Announcer:
From the library of the New York Stock Exchange at the corner of Wall and Broad Streets in New York City, you're Inside the ICE House, our podcast from Intercontinental Exchange on markets, leadership, and vision and global business, the dream drivers that have made the NYSC an indispensable institution of global growth for over 225 years.
Each week, we feature stories of those who hatch plans, create jobs, and harness the engine of capitalism right here, right now, at the NYSE and at ICE's exchanges and clearing houses around the world. And now welcome, Inside the ICE House. Here's your host, Josh King of Intercontinental Exchange.
Josh King:
I'm just back from a quick trip to California, which included a day in Simi Valley at the Ronald Reagan Presidential Library and Foundation, where I was attending a conference. Walking through a Boeing 707, tail number 27,000, which served as the 40th President's Air Force One, I was reminded that the plane took Reagan to Reykjavik, Iceland on October 11th, 1986 to meet with Mikhail Gorbachev, General Secretary of the Communist Party of the Soviet Union.
On the table between the superpower leaders at that summit was a bold proposal. But at the last minute, Gorbachev threw a wrinkle in the negotiations. I'll let President Reagan pick up the story from there.
President Reagan:
Implications of these talks are enormous, and only just beginning to be understood. We proposed the most sweeping and generous arms control proposal in history. We offered the complete elimination of all ballistic missiles, Soviet and American, from the face of the Earth by 1996. While we parted company with this American offer still on the table, we are closer than ever before to agreements that could lead to a safer world without nuclear weapons.
Josh King:
The sticking point was the Strategic Defense Initiative, SDI, or as it was known back in the day, Star Wars. Gorbachev wanted SDI research confined to American laboratories. Reagan wouldn't relent, but offered to share the technology in the form of low Earth orbit detente. Gorbachev balked at that, and the two leaders went home empty-handed. Until recently. Space has always been the next frontier of the Cold War, but the Soviets beat us into Earth's orbit with the launch of Sputnik 1 and 2.
Then, of course, we're 55 years out from that date on July 20th, 1969 when two members of the Apollo 11 crew, Neil Armstrong and Buzz Aldrin, touched down on Tranquility Base, winning the race to the moon. As we reflect on the more than half century since Armstrong's small step, the superpowers are now engaged in a new competition to make the giant leap to fully harness the power of artificial intelligence. While space remains a potential Cold War battlefield, this modern contest transcends technological supremacy.
It's about securing economic, military, and geopolitical advantages. Today, countries are relying on the private sector to help propel AI technology forward, investing in AI solutions that boost national defense operations and efficiency. Among those companies is C3.ai, that's NYSE ticker symbol AI, led by today's guest, founder, chairman, and CEO, Tom Siebel. With over a decade of preparation for the ongoing AI boom, C3.ai has positioned itself as a leader in the industry.
In a minute, Tom, one of the real legends and driving forces behind the march toward artificial intelligence, is going to join us to discuss AI's impact on national defense and how the United States can position itself to win the AI race. We'll also delve into Tom's career, from his early days at the University of Illinois, to his current role as the leader of C3.ai. And we're going to explore the future of AI and the game changing developments that lie ahead for this transformative technology. Our conversation with Tom Siebel, founder, chairman, and CEO of C3.ai, is coming up right after this.
Speaker 1:
To prove that there is such a thing as free money, we gave people a little sample of Ibotta.
Speaker 2:
Hi, could I interest you in a sample from Ibotta?
Speaker 3:
What's going on?
Speaker 4:
It's just money?
Speaker 2:
Ibotta is an app that gives you cash back every time you shop.
Speaker 5:
What is it?
Speaker 2:
It's money.
Speaker 6:
I can just take this?
Speaker 7:
This is real?
Speaker 8:
This is insane.
Speaker 9:
What's the catch?
Speaker 2:
There's no catch. It's just cash.
Speaker 1:
Get cash back every time you shop with Ibotta.
Josh King:
Welcome back. Please remember to subscribe wherever you listen and rate us and review us at Apple Podcasts so other folks know where to find us. Our guest today, Tom Siebel, is the founder, chairman and CEO of C3.ai, that's NYSE ticker symbol AI, a leading enterprise AI software provider for accelerating digital transformation.
Prior to founding C3.ai, Tom founded and served as CEO of the global CRM software company Siebel Systems before merging it with Oracle, that's NYSE ticker symbol ORCL, back in 2006, currently serves as a member of the College of Engineering Boards at the University of Illinois and the University of California at Berkeley. Tom, thanks so much for joining us Inside the ICE House.
Tom Siebel:
Great to be here. Thank you.
Josh King:
We are back now for a real visit after your virtual bell ringing a couple years ago when you went public, how does it feel to be in this building?
Tom Siebel:
It feels great. I just came off the floor to see the energy of the floor and everybody back in business, to see New York back in business, to see people back in the office, the people in the banks back in business, the people in the shopping centers back in business. New York is very much alive and it's great to be here.
Josh King:
It is. It is indeed. We're feeling it more and more as the months pass and even up in Midtown the streets are bustling and it does feel great. Tom, as I said in the intro, AI becoming increasingly woven international defense. The importance of being first to fully harness the technology growing so significantly. You recently spoke with our former Secretary of State Condoleezza Rice about the AI race. I just want to listen to a little bit of what she had to tell you.
Condoleezza Rice:
There are a couple reasons for that, but the most important is that for all of its messiness and all of its difficulty, democracy has checks and balances. It has divided government in some cases. It has a free press. It has citizens that have views. And so even though we could go in the wrong direction with any of these technologies, the chances that an omniscient and omnipotent single leader is more dangerous I think is absolutely the case.
So I ask you to do just a little thought experiment. Suppose the Soviet Union or Nazi Germany had gotten to the nuclear weapon first, we might've been living in a very different world.
Josh King:
So that was Secretary Rice talking to you, Tom, imagine if the Soviet Union or Nazi Germany had gotten to the nuclear weapon first. Considering the historical context of major technological developments, how can the US ensure that it leads the way on AI?
