(bright music) This year at AI Ascent, we are doing a few things differently. One of those things is the what's next section that we talked about, these visionary ideas. Another is the what's now that Sonia has been talking about, about how are people taking things in AI and implementing them.

As part of this, we don't just have the amazing foundation model CEOs like Sam, and Arthur, and Daniela. We've also brought in two exceptional leaders in the space of taking AI and making it enterprise-ready at scale. One of those is CJ. CJ has been the president and COO of ServiceNow for the past seven years.

And in addition to ServiceNow being a Sequoia-backed company for over a decade, ServiceNow is in an exceptional business. It's the kind of business anyone in this room should aspire to be like. Pat, I believe sourced the business in 2012. Is that right? 2009. And what was the scale of the business then, Pat?

20 what? Million of ARR. Okay. Anyone have a guess as to where ServiceNow is now in terms of ARR? Andy, I feel like you know this. Billion? One more guess. 20 billion? Okay. Somewhere in that range is correct. So ServiceNow is the third largest software-as-a-service company in the world.

It has a market cap of $155 billion. It is about to cross 10 billion of ARR. It's at 9.75 billion of ARR. Remarkably, when you took over as president and COO seven years ago, it was at a billion of ARR. So it's 10x growth. And it was 13 billion of market cap.

So it's been a 12 plus X multiple from there. It's adding 600 million of ARR a quarter. So let that be something to aspire to. And I mentioned it's the third largest SaaS company in the world. Well, number one is growing at 11% a year. Number two is growing at 12% a year.

And ServiceNow is growing at 26% a year. So we all are good at math that ranking doesn't last very long with that kind of growth. And CJ told me virtually all of this has been organic. They've made some acquisitions like Element AI. We'll talk about that in a little bit.

They've been very ahead of the curve on AI. But really it's been organic. It's called ServiceNow, but it does not have services margins. This is an 82% gross margin business. The free cash flow is 30%. Operating income, 27%. This is a rule of 56 business on the rule of 40 terms.

And since IPO, since Pat sourced it in '07, I don't know what kind of multiple it's had. What was the valuation then, Pat? What it would be invested? 260 million posts. That's pretty good from 260 million posts to 155 billion, nicely done. Since IPO, it's been up 42 times, 42X returns.

So an exceptional story, and one that I think is both grounding and also aspirational for everyone in this room. One of the main reasons why we were so excited to have you come, CJ, is ServiceNow has been way ahead of the curve on AI. We first met actually in a conversation where you were talking about the AI vision with NVIDIA, with Jensen and the NVIDIA team.

And we might have a clip actually, but just yesterday at the NVIDIA conference, ServiceNow was a main feature. The specific use cases on the NVIDIA platform. And at the previous conference, it comes up as a main user. My first question for you, CJ, is please tell us how you got here to this exceptional role as COO where you do much of the product work.

And then please tell us how you've gotten the AI suite, what it is today, and how you've gotten the AI suite to where it is. Yeah, so first of all, thank you for inviting me. And he asked me a fun fact. And the fun fact is that I'm a failed standup comedian.

So I'm constantly working on the material to try something out, very high self-deprecating humor that you will see from time to time. And at the highest level, I want to say that we are extremely, extremely grateful to Sequoia. And here is a simple story. The story is that before we were going public in 2012 is when we went public with a modest market cap of $3.9 billion.

Okay, that's when we went public in 2012. And it was a meh IPO. People were like, you know, Facebook was pretty good. Back then it was called Facebook, the same year. And Workday was another one. And we were a meh IPO. But before that, VMware made an offer to ServiceNow for single-digit billions, single-digit billions, below five.

And Doug Leone and the Sequoia team convinced the management team, which would have been happy to take that offer because they're like, wow, we don't have to go public. We can be part of a great company like VMware at the time. And Doug Leone, to his credit, convinced not only the board, but also the management team that you can become big.

So we are extremely grateful for Sequoia's coaching at that point in time. And for the $2 billion offer which we could have taken versus today's $155 billion in just a matter of 12 years, it's incredible. So thank you to Sequoia and the team for sourcing us and believing in us and allowing us to get here because it would have been a very easy thing for Doug to say he was on the board and the return would have been amazing for Sequoia at that point in time and we still walked away.

