, the product. So, you know, I came with two things one, I understand business two, you know. Larry told me many years ago that if a technology company loses sight of the product, it decimates over time. And that's been true across technology, you can pick your take your pick across the, you know, the the tons of companies which haven't made it.
The business problem in enterprises eventually, if you look at a enterprise company with less than $1 billion in revenue, 50 to 65 of the cost is cost of sales, marketing and customer support. Which leaves no room for margin. If you look at the largest enterprise companies, that number goes to 30. So actually, it's all about taking that 70 and bringing it down to 30. Because R and D GNA don't change a lot. Fast, a billion to 10 billion or 100 billion.
You still maintain twelve to 16 of R and D cost, and you still maintain, you know, if you're, if you're like efficient, like some of the large players in the market, or four GNA, or you're at six or eight. So your maximum leverage comes from sales and marketing, customer support and marketing. And you realize, well, what is that? That is the ability to convince one customer that you're really good at what you do and be able to expand in that customer's environment with that trust that you can do more and more things for them.
So that's the insight it came in with saying, why is it that, you know, Sarah's friend has 118 vendors, who's in their infrastructure? Because each of them gets vetted, VOC, this is security. It's not like saying like buying some like random thing on the side
there's a big procurement process and
and there's a testing process, right? Because you could be a bad security actor and could take my whole enterprise down. So if you go through that entire process of validation, verification and trust, why shouldn't I take that right that the customer has given me, earn their trust and expand my capabilities across the platform? So that's the insight we came with and realized, you know, we walk in, we say, sell firewalls, Hey, you want to firewall? No, I'm sorry, I'm good for the next five years.
Oops, I had a sales guy, he was focused on the account, that was his target. He spent three months getting a meeting, going through the process. Now he doesn't have anything to sell him. So I'm like, Okay, you don't want a firewall, you want this, you don't want to get this, you want this. Eventually, enterprise salespeople have to have enough of an offering set that allows them to be able to sort of present something to the customer.
The good news in cybersecurity, there's always some transformation going on because some technology is coming to its end of life form. And the company hasn't innovated. And you're off to the next set of companies, so that's the insight that came
that's a good, that. in. Yeah, it's a great insight. And if you actually look at the margin structure for some of these things that you mentioned, sales, customer success, et cetera, or support. A lot of those things now have AI apps that are, uh, making people dramatically more efficient. What do you do? You view the future of those sort of functions being very AI enabled and AI heavy? Or how do you think about that transformation across companies like Sierra, Deckagon, Rocks and others?
Something that I think I think,
you, I, I'll answer the 1st half, the question, the 2nd half is your domain, and Sarah's domain. I'm gonna leave you to answer that question for him because I don't know the answer to the other companies. But I think, look, if you, if you fundamentally look at it and look at organizational efficiency, perhaps perhaps the best word. People are scared of AI based efficiency. I seriously doubt that an AI agent will convince the CIO or CEO faster than my human that goes and hangs out with them and shows them the product.
So I think my sales teams are very happy in their existence, they don't believe they're imminently threatened by AI. So that's a good thing. Interestingly, on the product development side, I think people will be,
Just to come back to that, sorry to interrupt you. But uh, there's other forms of sales enablement, you know, uh, a deck per customer that you customize, or, yes, but those are uh, sort of SDR, marginal. sort of, Yeah, those you know,
are marginal. That's process efficiency, which will come through it. But, you know, guess what, before you get there, you're probably gonna need, you know, 15 AI savvy people to build the workflow. Because an AI model is not just gonna spit out a Palo Alto Sasi proposal by itself, it's gonna be have to be And do trained.
you think that's customer segment specific in terms of leverage you can get? So a lot of people think, for example, the SDR level seller, where you're sending a bunch of emails out, or you're sort of cold calling or doing things like that. Where it's a s a lower ACV customer, Yeah, but is that I think that's a
only one side of it, right? Like, if you stick to SDR, we'll go back to the other problem in a 2nd, but stick to SDR. At what point in time does my email start getting read by an agent, which is called the block all SDR, block all s people trying to sell me shit? Because I'm overwhelmed by the number of people who think they should write to the CEO because they have some company which has five people automating something.
So I think there's a
that's interesting. Does that actually decrease the effectiveness of the sales teams over time? Then, in other words, if effectively you have agents screening things, does that create a block for certain types of sales leads?
well, I think the question, the sales lead, is the isn't the issue, right? The question is generally either I have a need, I know it, in which case, hopefully my blocking agent knows my needs. And eventually says, Ah, you know what? Actually, we've been looking for this thing. And here's an email which satisfies our thing, so maybe that'll do it. Or, uh, as you do in marketing, you never need a new watch, and you never need a new car, but you just buy it.