Condoleezza Rice:
Well, we are at war in the AI front and we're primarily at war with China. And if we look at the next generation of the kill chain, whether we're dealing with space, whether we're dealing with cyber, whether we're dealing with subsurface autonomous vehicles, hypersonic swarms, contested logistics, all of these technologies will be driven by AI. And so we have a kind of open and not kinetic warfare with China, and this will be the ultimate test I think of the free enterprise system versus the totalitarian state.
What we have in the case of China is top-down command and control totalitarian state with their five-year plans and they're mandating leadership in AI. And they have a billion people operating in lockstep, and these people are extraordinarily well-educated. Many of them were educated in our finest universities. They have tireless work ethics and they have a very well-coordinated plan to establish maintain leadership in the deployment of AI as it relates to defense intelligence systems.
Now, in the Western world, we have a much messier system with the free enterprise system, and we're largely dependent on the private sector and where inventions take place in the storefronts in Cambridge, in England, and in the Bronx, and literally garages in Palo Alto, California. And the question is, to what extent is the United States government going to be able to marshal the resources of the private sector and the innovation of the private sector to maintain and sustain a leadership position in artificial intelligence as it relates to defense?
That's the battle that we're in where we have the privilege of being able to serve the Department of Defense and the intelligence community in many ways. And so we're doing the best we can to transfer our technology base to them and enable them to take advantage of it.
Josh King:
I want to go a little bit deeper on that, Tom. I looked on your website in the area of the US Department of Defense using C3.ai at scale, things like accelerated mission data collection and processing, generated hypersonic missile trajectories, even increased supply visibility and expedited security clearance investigations, which from personal experience is something that DOD can really use. What excites you most about some of these approaches?
Tom Siebel:
Well, they're all exciting. They're all necessary. None are sufficient. I think one of our most interesting applications, I think everybody listening will be able to understand, is AI predictive maintenance for the United States Air Force. Now, the United States Air Force operates about 5,000 aircraft. Most of these airframes, average age of these airframes is 27 to 30 years old. On any given day, about half of them will deploy and half of them are down for unscheduled maintenance.
Well, by aggregating all the data from these weapon systems, the telemetry from the airframes, the maintenance systems, the parts systems, the inventory systems, the weather wherever they were where they operated, and these are massive data sets. I mean, just a B-1 bomber airframe has 42,000 sensors emitting telemetry at eight hertz cycles. That's eight times a second.
So this is one of the largest AI applications deployed on Earth where we take all these signals, we run them through machine learning models, and we can identify system failure or subsystem failure before it happens, 50 to 100 flight hours before it happens, auxiliary power unit, flap actuator, igniter in the afterburner, whatever it may be, and then we can fix the plane.
We can dispatch the parts and the personnel to converge with the plane on its current flight profile, fix it that night in Munich or Stuttgart, and the plane doesn't fail. Net-net we're increasing aircraft availability of the aircraft being used by the system. It's called PANDA, that's their name for it, by increasing aircraft availability by 25% on any given day. Get your mind around that. 25% of the United States Air Force is probably larger than any other Air Force in the world, just that increment. So this is highly impactful technology.
Josh King:
The B-1 has a long history. It's been around a long time, so all that data and the decades of information that you have can be highly predictive. I can see that. Another platform across branches of I think the Marines, the Navy, and the Air Force is the F-35 Lightning II, much newer aircraft with a much less mature data set behind it. How do you apply what you're doing for platforms like the B-1 to something that tells you a lot less based on its service history?
Tom Siebel:
Well, it's fundamentally the same technology and it is the same technology. It's the same machine learning models. We're just training the machine learning models on a different data set. Now, the F-35 program is roughly a $1 trillion program. These things cost about $100 billion a copy. And you can think of it as basically 200 computers, a pilot and an engine and a couple of wings. That's basically what it is.
So the amount of telemetry that these things emit are massive, and we were very much a part of the deploying. They've had massive reliability problems with the F-35 program, and I think that's a responsibility of the manufacturer. But we've been applying predictive analytics to that airframe to fix that maintenance problem and make these aircraft more reliable and more effective at what they do. That is core to our mission.
Josh King:
We're going to come back to some of the defense applications a little bit later on our conversation, but let's stay right here at 11 Wall Street for a second. It doesn't feel like it's been nearly four years since amid the height of COVID, you rang the closing bell virtually to mark C3.ai's IPO back in December 2020. Reflecting to that time as millions stayed hunkered in their home, you said, I'm going to quote you here, "I never in my wildest imagination thought we would go public this year."
Yet C3.ai, under that ticker symbol AI, joined the ranks of our 2,400 listed companies. What has the journey been like then in the last four years from that time, December 9th, 2020, through the pandemic and the ongoing AI wave to what brings you here on this journey to New York?
Tom Siebel:
Well, this has been a long and very carefully thought out journey to establish a leadership position in enterprise AI. Understand, we started in January 2009 and I was talking about enterprise AI in 2009, '10, '11, '12, '13, '14, '15, '16, '17, '18, '19, and no one in the world was talking about enterprise AI. We began to work and spent billions of dollars building this C3.ai platform when AWS was just a tiny little enterprise.
This is before Azure. This is before the Google Cloud, and this is before the GPU. We spent over a decade building that platform, and now we've built 90 turnkey enterprise applications to address the value chains of financial services, health, government services, defense, intelligence, manufacturing, healthcare, what have you. And so it's been quite a journey. We spent 15 years preparing for this moment.
Now, after the announcement of ChatGPT in November of 2022, all of a sudden the AI winter is over. The AI winter, which goes back to like 1950 when all these technologies were invented and really nothing happened for almost 75 years until we finally saw the elastic cloud become preeminent, big data, internet of things. And these enabled us to deliver today 90 turnkey enterprise AI applications.
And so it's not AI winter anymore. Look out the window, it's sunny. It's warm. It's going to be warmer tomorrow, and we're in the midst of the most rapidly growing and indeed the largest market in the history of enterprise application software. And C3.ai is fulfilling its mission to establish a leadership position in that space.
Now, going public as we did with your considerable assistance to the New York Stock Exchange in the beginning in December of 2020 allowed us to raise the capital to invest in technology leadership, invest in thought leadership, invest in market leadership, and assist us in establishing a leadership position globally. So that was a very important event in the development of C3.ai that has accelerated the company considerably.