- CJ texted me before and he said, "Excited to see you. "Can't wait to be part of the cult that is Sequoia." You're already in it now. - I am, I am, I am. But I just wanted to be thankful and grateful and you guys work with great people at Sequoia and I never take that lightly and on the path to our growth, we never forget our friends and supporters.

So just wanted to start there, okay? Number two, on the scale, so I joined the company, so Frank Slootman hired me and Frank Slootman was also placed by Sequoia at ServiceNow. And Frank had two choices at the time, which is I think public knowledge, but Frank's data domain gets acquired and then he decided, and I make fun of Jess all the time on this topic, but he decides that he's gonna be a VC and he joined Greylock.

And if you have met Frank Slootman, which I'm sure some of you have, he is not a Sand Hill guy. So he said, "No, I need to go back to operational CEO job." And it was Sequoia that convinced him. Frank had two offers, Palo Alto Network CEO or ServiceNow CEO and he picked ServiceNow.

Okay, at that point in time. And that was also Sequoia was behind it. So when Frank hired me, we were doing billion plus in ARR and Frank said, "CJ, by 2020, "if we can get to four billion of ARR, "that'll be a massive W and let's go for it." And our founder who created the company close to at age of 50, Fred Luddy, who became bankrupt.

So this is not a classic Stanford, Harvard story, but it's like he went to Indiana University and it's originally from Indiana. And he said, "I'm gonna create a platform "that can solve multiple use cases." And so we always knew that TAM was pretty much unlimited and the platform provided everything you need to create a product or multiple products.

And so when I joined, we were one-ish billion. We had one large product and two or three small products. And since then, it has been brutal execution on which buyers we go out of. Like the simple question on when my product team comes and says, "Hey, we can create this great product." My first question is, "Who is the buyer?

"Do we have access to that buyer? "Is that buyer next door or two doors down "or five doors down? "And will current buyer introduce us to that buyer "or those two buyers don't talk to each other?" Those kind of simple questions on who is the buyer, who are we competing against, okay?

And even though this is very simple, the third one, what is the size of the price? That if we nail this use case, can we create a billion-dollar ARR product? Again, another product. So we have been doing that level of precise execution, and that's what has helped us on organic innovation.

We haven't bought revenue. We are the only SaaS company that has not bought revenue on our path to 10. And we always buy amazing companies, which has great people, and then we make them work on our platform. But that's how we have scaled now from 2016 billion-plus to 2023 when we exited the December quarter.

We already reached 10 billion of ACV. And then, of course, the revenue trails a little bit, and we guided for 10.75 billion growing at 21%. I mean, these are some of the numbers, but it has been, hey, you have an underlying platform that's cloud-based. What products do you create?

What are you solving for? Who is the buyer? That brutal focus on who is the buyer. Do we have access to the buyer? And what is the size of the price? And without that, with my product team and engineering team, and I joined as a head of products and engineering.

Frank hired me and left after that, but that's been the focus. And on AI, it's as simple as we had a fundamental belief from supervised machine learning and as AI evolved all the way to Gen AI today. We have been very focused on AI in service of our use cases, because if we can infuse AI in our use cases, it's a very easy conversation with a JPMorgan Chase or a Citibank or United States Army, that, hey, you are using service now for this use case, AI will help accelerate X or accelerate Y.

And so we have been acquiring or gaining small teams that are AI experts at various stages all the way from 2016. 2017 was our first one. 2016, we started the journey. And then when we bought Element AI, they were trying to be the next Google of Canada. And they had somewhere between 170 to 180 engineers between PhDs, data scientists, and engineers.

And they were in this amazing team, a lot of very well-known people. Yoshua Bengio, who won the Turing Award, was part of that team. There are people who have written some seminal paper on transformer model. That's the kind of team we got. And the call I got from Allen & Company was, hey, this is a great team.

They have no revenue, zero. And they are trying to figure out what use case AI can be applied to. And this was during pandemic. And I went to my boss, our CEO, and I said, hey, man, these people don't have any revenue, but it's a great talent. And we need to spend some money.