Because somebody put in front of you. I don't think it happens that way in enterprise, so we can't generate demand for something we don't know we have a need. But sometimes you can in cybersecurity. So look, I think those will be marginal efficiency outcomes, because how many SDRs do you have? Eventually, you do the selling, you do lead generation. We're a large enterprise business, we go through a large, rigorous testing process. So, yeah, I can get a phone call, but if my product doesn't show up, then it doesn't matter how good the lead was.
it's interesting because I hadn't thought of it as just like becoming, you know, if we have agents doing this sort of low level communication. 1st, it just could become a more efficient market for finding the problem. Match, right?
Yeah, maybe. So I think if you go back to like, the organizational efficiency question, yeah. I was talking to my CFO this morning. He was like, when are you going to take a whole bunch of the stuff that you're doing with human beings and replace them with some some form of, you know, at least 50? Agentic analytics or apps? So I think we're going to get there. I think the Max replaceability will be in, let's call it administrative areas.
I have 200 people doing documentation. Do I really need 200 people doing documentation? If models can do 90 of the work, I probably need AI savvy people to write the guardrails around these models and templates, and they can produce them. So we'll find some efficiency, you know, pick your number, 200, 300, 400, based on points of efficiency. The larger you are, the better you are, because you're gonna make, uh, save more money. I think sales doesn't change as much, I think product innovation is the better companies will innovate faster.
I have technical debt. I'd love to get rid of a whole bunch of Salesforce code or a bunch of SAP code that I have and get a more efficient. I can't find good people to go do it because nobody wants to work on it, right? They're all working on AI. So can I get some version of, you know, agentic AI apps that are going to get rid of, you know, a few hundred of those people? Great, so I can get inefficient product development ideas out of there.
But I'm not going to let go of any good product person because I'm going to get them to move faster because that's my competitive edge. So I think we won't see as much attrition in the vibe coding outcomes or the product development outcomes, we'll just see faster innovation.
So basically, the statement of AI displacing all of human labor, and, uh, that's very overstated from your perspective, relative
I think on those two segments, I think customer support will have to go through the revolution. And what I mean by that is, I always joke internally, I tell my people customer support exists because we build bad products, if you have great products, why would you have to have customer support? It's complicated, it's hard to onboard, it's got too many dials, and that's why it takes me so much time to make it, and then eventually it's not efficient.
So we have most recently moved customer support closer to our product teams. I told my product teams the day you have a bug or the day you have a customer issue because of the product issue, fix that before you build a new feature. So I I think, in concept, from a North star perspective, we should be able to take 80, 90 of customer support out in the next two to five years across the the landscape. That's what every good company should aspire to do.
And when you take it out, it means your product quality should get better. If I'm really using a good vibe coding agent, why shouldn't it write better code? If it's writing better code, why shouldn't it find flaws in my code much sooner? Why shouldn't it run simulations? So I'm expecting product quality gets better, I'm expecting the fixes come faster, I'm expecting that, you know, we can diagnose the customer's problem with data, as opposed to humans calling, asking all those questions.
So
I think there's a lot of anxiety from engineering leaders that the product quality is not likely to get better. There's just going to be way more of it. Because
more bad quality product or more more, pro?
just more product, right?
That doesn't mean it has to be bad product.
That's fair. Yeah, I I think the the the combination of generated code that people are not fully understanding from an architectural point of view, fully reading and reviewing. And that volume overwhelming. Like the processes that we have today in software development is an anxiety and Well, let's, quality team's let's, let's, let's, biggest facet of it.
let's parse that for a 2nd. It is the worst right now that it's ever going to be. It's only going to get better. So what's
you're the most optimistic security person I know, that's great.
is that? Am I restricted to being optimistic in security? Well, I think you have to that's not like a you have to think
super, it's not a super optimistic group. I'm just saying, Look, I think yeah,
at the end of the day, being one of them, we have to do best, we have to do the best job we can, and eventually life takes over. So it's fine. But I think from a from a quality perspective, anecdotally, I've seen examples which have huge promise. I've seen the vibe coding agent find a vulnerability in security code at Palo Alto, which we wouldn't have found unless it was out in the wild. Which is a good thing for us.
I've seen it take 500 lines of code and and come back with 75 lines of code, which would be much more efficient in terms of and in doing the task that 500 lines of code is doing. I've seen it come back with code explainability, which was written 15 years ago, we can't find the person who wrote it. So there are some amazing examples of what the art of the possible is now. If that's just what we get today, I think one, two, three years from now, that's just going to get better and better.
