This is Tim Horan, the Cloud Analyst here at Oppenheimer. But I'd like to say I'm also cloud now and AI, if that is becoming a sector, because after I'm hosting this today, I have to go. I'm doing an AI podcast, and tomorrow, I'm speaking internally. I have to do another AI presentation. So it's definitely a very, very topical time here to be talking to a legend in Silicon Valley, Tom Siebel, who's been very, very early in the most important technology trends that consistently in America and globally. And Tom, basically invented the SaaS industry and kind of is one of the few pure play AI companies that are out there. So Tom, thanks so much for joining us, and appreciate it.
If you want to bring up some slides at any time, feel free to do so. But maybe, Tom, I was just gonna start out with, this might sound like a crazy request, but can you describe your company, like an elevator pitch? And then maybe in a little bit more detail, what do you think your core competencies are? And kind of what's the moat around what you guys do?
Our core competency. Good morning. Thanks for having us. and our core competency, enterprise application software. So we've been in this business now for enterprise application software business for over four decades. I was one of the early guys at Oracle, where we built, you know, the Oracle relational database and the applications. And as one of the people who ultimately ran that business in 1993, we thought about the application of information technology, and communication technology to sales, marketing, and customer service. At Siebel Systems, we invented the CRM market as you know it today, and by, I think, 2003, we had 85% market share globally in sales, marketing, and customer service. Ultimately, that company was sold to Oracle.
In 2006, after Siebel was sold to Oracle, we started thinking about what was next in enterprise computing, and we thought, and again, we're thinking about applications, okay? Applications, ERP, CRM, supply chain, what have you. And we thought next was about when we looked at the infrastructure layer, we thought next was gonna be about elastic cloud computing. We thought next was going to be about the Internet of Things. We thought next was gonna be about big data. So we started this work in 2008. Understand in 2008, AWS is, like, this big. Azure doesn't exist. Google Cloud doesn't exist. The GPU doesn't exist. Nobody talked about enterprise AI.
So we started in earnest in January 2009, and in the next 12-15 years, spent $2 billion-$3 billion building a software stack that the idea of the software stack was we would provide all of the services necessary to design, develop, provision, and operate enterprise AI applications. And so this was, you know, a pretty substantial piece of work. And let me see. I might be able to even share a slide here. I don't know. Is this gonna work if I go here? No. So then after we built the platform, we then built 90 turnkey enterprise AI applications. Give me just a second. I'm gonna show you a slide. There we go. Get to there. Get to there. Get share. Share. That slide.
Am I sharing a slide?
Yep, it looks great.
Perfect. So this, so this was the first platform, and the idea was this is the first AI platform, and this is $2 billion-3 billion worth of software engineering. And the idea was to provide all the, the services necessary and sufficient to design, develop, provision, operate kind of massive-scale enterprise AI applications. And then we used this platform to build 90, count 'em, 90 turnkey enterprise AI applications for financial services, manufacturing, aerospace and defense, health, telco, oil and gas, utilities, what have you. Well, when we get into 2015, 2016, 2017, 2018, 2019, even 2020, hey, C3, I'm the only guy in the world who's talking about enterprise AI, guys.
We're, we have heads down building these applications, and then we get beyond. Now we are in the middle of 2024, and the world isn't talking about anything but enterprise AI. Okay, and we have, you know, 90 turnkey enterprise AI applications to service this marketplace, which would be probably about 89 more than anybody else in the world. Now, we have deployed some of the largest enterprise AI applications on Earth into the, some, some of the world's largest corporations: Shell, ExxonMobil, United States Air Force, Defense community, Intelligence community, what have you. Now, after November 2022, after November 2022, with the introduction of, ChatGPT, everybody in the world who has a software stack that they built in the last century, you know, puts AI on their box.
You know, Dot Ai, Salesforce, ServiceNow, oh, you know, all of these guys. All of a sudden, now, everybody's in the enterprise AI business, which is kind of curious. But I think that, you know, really, the, the slide that says it all is, you know, is based around, you know, who is receiving value. Okay, these are, you know, Net Promoter Scores, that you can see on the website of Comparably, of C3 versus everybody else in the world. The bottom line, we have the most satisfied customers on the planet who are delivering, who are receiving very high value. It looks like, I think it's generally believed now that enterprise AI is a pretty big market sector. We are, enormously well-positioned, to establish and maintain a market leadership position in this market.