And to Bill's credit, he said, absolutely. If you believe this is a great talent, let's take them. And they showed me Chad GPT 1.0, 1.5, 2.0 demos in 2020, late and early 2021. And then when this whole thing just blew up in 2022, we exactly knew where we could apply LLMs to our use cases.

And again, I don't know if it's a term, but SLMs. We are very use case specific LLMs that we apply in service now. And we started our monetization strategy in September. So that's the story. I know it was a simple question, but I had to give a pretty long answer because there's a lot of history to it.

- Really incredible. And I didn't realize, frankly, until this conversation that you guys are the only SaaS business ever to cross the 10 billion ARR point fully organically. That's remarkable. And frankly, I've talked to a bunch of people in service now. You are so much of the product brain.

It's such a pleasure to get to learn from you. And this question is on product. So you guys had a little bit of a headstart a few years, 'cause I know you and Bill had been talking about AI even ahead of the element acquisition. But then with the element acquisition, you got to think about how you're gonna integrate it into your product.

Tell us about how you got up the curve and now how AI is in service now products, maybe a couple of examples. - Yeah, so I'll just take the recent example. We are a big fan of open source community when it comes to AI. Even on the NLU models, we worked with Stanford to figure out which libraries we can use, which was four or five years ago.

But we are a big fan of open source community. And the team in Canada, working with Hugging Face, figured out for which use cases of service now you can apply AI. So then we said, okay, now listen, you talked about our gross margins. Our gross margins are 82%. And all of you run the companies, your profitability starts at your gross margin level.

That's your first step or staircase. 82, then you add R&D cost, sales and marketing cost, G&A cost, and then you get to profitability. So we are world-class in terms of our gross margin at scale. So I don't have the luxury. It's a constraint-driven optimization problem that I don't have the luxury to say, I'm gonna run open AI everywhere in my farm, in our cloud, because we are 100% cloud company with 170 billion, who knows, 2 trillion parameters now, with 4.0.

I don't have that luxury. So the constraint was, can I run smaller models faster with lower latency for service now use cases? And that was a constraint-driven innovation. And we partnered with Hugging Face, our science and research team, and we came up with the first model on text-to-code. And we are not trying to do text-to-code like GitHub Copilot with Java or anything.

Our text-to-code was specifically service now code, how you configure service now. And that was our first breakthrough working with Hugging Face. And then, once we do that, and you talked about Jensen, he's a big fan of Canada. So when we acquired Element AI-- - Me too, Canada. - Yes.

So when we acquired Element AI, it was the first phone call he made, and said, "CJ, love the Canadian talent. "We should do more together." And that was in 2020, because his history with you, Toronto, and ImageNet, and all of those things. So what we did is, that Canada team working with Hugging Face, we figured out smaller models, a one-tenth the size of OpenAI.

And I told Jensen, "Hey, man, dude, "you are constantly pushing the next rev, H-class, "plus plus, I need these to run on A-100s." And that's what we'll work for, because we are a public company. So you have 1% gross margin, de-sell, and the number of questions I get from investors, like you get from VCs all the time, are not fun.

So, yes. So we basically said, "I want smaller model that can run on A-100s, "and I can replicate that in every cloud." He's still always trying to push me. He's a great salesman, even though he acts like he isn't. He's always trying to push for H, to say H is faster, more efficient, which he's right.

But we wanted something to run on A. So the smaller model, smaller models is where we are going with use cases. - Well, I hear you're really gonna need Blackwell. - Yes, I know. - The highest margin model. - $30,000, or something? - Exactly. What's that to you? - Yeah.

- Fabulous. I have one more question for CJ, and I'm giving you that heads up so that you come up with a couple questions top of mind. CJ, we've got a room of builders here, and they're building many consumer companies, but also many enterprise companies. Frankly, you're a dream customer for a lot of the companies in this room.

What can you tell people building products in this room to help guide them towards being a great AI builder that ServiceNow might consider partnering with and being a customer of? - Yeah, so I said there are two places. So just, when you look at me, look at my forehead, and it's a $1 billion spend I have in my cloud and software.

So if you wanna sell something to me, make it quick, and I'll buy, okay? But I spend $1 billion, no jokes, on cloud and software and on the infrastructure a year, and growing at 25% in line with our revenue. So that's how much I spend, so I can be a great customer of yours, or at least a prospect.