So from that perspective, do I believe product quality gets better? 100? believe that product quality gets better. I don't think there's any debate on that topic right now. It doesn't look there is no solution for stupidity in the world. So if you don't have good people reviewing this stuff, then yes, you can end up with bad outcomes. Not because AI is bad, it's because you didn't set up the right guardrails, the right process to do it.
So I think we're going to get good quality outcomes, I think we're going to see better and better capability. I think the most vulnerable area in any enterprise are large amounts of humans doing repetitive tasks, which are either generic, which is the easiest to replace because everybody's doing them. We don't need to do them. Or even if they're specific and have critical mass. If I have two zero people in customer support, I should be able to optimize a lot of that and hopefully deploy those people in much more meaningful things.
I mean, imagine I have a child, you have children. Do you really want them to grow up and become customer support people? And so every time, what do you do? I just pick up the phone and listen to somebody grumbling on the other end because something they bought is not working. That sounds like a really horrible job, so I think those jobs should go away. What's wrong with that?
And they don't have the power to fix it.
Yes, that's worse. It's like, you know, it's getting punched in the face and not being able to punch back. So if somebody's taught them, customer's always right,
can we talk a little bit about leadership? You're a very unique leader. So this is a time of, like, at least a handful of companies growing very quickly. Um, because they create it's a game to some.
trades in billions of dollars.
Y, Yeah, yeah, uh, well, and like, I, you know, an optimist myself, I'd say, like, in AI, wrappers or not, they're creating a lot of value, and consumers and enterprises are buying very quickly. Google from 2004 to 2014, Palo Alto, seven years in, you've added, like the size of a Palo Alto originally every single year since. In terms of enterprise value, that's wild. Like, what, what makes you a great leader in terms of growth at scale?
Like, what do you? What? What advice do you have for some of these people who are like, Guinness Book of World Records in terms of growth? The 1st few years?
no, I look, if you step back, it's interesting. Every business that you've identified or you've looked at, or we've talked about, has a much larger TAM than any of these companies is able to touch. The the the markets are growing, you have the opportunity to take share and grow in that market. So I'm a huge fan of growth businesses. I hate the idea of going restructuring something which is on a declining curve. It be sort of, you know, scaring, so it's good to find the right market from a growth perspective.
I think, um, I always have this principle that nobody wakes up the morning and goes to work to screw up. Nobody wakes up saying, Oh shit, I want to have my worst job possible, no chance in hell. You know, you can find people, you can get a group of people together, they can be innovative as hell and go put, you know, a rocket into space faster than NASA. These are all humans, they're all people out there.
There's no difference in many of them than people who work at Palo Alto, the people who work at Google or elsewhere. So what creates the difference between great companies and companies that are not, uh, as good? Because I'd say, within reason, the people you can find those people in every company. I think it boils down to understanding the market, setting the right north, star, getting enough buy in. Talking about the why, not just the what you need to get done, and getting people really excited and bought into it.
And then making sure they have the resources to get their task done. If you do that, then my job is just like, set the strategy, set the north, start, put the right people in place. And then basically act as their shield and keep blocking. Bad things or friction from slowing you down. So if you can do all those things in a way, you know, there is a high probability you can create good outcomes. There's no guarantee.
Are there any unique structures or approaches or tactics that you do that go against the grain? Uh, we've talked to Jensen a few times and he's pointed out, for example, he has like 40 direct reports, he doesn't do one on ones, you know, it's a I actually bit of a
read that, I actually tried, I actually expanded my staff meeting from eight to 25. After I read that let it go, book, and said, it's interesting, and it solves a different problem. So, at least from my case, I've discovered that I'm not always sure that these people communicate the why's to their teams. That at least eliminates one level of confusion. So why does he want this? Oh, actually, I heard him directly. This is what he wants to get done.
Because sometimes, you know, you have this notion of you have to be a player, you have to be a coach, sometimes you have to be directive, sometimes you have to be, you know, encouraging. It's like, we're going to climb that mountain, we are going to climb that mountain, and it's like, you know, if you go back, so we're going to climb a mountain. Promise people will end up on that one. So it's important for people to understand the communication part.
So I've discovered that communication actually is underrated in organizations. And usually, the way I do my sort of call it 360 degree test is I meet 50 employees every two weeks and I ask them questions. Then I discover, Oh my god, these people are asking questions which I thought was abundantly clear. Why we're doing this, what we're doing, this is what I discover eventually, that by the time you get to the person, four, five, you know, levels removed from you, they actually don't understand exactly why we're doing certain things.