We went public in 2020, December 2020, I think, as I recall, we raised about $1 billion. We've been focused on, you know, we restructured our pricing model from subscription-based pricing to consumption-based pricing, which had a short-term effect of breaking our growth rates down from, like, order of 60% growth to, like, below 0%. And now we've seen it with the consumption-based pricing model kicking in nicely. It's returned in the last five quarters, as I recall, I think it went from -7% to 0%, to 3%, to 7%, to 11%, and I think last year's last quarter's revenue growth, if my memory serves me correct, was 20% year-over-year.
We believe we've announced that we expect to see, in this fiscal year, 23%, year-over-year revenue growth. So we have a rapidly growing software company. We, you know, are very well capitalized, and we're in a position to establish and maintain a market leadership position in the enterprise applications layer going forward. So this is... So we're not focused on silicon, like NVIDIA. We're not focused on infrastructure like AWS, Azure, and Google Cloud. We're focused on applications, and in the long run, when this becomes a, well, not very long run, okay, a $2 billion addressable market opportunity, the bulk of that value will be in the applications, not in the silicon and not in the infrastructure. So in a nutshell, this is what we do, and this is, this is what the business is about.
And, just stepping back, I want to talk about the platform layer, then maybe then the applications layer, and I definitely got to talk to you about what's going on in overall AI. So I'm not even sure where to start, but actually, maybe we'll just start in the overall AI market. There's just a lot of concern right now that, the hyperscalers or maybe the industry is spending too much on CapEx, and the application revenue is not kind of coming through, you know, for whatever reason. And I, I'm oversimplifying things, but, I guess, do you share investors' concerns?
Well, I think AI presents a headwind for a lot of these software companies. I really do. You know, I mean, they position themselves as, you know, a lot of these companies that have, you know, 20th century software stacks, you know, that remain unnamed, but a lot of them, you know, are on this slide. Okay, there's no AI in this stuff, and so you know, the dog doesn't hunt. I think traditionally what people have done, beginning with, you'll remember the Apache open source Hadoop stack, and then later with the components from the infrastructure providers, is they've attempted to build these applications, and that's not worked out so well. Nobody wants to build applications, guys. I mean, how many companies want to build their own Bloomberg Terminal?
How many companies want to build their own ERP system? How many companies want to build their own CRM system, and how many companies want to build their own, you know, whatever it may be, I mean, relational database? I mean, nobody does that stuff. What in the world makes you think that they're going to be able to build these enterprise AI applications? They're more than an order of magnitude more complex than these other applications. What the people want to do is they want to buy turnkey enterprise applications that solve real-world business problems, like customer churn, like inventory optimization, like fraud, like grid optimization, like optimization of the supply chain, like readiness for the defense community. And so that's, in the long run, that's where this market goes.
This is the game that we're playing, and I think it'll play out quite well.
What are the other roadblocks to applying Gen AI technology to the enterprise, and why are you able to do it better than others?
That's a great question. You know, this GenAI thing is really big, okay? This is a powerful technology, and it's very confusing because there's so much out there. You know, whether we're dealing with Llama, Anthropic, ChatGPT, Mixtral, every one of these technologies, they're very interesting. They solve an interesting class of problems, but they have a number of problems associated with them that make it so that enterprises won't install them. For example, the answers are random. For example, when you're dealing with structured data and unstructured data, you can't tell where the answer came from, so you can't. It's not traceable to ground truth. None of them, of course, enter any of our enterprise access controls.
We have enormous LLM-caused cybersecurity risks and attack vectors that are now being well documented by Carnegie Mellon and other places. We have this problem with LLM-caused data exfiltration, where all of a sudden, for example, the classic case is the Samsung case, where all their intellectual property got broadcast on the internet. We have these hallucination problems that are just fascinating, and finally, and not insignificantly, we have these IP liability problems where people are using large language models that are trained on the internet, the corpus of the internet. Well, guys, I mean, the corpus of the internet is all somebody else's copyrighted information, be it Bloomberg, the Weather Company, whoever it might be, or Stephen King, and those people want to be paid... So, for any one of those reasons, this dog doesn't hunt.