Josh King:
To go well back before, even before the AI winter, if you want to describe that period that way, in your recent fourth quarter earnings call in May, you compared the current AI market to the personal computer market in the '80s and '90s, noting the market frenzy around AI infrastructure similar to the excitement around the physical computers decades ago. And fast-forward to that evolution, Tom, in one of your recent keynotes I think last year, you played a clip of Steve Jobs at an Apple keynote in 2007 talking about his new gizmo.
Steve Jobs:
IPod, a phone and an internet communicator. An iPod. A phone. Are you getting it? These are not three separate devices. This is one device.
Josh King:
Why do you think we're seeing history repeat itself with the unfolding AI era and the focus on infrastructure mimicking those early days of personal computers?
Tom Siebel:
Well, when Steve introduced the iPhone in 2007, the value was perceived to be in the hardware basically and in the silicon underneath it, and the market reflected that. If we think back to the personal computer market in say the '80s or 1990, an IBM PC XT cost $3,700. That's $22,000 in today's dollars. On top of that, we would have some software infrastructure from Microsoft. And on top of that, we'd be running two or three or $400 worth of software from VisiCalc or WordStar or one of these companies.
Now, fast-forward to 2024, this PC on your desk more powerful than any machine that existed in 1970 or 1980 cost a couple thousand dollars, and that's a couple thousand dollars for it's a lifetime. By the way, date back to when the IBM PC XT was there, I mean, all of the value was perceived of in the silicon and the infrastructure, and this is when Intel was king. That's where the value was perceived by the markets. Now, if we look at the AI market today, the value is perceived of in the silicon and the infrastructure.
And I'm not saying that I don't think Nvidia is a great company because I know Nvidia is a great company, and I'm not saying Microsoft isn't a great company because they are great companies. But when this plays out, just like the PC today that on a business person's desk cost the company a few hundred dollars a year to own and there's maybe $8,000 a year in software running on this application from Bloomberg, from SAP, from Salesforce, from others. We have seen this movie before.
If we look at the value of the iPhone today, it's not in the hardware, it's in the applications. I mean, look at the size of their software business and service. It's huge. So the AI market will play out the same way. So we're looking at a multi-trillion dollar market opportunity. In the long run, 80% of that value stack will be in applications. Silicon will be commoditized. Infrastructure will be commoditized. They're always commoditized. And the value stack will be in enterprise AI applications, and that's what C3.ai does.
Josh King:
You say, "We've seen this movie before." Having an appreciation of the history of your specific industry history and business in general as well as a vision toward its future is so critical for the success of A CEO, I think. I saw that A former guest of this show and my old friend Alan Murray recently joined the C3.ai board of directors after his long run at Fortune. What kind of perspective does Alan, as one of the world's most eminent journalists, bring to your boardroom?
Tom Siebel:
Alan brings historical perspective. I mean, he's a apogee of the result of liberal arts education. He understands history. He understands literature. He understands philosophy. He understands journalism. He is connected to every CEO in the world. I've had the great privilege of knowing Alan for decades, and we're very, very pleased to have him join our leadership with Condoleezza Rice and others on the board of C3.ai.
Josh King:
He always, as you said, has his finger on the pulse of the next big thing. We saw that in play at all the conferences that he'd hold and host. How are the current trends and the anticipation around AI infrastructure align with C3.ai's focus? Are you interested in infrastructure related products or are we concentrating on areas that might not be at the top of the value stack now, but you see the future of infrastructure?
Tom Siebel:
No, all this infrastructure that's being laid out by Nvidia and Microsoft, AWS and others, these people are out laying pavement for us a thousand miles down the road. What this infrastructure is going to do hard stop is run enterprise AI applications and we deliver those applications. People don't buy these GPUs to watch them glow. Let me help you out. They buy them to run enterprise AI software. So this is the game. And C3.ai after 15 years of work is really in a situation of finding itself in an enormous market tailwind where there is insatiable demand for these applications.
And I believe there's certainly no other company in the world that provides 90 turnkey enterprise applications. I'm not certain there's another that provides three. This is professional experience of a lifetime to be able to serve this market and deliver high quality products and meet this market need so that companies can deliver higher quality products at lower cost, lower environmental impact, into the hands of more satisfied customers so that governments can provide higher quality customer services and social benefits at lower cost and higher levels of constituent satisfaction.
And so that the United States Department of Defense and its allies can protect and defend freedom and liberty at global scale.
Josh King:
We're going to get to this professional opportunity of a lifetime that you see in 2024 at C3.ai more in just a minute, but let's rewind the clock a little bit and have our listeners understand a little bit more about Tom Siebel. C3.ai went public at the end of 2020, but you founded the company in 2009, about 16 years after you established Siebel Systems.
Siebel Systems was the leading CRM application of the day and in 1999 the fastest growing company in the US. Take us back a couple decades, Tom. What first motivated you to start Siebel Systems and then eventually merge it with Larry Ellison at Oracle?
Tom Siebel:
I did my graduate work in relational database theory at the University of Illinois. This is before the existence of a commercial relational database system. And I had the great good fortune of being invited to join Oracle Corporation in its formative stage. And so this was with Larry Ellison and Bob Miner, Ed Oates and others, which was the founding team, to play a material role in the commercialization of relational database systems. At the time, they had about 20 employees, and I worked there for about a decade.
It was the professional experience of a lifetime. And we were enormously successful at commercializing relational database technology of bringing it to market and of establishing a clear market leadership position globally in relational database technology, which proved to be an important technology as I thought it would. When I was at Oracle, I thought a lot about the application of information technology to sales, marketing, customer service, which say as of 1990 was largely unserved by information technology.
And I thought if a company were able to bring those technologies to bear, to enable organizations to operate those functions more efficiently, that there would be a market for it. So I left Oracle and in 1993 formed Siebel Systems with the idea of establishing and maintaining market leadership in a market that I'm largely credited with creating, I think I did create, called customer relationship management or CRM. Fast-forward by 2000, I think we're doing about $2 billion in revenue.