They have fundamental questions around what we're doing, and that causes them to do it differently or not do it. So that becomes sort of an issue of a communicator. You asked me about, you know, what do we do, what have we done differently compared to other people? I think communication, talking to people, making sure they're all bought in. But I think from a business strategy perspective, we've taken a very different approach to MA. I think we bought 27 companies so far, and about to buy our largest one if we get approval to get it done.
And I don't call it M. A, I call it product development and research in a highly innovative market where all of you guys are kind enough to support, uh, innovation.
it's distributed R. D. Yeah, distributed R. D. We We're in the
get to.
service for you.
Well, you guys do reasonably well for that. So, you know, you can say thank you, and I can say thank you, and I say, it's like, if you don't win and I don't win, if you don't win, I don't win. That's fine. So we're happy to be in a situation where you get to a certain stage and we get it past that stage and take it to scale. Because I, I don't believe we have all the smarts in the world.
There are lots of smart people out there who are trying to solve different problems, and I don't believe we can solve all the problems. Simultaneously, take AI, you know, we started off saying, Oh my God, when AI gets deployed, enterprise is going to want to make sure that their AI instance is sequestered and controlled and managed. Because they don't want data leaking, they don't want, you know, external inputs. So we built effectively what we called an AI firewall.
It's great. And we discovered this company, you know what people are actually trying to figure out these models that they have are hackable, have malware in them. So we weren't doing that, we were just protecting them once they were in there. So there's a company called Protect AI, which was actually assessing models, they were doing persistent red teaming against the AI models, specifically, not just the enterprise. Because, you know, going back to your comment on red teaming models, morph, their responses are non non predictable, non deterministic.
The same model could answer the same question one way today and could answer it differently one week later because it learned. Now, if you start getting non predictive responses, you have to inspect all the responses to make sure none of them is malware, right? So from that perspective, we do persistent red teaming of models, we do scanning of models. So we said, Oh shit, we haven't done it. We found this company and we said, thank you very much to the VC community for delivering that from an R D perspective and putting them part of our platform.
So we do rely on, like you said, R D as a service, uh, from the VC community, and that's been very helpful. But we always go for number one or number two. We never believe that you can take three or four and spit and shine and make it look like one or two, because one or two don't go away. They trade as one or two for a reason. Uh, we more often than not make the leaders of the acquired companies, the leaders of our business.
Because we believe they've acted faster than us in a much more resource constrained environment, shown tremendous amount of resourcefulness and hustle, they will deliver the outcomes and innovation. So I think from that perspective, we probably have a higher hit rate in the industry than most other M. A that has happened in the enterprise space. So I think we do that differently.
You at the beginning with Palo Alto and continue today, like, have been perhaps more ambitious than any other security company. Uh, I I think that's right. Um, how do you? How do you convince an organization to be more ambitious? Because, you know, you, um, uh, my understanding of the cyber industry before is you had like, endpoint businesses and firewall business, very sort of domain specific, right?
I don't think you have to convince humans to be more ambitious. I think we are natively and naturally ambitious. Like, you know, you meet somebody saying, Do you? Can you do more? I've never heard somebody say, I think I'm done. Like everybody says, I want more, I can do more, right? We live in a conglomerate society, we're all taught to aspire for more. So I don't think it's hard to make people at Palo Alto feel that we we have the right to play at a bigger table on a constant basis.
And people actually like the idea of ambition, aspiration and winning. Like, you know, trust me, if our stock wasn't up six or seven times in the last seven years, a lot more people internally would have questions on our strategy than they have now. So I think it's a sort of self fulfilling prophecy. It's a good thing across the board, but I think more fundamentally, if you step back, our industry is not fully formed. It has 118 vendors, it's fragmented.
Uh, you know, you take a look at the CRM industry, look at the ERP industry, look at the HR industry. These things operate on single platforms, right? Nobody has two salesforces deployed in an enterprise, nobody has two workdays deployed in an enterprise, nobody has two SAPs deployed in an enterprise. Why? Because you need end to end visibility, a singular workflow, a singular set of analytics to solve the problem. Our industry has started off because, oh my god, we have a threat, block the threat, you know, don't worry.
Yeah, so like we were playing whack a mole. So the point is, as I said, over time, as these requirements normalize, they somewhat sort of, you know what, do you say? Converge in the capabilities of the vendors. Then how does it matter if you take from one versus the other? And what? What becomes more important? You know, all these platform companies I talked about, it's not like they have unique features on a feature by feature basis compared to the competitors.
