C3.ai, Inc. (AI) Earnings Call Transcript & Summary

Okay. Good morning, everyone. I am Mark Murphy, software analyst with JPMorgan. And it is a great pleasure to be here this morning with Tom Siebel, the CEO of C3.ai. Tom, thank you so much for taking the time and joining us today.

Thank you, Mark.

Maybe you could spend a couple of minutes on -- just with a brief introduction of yourself and the idea behind C3.ai, just for anyone in the audience who might not be familiar yet.

Okay. Well, I -- this is my fourth decade in the information technology industry. I'm a computer scientist from the University of Illinois; did my graduate work in relational database theory; went to work for a little start-up at Menlo Park by the name of Oracle Corporation that, I think, was doing about $2 million in revenue at the time and had 20 employees in the United States. I worked there for some time and was ultimately one of the guys who ran that business. So that was quite a professional experience as we've established and maintained a global leadership position in the relational database industry globally. And then in 1993, I spun out with some colleagues, and we thought about the application of information technology and communication technology, the sales, marketing, customer service, what we now call CRM, a name that we coined. And that turned out to be a pretty good idea. That was a company called Siebel Systems. We started in 1993. I believe, even today, that was the fastest-growing enterprise application software company in history. So by 2000, we were doing $2 billion in revenue, had 8,000 employees in 29 countries. And we spawned what is today, a 20 -- excuse me, $80 billion industry in CRM. So that proved to be a pretty good idea. Then -- and we merged that company with Oracle in 2006 and began thinking about this effort in 2007 and 2008. And we began in 2009 with -- that was the genesis of C3.ai.

And you've been a fountain of great ideas, Tom. And I'll never forget -- in fact, the first time I met you was the IPO road show of Siebel Systems, and there's just nothing like it. So maybe you can bring us now -- bring us forward to C3, your vision for where the world is heading, how this market is evolving and just how -- because C3 itself has been evolving very, very rapidly, where this company has been and where you think it's headed?

Well, we, at large, if we look at the information technology industry, when I joined Oracle out of grad school in 1983, I think the worldwide industry was a couple of hundred billion dollars. And we've seen the advent of these technologies that have dramatically accelerated the industry, minicomputers, relational database technology, personal computers, cell phones, cloud computing, what have you. Windows 95 was important. [ Nematic ] laptop devices were important. The Internet was huge, right? And these accelerated the growth of the industry to today that -- you know better than I, but I think it's about $3.5 trillion or $4 trillion business. If we look at what's going on in this part of the 21st century, first half of the 21st century, we have a new vector of -- a new step function of technologies, okay, in the form of elastic cloud computing, the Internet of Things, big data and predictive analytics that are accelerating the growth of the information technology industry globally in this phenomenon that has come to be known as digital transformation. So the big idea behind C3 was to build a software platform that would allow even the world's largest organizations to leverage this new step function of technologies to be able to build enterprise AI and predictive analytics applications. We began in the energy industry, as you know, with clean energy. So this was about allowing companies to measure, mitigate and monetize their energy and carbon footprints in real time, something that now is evolving into the ESG market. And then from there, we went into utilities and -- which was an early -- huge early adopter of IoT with smart grid analytics. And then in 2016, we branched out into manufacturing, telecommunications, aerospace, precision health and what have you as C3.ai, and it's become -- it's been really quite a remarkable experience to see this market evolve and expand.

So Tom, when you think back on this, we have been aware now for many, many years. Companies have spent vast sums of money trying to develop, whether it's predictive analytics or AI machine learning technologies, they've been trying to build it in-house. And I think more often than not, these projects have failed, and they have fallen short of what they had hoped for. How do you think that this purpose-built platform, the C3 platform, is going to be able to bridge the gap? And why do you think so many of these in-house projects have been failing?