We had 8,000 employees in 29 countries and 4,500 large corporate customers around the world. That was Siebel Systems, the fastest growing enterprise application software company in history. We had established by the middle part of the first decade of the 2000s say 85% market share in sales, marketing, customer service globally. Siebel was acquired by my former colleague, Larry Ellison at Oracle in, as I recall, January of 2006. And I chose not to go back to work for Oracle having spent much of my career there and started thinking about what was next.
And so I got together with a group of former colleagues and friends from McKinsey and Accenture and Intel and Oracle and SAP and Siebel Systems and other companies, and we ideated the idea of C3.ai. And we thought next was going to be about elastic cloud computing. Again, remember, this is before the existence of AWS. We thought next was about elastic cloud computing, big data, and internet of things. And so we ideated this in 2006, 2007 and 2008, and we decided to start a company to address this opportunity.
In December of 2009, I sent an email out one Sunday, one Friday, raised $20 million by Sunday and the company was funded. So we started work in January of 2009, and we've been at work every day since then realizing this mission. I think we are on track. We are right on plan, and the market has played out about as we expected. So let's just say we got lucky again. We're in the right place at the right time. And if we don't completely goof it up, I think it's largely there's some likelihood this might be a huge market success.
Josh King:
Right place at the right time. Tom, in a 2001 interview with the Harvard Business Review, you said, I'm going to quote you, "It's unfortunate that CEOs are taught to believe that their most important job is to drive up the company's stock price or meet Wall Street's expectations. Those are secondary effects of a more primary goal, understanding what the customer needs and delivering it. You need to manage for customers and employees, not investors."
So I looked at your website again this morning, the logos that you name in your case studies, many of them listed here at the NYSE, names like Shell, FIS, Cargill, ConEd, Raytheon, and of course, many areas within the US government. 23 years later, do those words that you said to the HBR still reflect your mindset?
Tom Siebel:
Absolutely. At C3.ai, our time horizon is five years, 10 years, 20 years. We're focused on building one of the world's great companies. Our job is to deliver great products. Our job is to make sure that our customers are satisfied. Our job is to make sure that our employees are satisfied, that they're motivated, that they feel fulfilled with what we do, that we deliver great products in the hands of satisfied customers. If we do that, the market will take care of itself. So what the stock does today, I don't look. What the stock does this quarter, I don't really care.
I mean, candidly, this is a very, very well-capitalized company. I think that we have over three quarters of a billion dollars in the bank. We were cash positive last quarter. So we're looking at this for the long run. And I think that candidly from our perspective, you can turn the lights out on the market for the next five years and turn them back on. And I think that if we take care of our customers, we take care of our employees, we continue to demonstrate technology to leadership, the shareholders will be most pleased.
Josh King:
Since that time in 2009, the company has given itself a couple of different names all in the C3 space. It started at C3, then C3 Energy, C3 IoT, internet of things, and now C3.ai. During the transition to C3 Energy in 2012, you established the data science division to incorporate AI techniques into machine learning, predictive analysis, supervised learning, and also unsupervised learning into your applications. What was the vision for the future of AI at the time, we're back around 2012, and how did this early integration helped position C3.ai as the leader that it is today?
Tom Siebel:
When you have the advent of new markets, think when semiconductors were introduced or when the PC was introduced or relational database technology is introduced or the internet, it's very difficult to determine in advance how these things are going to play out. And you need to be on the balls of your feet and be ready for it. Now, it's counterintuitive, but the first application of enterprise AI really was the utility market, and the utilities worldwide had sped about almost $2 trillion upgrading the grid infrastructure, the grid being the largest most complex machine ever built.
It was to make all the devices in the infrastructure remotely machine addressable, transformers, substations, thermostats, you name it. And so they were set to be able to take advantage of these technologies to apply AI to optimize the power of power generation, transmission, distribution, and consumption to deliver safer, cleaner, more reliable energy into the hands of more satisfied consumers at lower cost and lower environmental impact. So C3 Energy really served the energy market.
Firstly, we used enterprise scale AI to address the market opportunity that was there at the time, and the market opportunity that was there at the time was almost exclusively utilities and oil and gas companies. Now, as we applied predictive analytics at massive scale to those value chains in Europe and in North America, it became apparent as this market developed now further say 2016, '17, '18 that there were other industries, virtually every industry that wanted to apply these technologies.
This is where we rebranded the company as C3.ai and to let it be known that we're open to do business in all markets, in manufacturing and supply chain, supply chain optimization, travel, transportation, healthcare, banking, professional services, law firms, health, you name it. We have been C3.ai I think since about 2016 or '17. As the market changed, we reformulate our product mix, our customer solution mix, and even our branding metaphors to meet the needs of the market.
So you need to stay on the balls of your feet when you're in developing markets that candidly, a market that didn't exist 20 years ago, that 10 years from now will be a, I suspect, multi-trillion dollar market upward.
Josh King:
I mean, staying on the balls of your feet and even thinking about the brand evolution requires some of maybe that liberal arts thinking that you mentioned like a guy like Alan Murray brings. You were one of seven children born and raised in Chicago for college and your MBA and your master's in computer science, you went not too far away in Champaign-Urbana. As a history major. What did you envision your career path to be?
Tom Siebel:
I studied the history of science. And I'll be honest with you, as an undergraduate, I had no idea what my career was going to be, but I enjoyed history. I enjoy history today. I mean, I'm constantly reading history and history of science. I'm glad I did it, and I think it clearly informs what I do, my decisions professionally in the last 40 years. It informs those decisions every day. I mean, history might not repeat itself, but it sure seems to rhyme.
And so these things play over and over again. And as you think about trends in information technology, you think about what's next. A good understanding of how we got here definitely helps you make well-informed guesses about what might happen next. And I think the guesses that we've made as it relates to relation database technology, as it relates to CRM, and as it relates to AI, let's just say that they seemed to be pretty lucky guesses.
Josh King:
Lucky guesses, indeed. Your legacy, Tom, at the University of Illinois continues to grow. You gave a $50 million donation this past April to establish the Siebel School of Computing and Data Science housed within the Thomas M. Siebel Center for Computer Science, which was dedicated back in 2004. How do you hope this school is going to prepare future Illinois students in this evolving area the way you were prepared?