Over time, those have normalized, but what they do have, they have an end to end visibility and capability that integrates the functionality. That's why they're there. So if cybersecurity has to survive in the long term as a mature industry, we also have to become sort of a singular enterprise platform. If you believe that, we're nowhere there, right? We've taken. We used to have 24 products when I came, we had four when I came, we took it to 24.
We have 44 magic, Quadrant and Stoplight mentions. We've turned that into three platforms now. Say, you know, if you're going the journey with us, it's going to take you two to three years to get a platform deployed. We're three right now. Because had we said one, oh my god, oh my god, your friend can't go from 118 to one. It's like it boggles the mind. So let's let's take it to three first, or one or three, and hopefully get into the next one and get into a third one.
So I think the idea from our perspective is if we can become the platform of choice in the industry. That's a very big ambition, very big north star. But you know, it's like, you don't get there if you don't start.
maybe when you look forward, both okay, Palo Alto, cybersecurity, AI. Three questions What keeps you up at night? What do you think most about
I think most about AI, not,
me too?
I'm glad we take part. That'll be something you'll Be working on. I think more about A I. From the vantage point that. If our view of the world how of how this is going to evolve. Is not. Within the guardrails of where it's gonna be. We may end up taking Palo Alto. In a different direction. Because remember. We exist. To help you secure technological advancement in a certain direction before you fully deploy it. I want to give like, you know, today our conversation with some of the big cloud providers was.
How is everybody thinking about agentic AI? I'm supposed to secure agents. The problem is. I can't get one person. To agree with the other person's definition of an agent. I'm like, what's an agent? Well. Are you going to use M C P protocols to deploy? Well, no, we just have connectors. What's a connector in in an L L M? A connector is effectively an A P I call, a microservices call. Why are you using A P I calls?
There used to be called A P I calls in the past. Why aren't you using M C P server client? And clients. Well, we're gonna get there. Right, well. Are you? How are you gonna do the inspection of identity? We're gonna register the identity somewhere else, like. What's an agent? How are you gonna run it? It's gonna be delegated. So there's like so many questions from an execution perspective. From how the industry evolves that. This kind quite keeps me up at night.
We talk about this every day. We have a team of people getting together every day for 2 hours and we read everything we did, didn't talk about saying, what do you think about this? What do you think about that? Because when you don't have an expert, then hopefully the collective wisdom of 6 or 7 smart people, those people who I bring together every week, every 2, 3 times. It's probably better. So We're constantly trying to paint a picture of how the world of AI is going to evolve.
So we're building our opinion. And based on that, we have to design a product. So that's that's extremely bleeding edge. Right? And if we want to be the cybersecurity partner of choice, we have to be able to go with the bleedingest capability and tell our customers, look. You have a problem. We're solving it faster than anybody else. And we can help you deploy A I securely. So that's kind of what, what It's kind of exciting. A little bothersome because, you know, you got to keep understanding.
You got to think, holy shit, were they just here? What do we do with that now?
Yeah. As as you've been thinking deeply about A I, have you thought about it from a broader societal perspective impact of the world? Like, there's are there specific threads that you think about most or worry about most, or optimistic about most?
I'm gonna stick with Sarah's characterization of me as one of the more optimistic people about these things. I want to expand that beyond cybersecurity like, I think it's exciting technology. You know, you've been in Silicon Valley for a long time. You've been in Silicon Valley for, you've been investing for so long. The intensity, the excitement is palpable right Like you can't turn around.
It's really fun again. Yeah.
Yeah, it's really fun again. Right? Like, you're, you're like, you can have dinners, you can debate all kinds of arcane topics, and I'm pretty sure you can find lots of people with different opinions, and there's so many different directions you can go to. You can talk about policy implications, you can talk about rappers, you can talk about L L M's, you can talk about infrastructure. It's like a whole new technological wave, which has so many Implications and majorly disruptive implications.
And from that perspective, I think it's exciting. Does every technology come as a double-edged sword? Of course it does. You know every revolution in history has come as a double-edged sword. So why not this one? You just have to believe in the power. There's more good people in the world than bad people. The good people will hopefully continue to. Make sure that the bad things get controlled. Now, is it gonna? Are we? Is some bad things gonna happen?
Most likely. Are we gonna find a way around it? Most likely, and hopefully. I mean, we powered through a pandemic. For crying out loud.
Awesome. Thank you, Nikesh.
Yeah. Thank you so much.
Thank you for your time. Appreciate
Appreciate it.
Find us on Twitter at NoPriorSpot. Subscribe to our YouTube channel if you wanna see our faces. Follow the show on Apple Podcasts, Spotify, or wherever you listen. That way you get a new episode every week. And sign up for emails or find transcripts for every episode at No-Dash-Priors.com.