Well, this is not new to enterprise, AI, Mark. I mean as you're well aware, when we introduced relational database technology to the market in the '80s and '90s, the alternative to buying it from Oracle or back then, there was a company called Relational Technologies or IBM was -- what most companies were trying to do was build their relational database systems to run their companies themselves. I mean AT&T tried it, the big banks tried it. I mean how many companies succeed the ability of their own relational database system? That would be like 0. When we introduced ERP technology and CRM technology in the '90s -- particularly in the '90s, the alternative to buying an ERP system from SAP or a CRM system from Siebel was to build it yourself. Now a CRM system would be 2 orders of magnitude simpler than this. But everybody tried to build it themselves. IBM, Microsoft, Merck, all the pharma companies, all the chemical -- how many companies succeeded to building their own CRM system? That would be none. So after their IT organizations kind of tinker with this stuff, they got -- the IT organizations frequently think that they are -- they're software development organizations. They get a little confused. And these projects inevitably come crashing down, and they buy their ERP systems from SAP or Oracle. Today, they buy their CRM systems from Microsoft or SAP or Oracle. Nobody builds it themselves. So now as we engage, as we embrace this enterprise AI opportunity and the ability to apply predictive analytics, the business processes of telecommunications and oil and gas and manufacturing, travel, transportation, what have you, the knee-jerk reaction of the IT organization is to hire 30,000 people in Bangalore and try to build it themselves. And to our knowledge, no one has ever succeeded. There have been 2 very large commercial efforts that you're well aware of, one that was financed by GE over the period of about a decade. And I think they spent $6 billion or $8 billion trying to build this enterprise AI platform called Predix and that came crashing down around them to the point of making a material contribution to the destruction and dismantlement of one of the world's great companies, GE. Other large commercial example that was hyped on every Super Bowl ad in -- for many years was IBM Watson, where these guys -- I mean IBM spent scores at billions of dollars trying to build this thing called Watson that, to our knowledge, there's not one successful installation in any place on the planet. And today, if they still market that product today, it's a very well-kept secret.

Yes. Yes. Well, one success I can remember was it'd be human beings in jeopardy a long time ago. I don't know what happened after that. But one of your customers, Tom, described Predix as a disaster of epic proportions. So what was it that -- what went wrong with Predix and Watson? I mean these are -- these technologies -- you have some very big sophisticated organizations trying to back this and commercialize this. What do you think went wrong?

Well, for those who are interested in looking at this further, there's a white paper on our website called The New Technology Stack, which describes exactly what they did. And so essentially what GE tried to do and IBM tried to do and companies like JPMorgan Chase are trying today and others is use structured programming to cobble together hundreds of enterprise and extrapise data sources, millions of sensors and hundreds of open-source software components like HDFS, okay, and Spark or whatever and try to cobble these things together into a uniform cohesive platform that can be applied to solve a range of enterprise AI applications. Another problem with that, when you work out the math in terms of the number of API connections you need to follow, it's about -- the complexity of this is order of 10^13th. Well, mortals can't deal with 10^13th, particularly when you're dealing with 10,000 programmers on 4 continents and 12 time zones. So it's a non-tractable information technology problem. And I believe that every one of those problems is destined to fail. And every one of our large customers in the world, whether it's the United States Air Force, the Army, Defense Intelligence Agency, Royal Dutch Shell, Enel, they spent many, many years, and in some cases, hundreds of millions of dollars trying to build the systems themselves before they scrapped it and standardized on the C3 AI Suite.

So Tom, that's a good segue into what exactly is the C3 AI Suite? Because I think -- in particular, we have some generalist investors, right, tuning into this session. It's not always clear to them. What exactly is the product? We hear abstraction layer. We hear model-driven architecture. We've heard these comments that entire other software companies are kind of a feature, right, inside the C3 platform. So just how are you describing it? Have you found a simple way to describe it? And how is your team able to just cover so much ground with this technology?