In terms of what works is, if you understand ServiceNow, which if you go to our website, you will not understand what ServiceNow does, but if you understand ServiceNow, we do basically a lot of workflows, as in tasks that get orchestrated digitally in a certain sequence between human and machines.

That's what we do, right? Because people at a large bank or customers tell me that they can get a Tesla faster than getting a PC from the bank or a Mac from the bank when they order something, right? I mean, that's the reality of large corporations, large governments, and so on.

So when you request a PC at a bank, say, and banks try to be very efficient, the processes that it goes through, does it require two levels of approval? Some banks have, for Mac, four levels of approval. Then once those approvals are done, it needs to go to shipping department.

Do they have inventory? They need to base image it, they need to put security creds on it, and then it goes to, okay, what is going to her home address, or it's going to, these are the workflows, and these are the things that we automate behind the scene. So all you say is, I want to pick this PC, I want to pick this monitor, I want it to be delivered via FedEx tomorrow morning.

That's the idea, right? But because of these complex workflows, and the banks want to harden the image of the Mac that they give you, if you're doing on a trading floor, that's what it takes. So that's what ServiceNow does, and then we infuse AI in making it simpler and faster.

So for us, if you understand ServiceNow, and you can say, hey, CJ, for your use cases, here is the great technology that we have built, and you can consume this technology, whether it's your LLMs, or whether it's a use case specific AI that you have done, or some kind of analytic, whatever it is, then you have my attention.

That if I can make the use cases for our customers better, I have your attention, and I'll buy your product, to make us go faster, so we can deliver for our customers, right? We have only 8,000 customers, only 8,000, and if you think about 8,000 customers, 10 billion ARR, you can do the math pretty fast, but we have only 8,000 customers, and so I'm obsessed, like OCD level obsessed, that if you come in and say, here is what it can do for your use cases, I will listen to your pitch every day, okay?

So that's one, and I have enough money to spend, if we can serve our customers better. And number two is, we have a great go-to-market team, so besides our engineering, AI, science, research team, we have a great go-to-market team, and if you have something coming back to the buyer, next door down person, or two doors down person, and you want to leverage, CIO is our prime buyer.

If you think about CIOs, 10, 15, 20 years ago, till ServiceNow came, CIOs were serving other C-suite. Hey, for sales, I need to put Salesforce in, for marketing, I may need to put Adobe in, for finance, SAP, I need to put for the CFO. We were the first platform, we said this is the CIO's platform.

So if you say that CIO is your buyer, and you want access to the CIO, there is no better company to partner with than ServiceNow, right? The two companies that really sell to CIO well is ServiceNow and Microsoft. These are the two companies that sell really well. - Brilliant.

I'm sure quite a few people in this room are interested in that $1 billion of cloud spend. All right, we've got time for, I'd say, let's say three to four questions. Michelle, hello. - Thanks, Vijay. Many of us have dreams of an Act Two. Some of us might already be thinking about Act Two in terms of a product, any advice in the early days of how you think about resourcing and philosophy around experimentation versus intentional bets around Act Two's product development?

- Correct. So I will share a story that the reason our IPO was very mid is because people said that our TAM was only 1.8 billion. So one of the industry analysts said, "These guys are not gonna do well, "and their TAM is limited." Same thing happened to many, many companies where people say, "I don't know if the TAM is there." So for us, that actually created a chip on our shoulder because we believed that the TAM for ServiceNow was a lot bigger than what the industry analyst community said it was, which was sub-2 billion, okay?

And this was in 2012, not too long ago. So one thing is on your core, core being core, you really, really have to understand what is the TAM, which is an art combined with science, before you start saying, "I wanna go multi-product. "I wanna go now to different buyers "or the same buyer, but multi-products," whatever the strategy you wanna go after with the buyer access, but the core has to be core.

The reason you exist is for something. You try to solve a problem. So really understand the TAM behind that core and then figure out, before you go into the next act, why are you really going after that next act? So we prevented going after the next act till we hit one billion in ARR.

And then overnight, we flipped it, and we said, "We are gonna go after these three "buying centers, security, HR, and customer service, "in addition to IT, and here is the go-to market for it, "here is the buyer for it, "and we are gonna rely on CIO "to make introduction to those buyers," because we nailed the CIO.