Now, when we combine the, see, I think I have a slide that shows this. When we combine. You see, in the C3 AI platform, we've solved all the problems associated with, you know, with cyber, with access control, with N-factor authentication. I mean, all of these problems that we've solved. So when we combine Generative AI with the C3 AI platform, number one, we have omni-modal, not multi-modal, omni-modal data infusion. We can take any data, telemetry, structured data, unstructured data, ERP data, you know, whatever it may be. Now, we've solved because we. The way that we use the LLM, we've separated the LLM from the data. The LLM has no access to the data, and so as a result of this architecture, we, when we use the LLM.
By the way, this is LLM-agnostic, so we can use Llama, we can use Chat, we can use Anthropic, we can use Mixtral, whatever it might be. When we deploy these applications, we, the responses are deterministic. No matter how many times we ask the question, we get the same answer. Every answer is traceable to ground truth. All of our enterprise access controls are enforced. There's no incremental cybersecurity risk. There's no LLM-caused data exfiltration. It's hallucination-free. We've eliminated the problem of IP liability exposure because we're not using these language models as they were trained on the internet.
We're using them just as a blank slate, and then we're building, if you will, an enterprise language model or a small language model on the data of the entity, be it the Air Force, or Bank of America, or Koch Industries, or whoever it might be. And finally, these solutions are LLM-agnostic, so that, as vendors out-innovate one another, as they are at just breathtaking pace, our customers can take advantage of whatever the hot ticket is of the day. So we've solved all these problems, and so for us, this is a huge opportunity, and we have our... We have 30 Generative AI solutions in the market today.
We have Generative AI for Salesforce, Generative AI for Workday, Generative AI for SAP, Generative AI for ServiceNow, Generative AI for all the C3 AI enterprise applications, Generative AI for public benefits programs, like Affordable Care Act, Obamacare, Medicaid, Medicare, what have you. So this is a huge and rapidly growing business. I think this represents, you know, Generative AI, I suspect, doubles the size of the addressable market for enterprise AI, and in its own right, is probably a trillion-dollar addressable market. So this is a genuine big deal.
What are some of the cooler innovations you're seeing? I mean, we've seen a lot of upgrades to the large language models in the last couple of weeks. You're closer to this than anybody. Are these major, you know, improvements, do you think?
Well, I think what we're gonna see in innovation in large language models in the next five years is gonna be just breathtaking, and I don't think... You know, everybody kind of assumes that it's one of OpenAI, or Anthropic, or Google that's gonna win this. I don't know why you would think that. I mean, this could be anybody out of the Bronx, or Paris, or Munich. But, I mean, we're gonna see massive innovations. For those of you who are interested in this field, you know, there's a book out called, you know, What Is ChatGPT Doing ... and Why Does It Work? by Stephen Wolfram. Yeah, it's available on the internet for $14. I strongly recommend this book. I mean, it'll explain to you exactly what these large language models do, what they will never do, how they work.
You'll also understand that when we get into these generative pre-trained transformers, that there really is voodoo involved. And the fact of the matter is, while they're highly beneficial, we really can't explain how they work. That may be difficult to believe, but it is true. But I strongly recommend that book to anybody who's interested, and I assume everybody on this call is interested.
What are you seeing for the cost to inference these AI models? How's that been trending in the last year or so?
I mean, I think if you want to do the corpus of everything that's been written or the corpus of the internet, you know, it can get expensive. You know, I think that, you know, there's one of the companies listed on this page, you know, that I showed you before, that built their own language model. It cost them about $10 million in inference. Okay, so, you know, so I suppose it can cost anywhere between $10 million and $100 million to train one of these guys. But when you're training on an enterprise language model, let's say you want to use SEC.gov to take all of the S-1s, all of the 10-Ks, all of the 10-Qs, because you want to use it, Generative AI, in a law firm to generate S-1s.
I mean, it's not a significant problem, and so... Or you're dealing with, you know, maybe, you know, hundreds of terabytes of maintenance data associated with all the weapons systems in the United States Air Force. The, you know, the inference costs there are, they're honestly trivial, and so it's insignificant.... So when we deal with real-world medical applications, defense applications, supply chain applications, manufacturing applications, the training and inference costs are really not that significant.