Well, we sat down with a group -- a relatively small group of highly experienced professionals in January of 2009 and basically spent 10 years grinding this out. And if you look up model-driven architecture in Wikipedia, there's a pretty good definition. And what we did that's unique is we applied the concept of a model-driven architecture to be able to build a platform for enterprise AI applications. Now we own all the patents on that technology, okay? And so -- and there's some very rich omnibus patents that are out there in the public, okay? And so what a model-driven architecture is, is the logical extension of what an object management group, the object management group used to talk about at the beginning of this century, in a kind of completely service-oriented architecture where everything is an abstraction wrapped in a restful interface. So what we have done at C3 in this C3 AI Suite is to provide all of the services necessary and sufficient to design, develop and provision and operate these enterprise AI applications. And all of these services were designed to work with one or the other. So you can think about it as an orchestration layer that sits on top of Azure or AWS or the IBM Cloud or GCP or runs on Bare Metal and enables us to -- it reduces the complexity of building these large and complex applications by order of, say, from 10^13th to 10^3rd. Well, we can deal with 10^3rd, and we're able to build -- bring these very large complex applications up at the Air Force, Bank of America, Shell, what have you, in a very fast time. Now let's think about when we talk about what the basis of other companies being features within our product, I think that's an important question to understand. So we have hundreds of services in the C3 AI Suite. For example, data aggregation and visualization. This would be 1/5 of our product and allows us to aggregate very large data sets, enterprise data sets, extrapise data sets, stock prices, whether terrain, social media, information from millions to tens of millions of sensors that are coming from sensor constellations that are providing data, say, at 90-hertz cycles, where we can aggregate those data into unified federated image and then visualize all the relationships between these components. Well, that would be maybe 1/5 of our functionality, and that would be the entire functionality of Palantir. And I'm not saying Palantir isn't a good company. They might be a great company. I'm not saying they don't provide a good service. I think they're -- see look to me like a professional services company that provides a great service to the world. But that being said, it provides maybe 5% of the functionality of what we do. Features within our product, a feature we would have -- people talk about data virtualization. And that would be Databricks. Databricks does data virtualization, and they use an open-source technology they called Spark, developed at UC Berkeley, a project called AMPLab by a professor by the name of Michael Franklin. Yes, the Thomas Siebel Chair of Computer Science. And the -- and so instead of using Databricks, we use Spark is much the equivalent. If a customer wants to be this Databricks instead of Spark, fine, they just put it in that container. Let's look at one of our features. As you know, democratization of AI is one of the big vectors in the 21st century, enabling analysts to do kind of very -- like no-code AI, drag, drop, point, clip, WYSIWYG and build AI solutions. But one of our features of our product is a feature called Ex Machina, okay, that enables analysts to do very, very serious AI and share the work that they develop with low code developers and deep code developers, Well, Ex Machina is a feature within our product that's functionally dramatically superior to Alteryx. I mean Alteryx is a fine product. And by the way, for anybody who's interested in Ex Machina, you can download it on the web for free today, okay? Just go to C3.ai, download it, use it and send me an e-mail and tell us how to improve it, okay? And -- but it's available for free. You can download it. It enables you to do a very, very rich data science and is functionally dramatically superior to Alteryx. One other area, and I'll just leave it at this, that people often talking about is AutoML. AutoML is kind of all the rage with products like DataRobot and H2O. And so Autorobot is just 1 of 1,000 features in C3.ai. And if you want to use DataRobot or H2O instead of our AutoML capability, you just wrap it in the rest of the interface to clear it as a first-class object and you're using it. But the point being, we provide the -- we provide all of the services necessary and sufficient to design, develop, provision and operate these applications. And you can conceive of the C3 AI Suite is kind of the union of all of the other AI software solutions that are available in the market.

So it sounds so simple when you describe it like that. So one of your partners, Tom, had said -- he said, the product is end-to-end. And as a data scientist, I don't have to worry about basically what's happening behind the screen, right? So that he's saying it's shielding the data scientist from all that back-end IT complexity. Is that a fair description?