Tom Siebel:
I owe a lot to the University of Illinois. That was a very important formative experience in a liberal arts education, a business education, and in engineering. And this is a land-grant institution, a public university. I am big advocate of the public education system. And like all of us who want to give back to the community and all of us who when you find yourself fortunate enough where you could maybe make a material contribution, I'm very interested in education and allowing broad based education to the public at large.
My association with the University of Illinois over the decades has been enormously rewarding and it is a humbling experience to be able to play a formative role in establishing the future of the University of Illinois so it can hopefully maintain its preeminence and maintain and continue to enable the public at large to become well-educated and succeed at what they do.
Josh King:
I mean, we had Steve Schwarzman in this room on this program talking about some of his initiatives at Yale and MIT, and certainly his material support comes with a vision that he has for those institutions among the others that he's supported. With your work at University of Illinois, the schools that you have created aim to focus on the intersection of computing and data science rather than thinking of them as separate disciplines. Why do you think it's important to emphasize the overlap and collaboration between these two growing fields?
Tom Siebel:
There was a trend in the earlier part of this decade, just many preeminent members of the academy, to separate data science from engineering and separate data science from computer science. I don't subscribe to that school now. The pendulum seems to be swinging back with the University of Illinois and now a couple of other leading members of the academy that I'm aware of that are merging data science, computer science, and engineering structurally.
And so at the University of Illinois, we have merged those structurally. Hopefully that will be a bell cow for others that keep these things closely entwined because I believe they belong in the same place. We're a massive consumer of human capital from these institutions of data scientists, of computer scientists, of design engineers, even liberal arts majors.
I think what we're doing at the University of Illinois is now we're forming a center for data science where data science will be promulgated across all disciplines at the university, including the Center for Design, where they have a wonderful design school. So our hope there was to establish leadership at the University of Illinois in a manner that hopefully others will follow.
Josh King:
Hopefully others will follow and hope for liberal arts majors like myself. So thank you. After the break, Tom Siebel and I are going to dive deeper into AI, take a look at C3.ai's recent fiscal fourth quarter earnings, brewing legislation and regulation around AI, and perhaps the future of AI. All that and more is coming up right after this.
Speaker 10:
Sail through the heart of historic cities and unforgettable scenery with Viking. Unpack once and get closer to iconic landmarks, local life, and cultural treasures. Because when you experience Europe on a Viking longship, you'll spend less time getting there and more time being there. Viking, exploring the world in comfort.
Josh King:
Welcome back. If you are enjoying this conversation and want to hear more from guests like Tom Siebel, founder, chairman, CEO of C3.ai, NYSE ticker symbol AI, please remember to subscribe to Inside the ICE House wherever you listen to your podcast. Give us a five star rating, if you will, and a review. Before the break, Tom and I were discussing his journey founding C3.ai, the company's evolution and how it's emerged as a leader in the global AI evolution and revolution.
Tom, enterprise AI has been a cornerstone of C3.ai since its inception with application spanning manufacturing, supply chain, oil and gas. We talked about that a lot before the break. How have you and the company leveraged this first to market advantage to stay ahead of your competitors and you have competitors?
Tom Siebel:
Well, understand, we did over a decade and $2 billion of software engineering, you could build the C3.ai platform, which solves some of the maybe less glamorous, but absolutely essential problems associated with the deployment of large scale enterprise AI, data fusion, data integrity, lineage, veracity, visibility, access control, authentication, encryption, motion encryption at rest. These are really difficult problems, and so they're grubby and not the sort of things that most people want to deal with, but you have to deal with these things to deploy applications at enterprise scale.
Now, for somebody else, now we've seen since November of '22 with ChatGPT, enterprise AI is now on the lips of every CEO, every military leader, and every leader in the academy in the world. And we have all these companies that with their software stacks that'll remain unnamed where they developed their software stacks in the last century and now they put AI on top of the box. So every company is an AI company. Well, let me tell you, that's a bunch of...
Josh King:
You can use the word.
Tom Siebel:
It's poppycock. It's poppycock. In order for those people to be in the AI business, they needed to do the thousands of person years of work that we did. They needed to spend the billions of dollars of work that we did to solve these hard problems. You can't simply take a large language model, stuff it in the box with your CRM application or AR application to declare that you're an AI company. I mean, this is laughable. Don't get me wrong, these are fine companies and these are friends of mine.
Many of them worked for me over the years at these companies that make these HR applications and these customer service applications and these CRM applications. Fine companies and led by great people. But they needed about 10 years of work to be able to support an enterprise AI application that they haven't done or get the technology foundation from us. And so we're in a unique position. The AI market is, I think, very confusing. It's confusing for investors.
It's confusing for business people, and it's confusing in the academy because everybody's AI today. I mean, if you can use ChatGPT, all of a sudden you're an enterprise AI company. Come on, any sixth grader can use ChatGPT. That an enterprise application does not. In enterprise application, we have access control, security, encryption, cyber. I mean, really difficult problems that we need to solve. We're unique in the space that we are really the first native AI company. The first. We started in 2009.
We've been playing out the same track record ever since and now. So we're a little bit unique, and it's a fun place to be. And we're doing everything we can to make sure that each and every one of our customers are successful. And I can tell you again, I mean, it's just the professional service experience of a lifetime to be able to serve. I think we provide some thought leadership in the space, and I know we provide technology leadership. This is a great professional experience for somebody who spent their life leading information technology companies.
Josh King:
I mean, talking about solving some of these large problems that you talk about, sometimes the origin story of solving a large problem begins with just a very simple little email or some outreach. And as I was watching that 2023 keynote you gave, you put up on the screen an email that you got from a friend on a Saturday afternoon, a leader who tends to work over the weekends, and a simple little one-liner.
He said, "Tom, I want to be the Google for DOD." How is generative AI fundamentally changing the nature of the enterprise AI market and creating opportunities for AI applications that may not have been expected as you rolled out your thought process and your reaction to this person and said, "Give me a couple of weeks to ideate on it?"
Tom Siebel:
I remember the email very well. It was a Saturday. I think it might've been September 17th. And it was, "Tom, I want to be the Google for DOD. The customer asks a question and get the answer." Hey, what a insightful question. Come on, give me a break. All I want to do is ask a question, get an answer. I think in the information technology business, and I take some responsibility for this because I've been involved in developing this human-computer interaction model at Oracle, at Siebel, at C3.ai.