So core has to be core, and you really have to understand the TAM before you say, "I'm now taking," because it's so easy to say, "Oh, typical thing I hear from entrepreneurs, "CEOs, smart people like yourself, "hey, we have a great product, "but I don't have a great sales team." And then they flip constantly chief revenue officer, right?

They do, some of you probably do. And I always, when I'm asked for advice, which rarely happens, but when I'm asked for advice, I say, "What problem are you really trying to solve? "Is it the chief revenue officer, "or you really don't know what product you are building?" So that focus, that maniacal focus on main thing being the main thing, and what is the TAM really in there, before you pivot, or before you go to second act, is something that we look out for.

So that's what we learned. We said main thing should be the main thing till it hits billion, before we go to the three other things. - Any Sequoia company can discuss act two after a billion of ARR, is what I'm hearing. And I think a very powerful insight there, CJ, when we get the next question set up, is just the power of the CIO.

I think a lot of people overlook that, that HR and security would look to them. Charlie. - I have a question, which is just, ServiceNow is a broad platform with many capabilities. You have other ways you interface with customers, like customer support. How do you think about prioritizing where you want to integrate AI?

- So, it has been hard in figuring out where AI could truly disrupt a use case, right? Because that's always the hardest thing, because it's still very bleeding edge. And on the buyer side, if you're talking to a large bank, if you're talking to a large government, right? And I'll share one story on this.

So US public sector, which is federal, state and local, okay? You have to invest a lot in US public sector for certification of your product. The cloud, you know, Microsoft has regions for IL-5, IL-6. So our chief revenue officer came to me and said, "We want to go all in on US public sector." And I said, "Okay." And we had to invest 100 million plus in infra before we can start really making money in US public sector.

And now, whether it's US Army, US Navy, Air Force, to all public sector institution, even on civilian side, are all ServiceNow wall-to-wall customers. And coming back to the question, we always try to figure out the pain point and can AI really disrupt that use case in a positive way that customers gets higher value from ServiceNow.

So if I infuse AI in a use case, because not all use cases are created equal, right? You have a product for multiple use cases. Not all use cases are created equal, but which use case will really provide higher value? Because when customers spend money on you, all they are looking for is how much value I'm gonna get out of this investment.

And that software ROIC, including AI ROIC, is hard. Like right now, we are trying to tell everybody that, okay, we have gen AI infused in ServiceNow products. Number one question is, how much will it cost? And what's the return I'm gonna get? And if you're not doing outcome-based selling, so if you're not doing outcome-based selling, it's like you are another guy coming in there and giving the AI pitch to the customer.

And truth is, nobody gives a shit. I mean, they don't. Because you have to be very, very clear and specific on here is where you will get the ROIC on that. So wherever the highest ROIC is, that's where we prioritize. That customers can say, okay, I could see, if CJ is saying $10 million productivity for this large bank, most likely, because if CJ is gonna be $3 million, but $3 million is still better than zero.

Yeah? - Andy. - You have such a close understanding of what the CIO wants. - Yeah. - Beyond HR support services, security, IT, what's the next set of use cases that the CIO is super excited about beyond what ServiceNow is currently harvesting? - Yeah, I would say, if you think, and one thing you all should know, this is what I learned, is in my seven years at ServiceNow, they constantly told me about CIO being chief irrelevant officer, okay?

And they said, all the power is with developers. And CJ, you are selling to the wrong door. You need to be like other people and sell to developers. Yes, developers will buy X, Y, Z, all that is fine. But the irrelevancy of CIO has been exaggerated. And right now, CIO is the most technical person on a C-suite at most of the large companies which are your buyers, right?

You may have a CTO, the product person, right? The product tech person, and you have the CIO. But product tech person is focused on innovation, not about what they will buy from you. So coming back to CIO's irrelevance, has been exaggerated and over-exaggerated year after year. And now, we mainly sell to Fortune 500.

In Fortune 500, the CEOs are only asking CIOs, give us the AI roadmap, give us this, give us that, and so on. So that's number one, that CIO is still very relevant and very important, okay? It's not that developers are not, but if you're selling to developers, I mean, I was one, very fickle people, right?