Yes. So 95% of what data scientists have to do in most installations is wrangle with data, okay? Worry about how the data relate to one another, what processes act on these data, ETL, encryption in motion, access control, queuing, event processing, whatever it might be, data science services. And so they have to worry about the relationships between all of these components. And we provide an abstraction layer where you're dealing with whatever the issue is of the company, customer, airframe, okay, propulsion, flight management system. And when the data scientists or the analysts for the application developer calls that object, they get all of the data associated with that object without having to worry about how the data are persisted, where the data are persisted, how the data are connected or what processes act on those data. So it reverses the complexity of what -- the facility data scientists can perform data science rather than dealing with kind of low-level data connectivity problems.

Okay. So now, Tom, you've been nonacquisitive at C3. I don't believe you've ever acquired another company. It's all organic and all homegrown. Why is that? What kind of an advantage do you get as a result?

Well, we have acquired a couple of companies in the history of the -- since we've been in business, but they've been relatively...

But they're quite small, right?

They're relatively insignificant. And our focus right now is on organic growth. And I'm not saying, Mark, that we're never going to acquire a company. But our focus is on growing the business. I mean if we stick to our knitting, okay, and don't get distracted by major acquisitions, I think this is going to be a hugely successful company. The goal that we're -- you know our goal, just like it was at Oracle, just like it was at Siebel. Our goal here is to establish and maintain a market leadership position globally in enterprise AI. I think there's some probability where we might succeed at that. And I think a major acquisition would be a distraction. If we stick to our knitting, okay, and do our jobs, I think that -- I think it's highly likely we're going to build a -- I think we have built a successful company. We can build one of the world's great companies. And I'm not saying we might pick up -- not pick up a piece of technology here and there, but I -- but make no mistake, we are focused on organic growth.

Yes. Nothing had fit our criteria of a material acquisition thus far. And we respect that. So I wanted to ask you, Tom, we had gone out, we blanketed your deployment partners. Big thumbs up from these pretty sophisticated partners. They've now admittedly -- here's what they said -- here's what one of them said. C3 is too expensive and heavy for SMBs. And he didn't think you could break even on just one use case with it. But they're saying the subsequent development is significantly accelerated. So everything after your first machine learning model, I suppose, it just becomes hyper accelerated. So their view was if a company is ambitious and data-driven, then that's going to be the one that needs C3. Is it -- are those fair statements?

Well, I think that if we like what's going on with enterprise AI, Mark, when we look back at this in 10 years, it'll be just like many computers -- personal computers, enterprise software, CRM, okay? When we introduce those technologies to market, and I was there for every one of them, okay, the initial reaction was that large sectors of the market would not use many computers, would not use relational database technology. Now let's fast forward. Today, it's inconceivable that you can run your business without an ERP system. It's inconceivable that you can run your business without a CRM system. It's inconceivable that you can run it without a relational database. I think that it will -- and I think everybody on this call can envision that in 10 years, it will be inconceivable that you can run a company without taking advantage of elastic cloud computing. I guarantee you when we look back at this market in 10 years, okay, if you have not adopted enterprise AI to improve your business processes in manufacturing, okay, in product design, in customer service, in supply-and-demand shade management, you will not be able to compete. And so this will -- it will be ubiquitous. Everybody will use it. Large companies, small and medium businesses and small companies will all be using enterprise AI.

Are you seeing that trend develop today, Tom? So when you're out there and you're having these discussions with the Fortune 500 and the Global 2000 that -- you must know most of them from the Siebel days. Which way are they leaning? Are they leaning towards, we just want to start with 1 or 2 AI use cases? Do you think they're leaning much more ambitious, data-driven and kind of with this long-term pretty expansive road map in AI?