I think we've really done a horrible job as an industry of building applications that people can use. We might have a lot of investors listening who use the Bloomberg Terminal, and I'm not saying the Bloomberg Terminal isn't a great product because it is a great product, but not many people can use more than 5% of the capability that's there. I mean, you need to control shift. It's really tough. Now, why not simply put... If you merge that with generative AI, all of a sudden, you have a Mosaic browser.
You remember the Mosaic browser. It came out of the University of Illinois in 1993, and then Google adopted it. You know it as the Google browser and everybody knows how to use it. It's a command line and you put in the question. You simply put in the question of tell me the correlation between sunspots and the price of IBM stock, or whatever it might be, and it'll draw you to the graph. You don't need 10 hours of training to figure out how to do that command.
Well, I think one of the most important aspects of generative AI is going to fundamentally change the nature of the human-computer experience. Think about this application that we talked about earlier that we have for the US Department of Defense related to predictive maintenance for weapon systems. Well, you can imagine that the human-computer interaction model there, the user interface, is a lot like the Bloomberg Terminal for that audience.
It's really only usable by a specialist, by maintenance specialist. Now, when you put a generative AI human interface on that like we have done, all of a sudden you have a command line where anybody can use it, the Secretary of Defense can use it, the chair of the Joint Chiefs can use it. You can ask any question about any weapon system or any weapon systems for the United States Military in the world instantly, say a minute, get the answer. For example, what are my readiness levels of my F-35 squadrons in Central Europe?
A question that somebody might've wanted to ask in the last year, and then it generates a map of Europe. It shows you exactly where your F-35 squadrons are. It tells you what the readiness levels of them are. It tells you what the source of ground truth was, where the answer came from. It doesn't hallucinate, doesn't data exfiltrate. It is cyber secure. It doesn't make up answers if it doesn't know it. And all of our data access controls are enforced.
Well, I mean, this as it relates to our transferring this technology and the change management issues that we need to do in these technologies in corporations and government entities, well, with regenerative AI, anybody can use them because everybody knows how to use the Google or the Mosaic browser. I think it's going to have a very important effect on change management and making these technologies available.
Another comment I would say. As it relates to AI, there was this belief that in order to survive as a professional, in order to operate in the new economy, that you needed to have a deep expertise in data science. Everybody needed data science. With all due respect, this is a bunch of poppycock. This is just not true. I mean, the way that we'll use these technologies, let's look at Microsoft, for example. These guys have done a wonderful job with Copilot. You don't need to know anything about data science.
You don't need to know anything about machine learning models, deep learning, neural networks, supervised learning, unsupervised learning, TD learning, reinforcement learning. You don't need to know anything to ask the question and get the answer. And so in legal applications, in medical applications, in business applications, in travel, transportation, this idea that we need to retrain all of the cab drivers in New York City in data science, this, again is a bunch of bunk.
We're simply going to have a PC on their, excuse me, a iPhone on their dashboard or an Android device on their dashboard that's going to tell them where to go, where to get the best fare, and the amount of data science that is running behind that application that they don't know about, I mean, it's mind-numbing. But the idea that we have to have this expertise in the population at large in data science is just not true.
Josh King:
In that quarterly earnings call, you mentioned earlier that you don't pay attention to where the stock price is at any particular day. But in that fiscal fourth quarter earnings call, C3.ai showed your 20% year-over-year revenue growth, and you said, "We exceeded all expectations for revenue, cashflow, and profitability. Let me be clear, there were no expectations that we did not exceed." What do you credit for the most recent performance of the companies you reported?
Tom Siebel:
I think a hugely expansive market, very large, and the market is expanding at a greater rate than anybody could anticipate. The levels in AI and enterprise AI are huge, and I would say the execution on the part of the organization was superlative. These people did a fine job in engineering, in products, in customer service, in sales, so the execution was quite good.
Josh King:
You also talked about the market dynamics associated with AI and how your primary competitor is trying to build versus buy. It's a debate that we also are thinking about and engaged with here at ICE. How do you foresee the build versus buy debate evolving in the marketplace? Do you think enterprises will lean toward buying solutions and products from people like you or will there still be a significant number of folks trying to do it by themselves?
Tom Siebel:
Again, we have seen this movie before. So when we introduced relational database technology to market in the '80s, everybody was going to build their own relational database. Now, I ask you, who succeeded at that? That would be nobody. When we introduced ERP technology and CRM technology to market in the '80s and later in the '90s, there is nobody that didn't try to build these systems themselves. IBM tried to build them. Microsoft tried to build them. Compact tried to build them.
Hewlett-Packard tried to build them. They all bought them from us after they failed three times because they didn't have... I mean, these are technologically articulate companies. This was before we get to General Motors and Pfizer. Nobody succeeded at building their own ERP application or CRM application. Nobody. Now, let's talk about enterprise AI. Honestly, this would be a order of magnitude, maybe two orders of magnitude more complex than CRM, ERP, or relational database.
This is really hard. The idea that some IT executive is going to succeed at taking 10,000 tinker toys that they buy from AWS or Google or get from the Apache open source, Hadoop Cloud, and cobble these things together into a system that's going to work at enterprise scale is candidly laughable. Now, everybody tries and all of our customers have tried and they've failed once. They failed twice. They failed three times. After they fail about three times, CEO gets a little cranky. Everybody's mad. They push the CIO off into a corner.
They bring somebody into their office that they designate to be a chief digital officer. He or she, the CEO, starts paying attention, him or herself, and they call us and they start getting things done. There's no way, no how that people are going to build their own enterprise application, software applications. This is like a pilot thinking that he or she is going to design something the scale of an Airbus. A pilot is trained to fly these plants, to take off and land, and push the autopilot button these are not aerospace engineers that can build subsonic intercontinental aircraft.
Josh King:
I mean, talking about calling us and getting something done, Tom, on that call, very entertaining. You said in the fourth quarter alone, we received almost 50,000 inquiries from 3,000 businesses, each with revenue greater than $500 million, all expressing interest in our generative AI application.