They churn and sell thing, and someday they like this, and I read this on frickin' Reddit, and now I like this. (audience laughing) That's a hard thing to sell. So CIOs right now, with the state of economy today, are focused on two main things. One, where can I take out the cost, just enterprise-wide using technology?

So where can I take out the cost? And second, how can I help in the path to revenue? If I can help on the path to revenue, whether it's quote to cash, or whether it's any part of the sales, front end or front office, CIO is constant. Like today, I'll tell you at ServiceNow, it still takes us a time when we propose a bill of material to a customer, and customer says, well, I want to change quantity here, quantity there, then we have to re-spin the order form, make sure it's in SAP, make sure it's in our CPQ system.

That process is still three, four hours sometimes, and at the end of quarter, three, four hours feels like eternity. So how can we do that fast? Well, CIOs right now are the two. The CFO and CRO are their biggest stakeholder, and if your product is in path to that, you'll always get that CIO meeting.

But you have to make it really quick, because I talk to seven, eight CIOs a day, and that's what I constantly see on the pattern matching. - Excellent, we have time for one more question, Peter. - All right, you talked a lot about use cases. I'm curious if you have any stories of use cases with AI that worked really well, and use cases with AI that really did not work at all.

- Yeah, so one of the things I would say, trying to read documents and understand documents, which is accounts payable, invoices, and this and that, it is still very hard for AI to crack that, and we have tried multiple different ways, and people talk about OCR and this and that, and that is a junkyard of technologies that we have tried and not been able to crack through, because if you can automate the paperwork, invoice matching, and other things, there is still a lot of dollars to be had on productivity, on booking the revenue, and so on.

Where we have seen the most is simple things like predicting X for our use cases, specifically for our use cases. So for example, say a large bank. Their policy is that if you use your computer, every three years you can refresh your computer, say your Mac. We can do that now with AI, and we can say, "Hey, Julie, you're four months away "from your entitled computer refresh, "and we have already notified IT, "and you will get your new computer on that third," so the depreciation and all those schedules work, and say yes if you agree, because we still want human in the loop, because nobody likes.

So far, companies don't like if we just make the decision. That's where we see that when we see the pattern matching and can we predict better and make it easy is where we are seeing the highest leverage on even generative AI. - Amazing. CJ, this is fabulous. I wanna tie together some findings here.

Really, this is unique in that we had two separate areas. First is application layer. We've been talking about this throughout the day, and here's someone who has done it incredibly successfully at massive scale. You talked about who's the buyer, what is the size of the prize, and who are you competing against, and also the overlooked customer.

In this case, the chief irrelevant officer. You said that really actually is incredibly powerful. I'll tell a quick story, which is at dinner with you and Bill from ServiceNow a little while ago, you guys said to me, hey, you get to a building, you go to the elevator, what floor do you go to?

I said, I don't know, what floor? The top. Okay, great, you get off at the top floor. Where do you go from there? Bathroom? No, the corner. That's the person who's the buyer, and that seems like it permeates your mentality. And two, actually focusing on the customer. Each of you in this room, if you succeed, will try to sell to the likes of ServiceNow, a billion dollar cloud spend business.

And you gave us a great guide here with AI. For you, you're looking at small language models, not just the biggest ones. You're looking at cost. You're looking at open source. 1% gross margin matters to you. That's how you build an 82% gross margin, 160 billion dollar business, and actually understand the customer.

Don't just go to the marketing website and try to guess what ServiceNow does. Talk to people who've built it. This is fabulous. CJ, any parting words for the team? - Nothing, it was a pleasure, and I'm always around if you want any words of wisdom. I've made a lot of mistakes as well, too many, in terms of scaling ServiceNow.

From a product perspective, engineering perspective, which use cases you prioritize, which you don't, how do you ring fence the team when you go for the second act, and really, really get focused? So, lots of mistakes as well, but we are constantly learning. - Fabulous. CJ, one last question for you.

Will you do stand-up for us later on tonight? - No, I will not do stand-up today. It's my stand-up when I used to do it. Nowadays, I'll get canceled very fast, and Bill will not appreciate that, so I will not do it. - Well, there we go. That one's off record.

Thank you, CJ. - Thank you. (audience applauds) (upbeat music) (upbeat music)