I think they almost invariably begin with 1 or 2 use cases. Look at Royal Dutch Shell. Royal Dutch Shell began with AI-based predictive maintenance for offshore oil rigs. And then they wanted this -- they wanted AI production optimization for LNG operations. Fast forward 5 years, Royal Dutch Shell has deployed a project called C3.ai, where we've partnered with Microsoft, Satya and Azure and Judson Alta. And we've built this with Azure and C3 AI to build what they call Shell AI. And they're reinventing all of Shell, upstream, downstream, midstream, renewables to make Shell into a carbon-neutral company by 2050. Bank of America. It began with cash management, okay? And then it went to a unified view of the customer, and then it goes to margin lending. Almost always begins with a relatively small use case and then expands into scores and scores of applications that expand the enterprise. And now Coke Industries, LyondellBasell, Baker Hughes, that's the way this goes.

So it's -- since you just mentioned Microsoft, and I wanted to ask you about this. You had said something along the lines of -- and I'm paraphrasing this, but it was something like I believe AI will replace everything that happened in enterprise application software. And so we see now -- so you're working with Microsoft, you're working with Adobe, you're working with Infor. And I think -- so there's this concept of weaponizing, right, some of these applications that have been around a long time, right? That's a different route to market for you. Can you talk to us about what's happening there?

Well, I do believe that enterprise AI represents an entire replacement market for everything we've done in enterprise application software in the last 3 decades. That will happen. I'm confident of it. But as it relates to partnerships, it is our model to go to market with long-term strategic business partners like Judson and Satya at Microsoft, by far our most strategic partner; Andy Jassy at AWS; Shantanu at Adobe; FIS; Lorenzo Simonelli at Baker Hughes; Infor. So we are definitely partnering with large, large market partners to address the markets. I mean Microsoft gives us, what, $150 billion worth of distribution capacity for all industries, particularly as we attack AI-enabled CRM together, okay? Watch this picture. That's going to be a big deal. The -- we think about our partnership with Baker Hughes in oil and gas, we have 12,000 people selling C3.ai for us all around the world every day, 12,000 people. So our strategy is very much about putting together a large, highly cooperative partner ecosystem to bring these products to market and assure the success of our customers.

Okay. So why are the -- Tom, why are the ASPs so high? I mean I think about this -- some of the total contract values you've been signing, they're just unimaginably large. Some of these customers must be spending more on C3 than they're spending on their CRM system or their payroll system and other systems. What's causing that to occur?

The economic benefit associated with the application. And yes, they are spending more than they are on CRM, in many cases, more than they are on ERP. But if we look at the economic benefit that accrues from these applications, let's look at Enel. Enel, as you know, is the largest utility in Europe. It's the largest utility in the free world. And if we look at the annual report from their CEO, Francesco Starace is looking for EUR 5.1 billion a year from this enterprise AI initiative that he's working with C3. I mean EUR 5.1 billion a year, Mark, how many times have you seen value propositions like this? We have one application at Bank of America where the economic benefit is $10 billion a year. We look at Royal Dutch Shell -- before we did the last expansion of our agreement with Shell, they were looking for EUR 3.8 billion a year in economic benefit. So if you can deliver this kind of economic benefit, then -- they may often pay us. While it might be a lot of money is inconsequential in the big picture.

So the other side of that equation, Tom, right, you have these huge contract values. The customer account is relatively low today, and you have been working on -- you've been making progress on that. What can be done to maybe help supplant the business with a higher volume of customers and orders? And I understood you sell in some boulders. Maybe you can sell some pebbles, right, that would just help diversify. Can you get -- can this get to a point where it's a diversified kind of high-volume model where then some of this bookings volatility, it's always been there for you. Maybe that bookings volatility is a little more stable?