So of the roughly 50,000 inquiries, how many turned into deals and why are they choosing your application over the burgeoning AI products flooding the market? And I did see a slide of yours maybe on the website that talked about the timeline from when the call comes in to when they can see a product that they can put to use.
Tom Siebel:
In February alone, I believe it was 2018 in February, we received 10,500 inquiries, the CEO, CFO, the person in charge of supply chain and manufacturing, customer service, whatever it may be, at some of the world's leading companies of the world. I think last quarter in the fourth quarter, it exceeded 40,000. At this time, I expect the number to exceed 90,000 in the first quarter of this year. That ends in July. Now, honestly, very few of those turn into qualified opportunities.
So we have to go through the process of filtering them down. And by the way, we're using generative AI to do it. But the interest in generative AI is unlike anything I've seen in my professional career. Now, why do they do business with us rather than do it themself? The problem with these large learning models are you can't enforce access controls. The answers are random. They hallucinate. You have enormous cyber problems associated with data exfiltration and people being able to access your data.
They introduce very serious institutional IP liability problems. And so for any one of those reasons, that dog does not hunt and that large language model does not get installed at name the company or name the government organization. It won't pass. Today, people are interested in cybersecurity. They're interested in access controls. They're very interested. So by combining the large language model with all the capability of the C3.ai platform, the $2 billion worth of software engineering that we did, we can take advantage of these large language models so that the answers are deterministic.
Every time we ask the customer, we have the same answer. You can see exactly where ground truth came from, see where the answer came from, both in structured and non-structured data. There's no risk of LLM induced data exfiltration or LLM induced cyber risk. It doesn't hallucinate. We have no IP liability issues, and they're LLM agnostic. So as these LLM providers, be it from X, Grok or Open AI or Anthropic or Google or whoever it might be, as they out-innovate one another, we can simply change large language models and take advantage of the latest innovation.
So to the extent that people do business with us is that it passes their basic tests for security, for data integrity, and for safety. And that's why we are achieving success in the generated AI space and our product is very much unique in those aspects that I described.
Josh King:
You can think about all those inquiries and the funnel that you have to go through and the number that come out the other end of the funnel resulting in engagements and deals, but where you actually have at least quantified for investors and other observers of C3.ai. Last year, you closed 191 new agreements across 19 different industries, including 65 with federal agencies such as the Air Force, Navy, Marine Corps, and the IC, the Intelligence Community.
As you sit back, given your decades of experience and engagement with organizations in the federal government, and you could ask one of the service chiefs or the CIA director to come in because you've got a great idea that they aren't yet doing, what are some of the things that cross your mind?
Tom Siebel:
Well, I've been talking to them about enterprise AI and generative AI since 2014, and I have met with all of the people in all of those roles, the Secretary of the Air Force, the director of the CIA, the director of the NSA, virtually everybody, and I would say that these are very bright people. They're very receptive. They're very interested in operating highly ethical AI with a human in the loop.
We are at the table with these people every day helping them take advantage of these technologies. It's a privilege to be at the table, and it's a privilege to be able to serve. I hope what we're doing is important, I think it is, and that's the beginning and end.
Josh King:
We could be facing a change in administrations. There are also changes in leadership. Mark Milley no longer the chairman of the Joint Chiefs of Staff. Change of command with CQ Brown. Lisa Franchetti now the new CNO. How do you pivot to align with the new decision maker from the Oval Office on down?
Tom Siebel:
Well, whether it's a Republican administration or a Democrat administration, we've worked with them all over the decades. We do have separation of the military and the civilian branch. I think you find that the military decisions are not very well controlled by politics. These people are interested. These are professionals. They have a mission. Their mission is to secure the free world.
And whether it's a Democrat administration or Republican administration, there is no question that they will continue to make huge investments in enterprise ai. You can take that to the bank. I will be surprised that no matter what happens to the election, I think our conversations with them will continue and likely accelerate.
Josh King:
Beyond securing the free world. And while government contracts do drive a significant portion of the growth of C3.ai, you also operate, as you've mentioned, in so many diverse sectors, oil and gas, utilities, agribusiness, even consumer packaged goods. What's the current level of interest among these industries regarding your products and how are you leveraging your solutions to meet their needs?
Tom Siebel:
It's growing. I mean, it's growing in manufacturing. It's growing in oil and gas. It's growing in agribusiness, who are huge, believe it or not, and sugar cane production in Central America. Who would've thought? Supply chain and delivery of protein. $100 billion worth of protein products for Cargill. I mean, if we get that wrong, North Africa starves.
So this is really important stuff. I think it's evident to everybody listening and everybody who's alert that this enterprise AI phenomenon, their high levels of interest is growing very rapidly. Budgets are expanding, interest is expanding. It's quite a professional experience to be able to be at the table.
Josh King:
I can't let you leave the library of the New York Stock Exchange and our conversation without at least touching for a moment on the discourse around the risks about ai. I want to listen to OpenAI CEO Sam Altman in his testimony before Congress highlighting the potential risks associated with generative AI. I want to listen to some of the concerns he expressed.
Sam Altman:
Worst fears are that we cause significant, we the field, the technology, the industry, cause significant harm to the world. I think that could happen in a lot of different ways. It's why we started the company. It's a big part of why I'm here today and why we've been here in the past and we've been able to spend some time with you.
I think if this technology goes wrong, it can go quite wrong, and we want to be vocal about that. We want to work with the government to prevent that from happening, but we try to be very clear-eyed about what the downside case is and the work that we have to do to mitigate that.
Josh King:
Sam emphasized the need for proactive collaboration with government to mitigate the risks. Do you think that intervention, whether through legislation or regulation, is necessary?
Tom Siebel:
Absolutely it's necessary. I mean, it's not a question of whether AI will be used malevolently. It is being used malevolently today. Let's see social media for details. I mean, come on. Look at the impact of social media on young women in the United States and Europe. Well-documented, well-established that these companies are using AI to manipulate young people, the level of the human brain. We have, I'm sorry, the limbic brain. This has to do with the release of neuroreceptors like dopamine and serotonin and what have you.