It's -- first of all, it's a great question. The answer is yes. First of all, as a private company, Mark, we did what we know best, which is large enterprise selling. And if you can go close deals for EUR 10 million, EUR 20 million, EUR 30 million, EUR 40 million, EUR 50 million, why worry about $50,000? I mean this way, it finances the business. You don't need to talk to anybody on Sand Hill Road. And it enabled us to build a very, very rich technology foundation, growing the business and finance it without going to traditional forms of venture capital that will have it. But now as we operate a public company, you know that we're dramatically modifying our distribution model. Today, we sell on platforms like the Microsoft -- the Azure Marketplace, okay? You can go to the Ex Machina website right now and download a free copy of Ex Machina. And if you like it, keep it and pay $400 a month, okay? And so I encourage anybody to do that. Please do it today. By the way, if you have suggestions, send me an e-mail and I'll make the product better. What we're doing with CRM is all kind of mass market sales into CRM, a market that we know pretty well. We did invent it, okay? And the -- in a previous life. And you'll see that now we've moved from only kind of major market enterprise salespeople who are elephant hunting to putting traditional enterprise sales organizations. We put deer hunters in place, mass market organization. We have rabbit hunters in place. We have telesales organizations in place. And so we're going all the way down the food chain, and it is our intention to -- and as you've seen also, our ASPs have come down dramatically. I think the year before we went public, our average contract value was, I think about -- and I could be off, okay. But I think it was about $15 million. You'll know better than I. I think last quarter, if I'm not mistaken, it was an order of $5 million. I mean that's a lot of progress.

Okay. Okay. Great to hear. So Tom, we have probably just a few minutes left. And I -- a question or 2 came in here from the audience. But let me just ask you about the demand environment. I think depending on the quarter, there's a pretty good chunk of your business historically coming from utilities and oil and gas. And then we look back on it, I know you remember, oil prices were negative $40 a barrel a year ago. And then they're somewhere around $60 on the positive side today. So is oil and gas a healthier industry with more money to spend today?

Unquestionably, yes. I mean when -- if we look at the first 2 calendar quarters of last year when we had the oil crisis. I mean it just -- when we had the COVID crisis and the resultant oil crisis, I mean our business got whacked in oil and gas. And it was a large sector of our business. And it's true. Historically, and I think in the short term, you're going to see some lumpiness quarter-to-quarter in our bookings that we're going to be smoothing out over the next 2, 3 years through the strategy that we have in place. But today, with oil at whatever it is, $67 or wherever it is today, is clearly a much healthier business. Our oil and gas pipeline looks is very, very robust. That being said, these oil and gas deals do tend to be very large transactions. These are -- when we get into the Aramcos, the Shells, the -- of the world, these are large corporations. So that can create a little lumpiness as these deals -- if a deal moves out of 1 quarter into another.

Understood. Yes -- Aramco, yes, that's a large organization. So Tom, looking outside of that, how do you envision this global demand environment? You've seen so many cycles. You've run companies through these cycles. How do you think about it in 2021 and into 2022 across the other verticals? I mean is that -- is it going to be supportive of strong bookings? We know it's lumpy. But -- or do you think it's -- is it a little touch and go, if we're being realistic? Is it a little touch and go on just some of the geos and inflation and tapering talk and all this?

Mark, I can't tell you what's going to happen this month. But if you look at this in a period, in a 5-year horizon, I mean, holy moly, I mean, as far as I'm concerned, the capital markets can remain close for the next 5 years. We have $1 billion cash in the bank, and we have everything we need to do our jobs. And -- but you run this out 5 years, I mean we're looking at a market that is in, oh, my goodness, hundreds of billions of dollars, and we're going to play an outsized role in it. So I am exceptionally optimistic about the growth opportunities in large enterprises, in medium enterprises, in small enterprises, in CRM, in Ex Machina, telco, retail, travel, transportation, government, yes, oil and gas, utilities, automotive. This, Mark, is going to be a large global successful enterprise, and certainly, a leader, if not the leader in enterprise AI, either -- any way it ends, we have a win.

Tom, it's a perfect note to end on. I want to thank you just very kindly for carving out the time out of your busy schedule. And it's just a real honor and a pleasure to be able to host you at this event.

Pleasure is all mine. Thank you, Mark.

All right. Take care, Tom.