But the consequences, loneliness, depression, suicide, body image issues. We know that these technologies are manipulated by rogue states and bad actors to interfere with the democratic process. See the last election cycle for details. See this election cycle for details. I don't think we have to be so worried about the idea of, gee, we're going to have sentient computing and your smart refrigerator is going to take over your house or your large language model is going to take over your life.
Get over that. When we train these large language models from places like OpenAI with Sam, I mean, he might be spooling up a couple hundred million dollars of computing capacity and 20,000 H100 GPUs. That requires a gigawatt of power to generate something that really doesn't have any intellectual capacity and might have the intellect of a mealworm. I mean, the human brain has 60 billion neurons, makes 100 trillion connections, and all these connections are analog, again, with dopamine and with neurotransmitters and neuromodulators, and operates on 17 watts.
And the limbic brain in a few seconds can come up with the... I mean, the idea that a computer is going to compete with that, it isn't going to happen anytime soon. So get over the sentient computing threat and the hell scenario from Space Odyssey. And I would worry about what's happening now. I mean, social media is the primary exchange trade for slavery in the world, exchange medium for slavery. There are 40 million people enslaved on the planet today.
Nobody cares. Everybody wants to talk about the 16th, 17th, 18th centuries. Were there slaves? Yeah. When weren't there slaves? But today there was never been 40 million before. Never been 40 million. We have 40 million people enslaved on the planet today. Primary exchange trade is social media, and nobody is doing anything about it. Nobody is saying anything about it. I mean, come on. And so, yes, I think there is a role for government and there is a role for entities to regulate themselves.
I am very, very outspoken about the threats of AI and the threats of generative AI, and we need to be afraid because uncontrolled, this goes to a bad place.
Josh King:
One of the things that Secretary Rice said in her conversation with you is they can spell AI in Washington, but I'm not sure they understand what it is. You have been complimentary of California's Governor Gavin Newsom, his Executive Order N-12-23. It was implemented last year. It requires extensive studies to be done of the risks for artificial intelligence. Should states and the federal government be mimicking Governor Newsom's approach prior to enacting any legislation regulation on AI?
Tom Siebel:
I thought the executive order in California was pretty well considered, and it wasn't one of these crazy regulatory schemes. It said this was a mandate upon the governor that every department in the State of California study AI and study generative AI and come up with an analysis of how can we use these technologies to benefit the people of California and what are the risks to the people of California and how we can mitigate them. So I thought that was a very well considered executive order.
Unlike some of these things, for those of you who want some reading, read the 553 pages associated with the EU AI Act. I mean, candidly, it's kind of crazy. It's this idea that we're going to establish this large bureaucracy to inspect all these dual use foundational models to ensure that they're safe. And the question is, how can we have a regulatory agency ensure that they're safe when honestly, Sam Altman doesn't even know how ChatGPT works. He doesn't know how it works.
We don't know how these large language models work. There really is voodoo involved in this. It's not deterministic. It's not pure math like it used to be in computer science, and it's impossible to... If we, who develop them, can't describe how they work, what's the probability of that regulator is going to be able to look at them and determine they're safe? That would be zero. At the same time, we could pass laws. We could pass laws to make it unlawful to publish an algorithm that can be used to create a public health hazard, as I described in social media.
It can make it unlawful to publish an algorithm. It'll make the publisher required, rather than prevent somebody from publishing an algorithm until it's been certified the government. What we're doing is we're just criminalizing science and criminalizing every 18, 19 year or 20-year-old at NYU or Princeton who is the carriage return to put... I mean, what's the chances they're going to stop? They're not going to stop. They're going to publish it.
But why criminalize science when we can put the people who publish them, people like C3, Facebook, others, and make them responsible to make sure that what they're doing is safe, doesn't cause harm to society? I think there's a very easy cure to this, pass a few laws, enforce the laws, and it'll go a long, long way to mitigate the very serious social risks associated with AI.
Josh King:
As we wrap up, Tom, earlier this week on Fox Business regarding AI and specifically enterprise AI, you said we are in the first half of the first inning and the first batter is on its way to the plate. If the pitcher is yet to even throw the game's first pitch, what do innings three, four, five and maybe all the way to the bottom of the ninth look like?
Tom Siebel:
It's unknowable. It's honestly unknowable. This future is vast. This universe is expanding at a very rapid rate, and where this goes is really unimaginable. But we need to know, we need to be sure that at every step that we're doing the right thing, that we don't take this to a bad place. If it goes to a bad place, it goes really bad. We can have healthier population. We will have more jobs as a result, not less jobs as a result.
We can have more productive society. We can have lower mental impact. We can have peace. We can have prosperity. These are good things. This could also get very Orwellian, and so we need to make sure it doesn't go there. To predict where this might go in 10 or 20 years, it's just impossible because it's so vast and so exciting and so enormous. So I'm not going to take a shot at that.
Josh King:
I won't ask you to take a shot at that, and I'll only end by saying we don't know what's happening in four years from now. But in the four years since C3.ai has been listed in the New York Stock Exchange, we have been honored and thrilled to have your company listed and your shares traded here. Thanks so much for joining us Inside the ICE House.
Tom Siebel:
Thank you, sir.
Josh King:
That's our conversation for this week. Our guest was Tom Siebel, founder, chairman, and CEO of C3.ai, ticker symbol AI. If you like what you heard, please rate us on Apple Podcasts so other folks know where to find us. If you've got a comment or a question you'd like one of our experts to tackle on a future show or to hear from one of our listed CEOs like Tom Siebel, make sure to leave us a review.
Email us at [email protected] or tweet at us @icehousepodcast. Our show is produced by Lance Glynn with production assistance, editing, engineering from Ken Abel. Pete Ash is the director of programming and production at ICE. And I'm Josh King, your host, signing off from the library of the New York Stock Exchange. Thanks for listening. We'll talk to you next week.
Announcer:
Information contained in this podcast was obtained in part from publicly available sources and not independently verified. Neither ICE nor its affiliates make any representations or warranties, express or implied, as to the accuracy or completeness of the information, and do not sponsor, approve, or endorse any of the content herein, all of which is presented solely for informational and educational purposes.
Nothing herein constitutes an offer to sell, a solicitation of an offer to buy any security, or a recommendation of any security or trading practice. Some portions of the proceeding conversation may have been edited for the purpose of length or clarity.