Are they coming down, the cost?
Well, I think that people are putting a lot of pressure on the, you know, there is a lot of pressure on the, on the cloud providers to lower their costs. And, you know, it is a highly competitive environment, and, these guys don't like to lose. And, they're all run by very competent people, and they tend to meet the price point necessary to close the deal. So I think we see. There's, it's unquestionable that we see the price of inference, and the price of cloud services, you know, coming down in the next decade. That doesn't mean these guys won't continue to print money. I suspect they will, but the price for our customers comes down.
And Tom, there is a lot of concern. I mean, their CapEx has gone up a huge amount. Are they gonna be able to monetize that capital spend, the hyperscalers? Do you have any thoughts on it? I mean, I know you're not inside with working with them directly, but, it does seem like a huge amount of money.
Yeah, good question. Not my business. I don't understand. You know, I don't really care about the economics of the hyperscalers. I, I am quite confident that the hyperscalers will be providing virtually infinite computing capacity, infinite inference and training capacity, and infinite storage capacity at acceptable costs. But what the economics are of their business model is just not something I pay attention to. And-
Yeah, and it's, frankly, it wasn't a fair question, but I'm getting asked, and I'll, I'm getting-
A legitimate question, just not a field where I have, I'm competent to respond intelligently.
So, I know you don't have any real peers out there, but one of the closer ones, I guess, would be Palantir, maybe. You know, they said they're on their earnings call two weeks ago, they're basically seeing unprecedented demand out there. You know, I know you haven't reported yet, but can you comment on what you maybe said last publicly about, you know, the demand and your ability to meet that demand?
Well, I suspect, you know, Palantir, you know. I don't know that much about Palantir. I suspect it's a pretty good company. You know, I'm pretty impressed that the investment community doesn't make them distinguish between services revenue and license revenue. I mean, it's fascinating that somebody would invest in a company that is hiding the ball like that. It looks to us like a services company masquerading as a software company. But that being said, they're certainly making a lot of revenue. I suspect they will continue to make a lot of revenue, and you know, good for them. And you know, they don't have to lose for us to win. I mean, this is a big market opportunity, and we're.
You know, I wish them all the best. You know, this is kind of an interesting slide. You know, if we look at, you know, if we look at where the value is in AI in the long run, you know, if we think back to the PC market or the mini computer market, you know, as these markets began. Let's look at the PC market in the late 1980s, okay? You know, a PC in today's dollars in the late 1980s cost about $17,000, okay? And on top of that, who was king? Intel was king. They were providing the chips. Microsoft was king. They were providing the operating system infrastructure.
And on that computer, you would run about $200-$300 worth of software that we used to get from WordStar or Lotus 1-2-3, right? And so we'd have $1,000, thousands of dollars worth of infrastructure, and all the value was perceived of in, as being in the silicon, okay, and be in the infrastructure. But in the long run, I mean, how much value is in the infrastructure today? I mean, come on, does Intel still exist? Do they matter? And, you know, this PC that all the, that all of us have, you know, on this PC, it's costing our company about $300 a year in terms of capitalized, in terms of amortized cost to own the hardware.
We're spending about $200-$300 a year on infrastructure, and we're spending about $8,000 a year on applications: Bloomberg, SAP, CRM, okay, et cetera, $8,000 a year. Well, in the long run, this is how it plays out in the AI market. Today, all the value is perceived of as being in the silicon with NVIDIA. Great company. I think NVIDIA will always be a great company. It's perceived of as being in the infrastructure. See, you know, Microsoft, AWS, Google Cloud. Are these good companies? No, they're great companies, okay? But in the long run, we know that the silicon and the infrastructure gets commoditized, and 80% of the value is gonna be in the applications, and this is the place, part of the market where C3 plays.
The name of the game that C3 AI is playing is simply to establish and maintain a market leadership position in enterprise AI applications. You know, we are a well-capitalized company. I think we have about, last time we announced, plus or minus something, roughly $750 million in the bank. I love these analyst reports that, you know, comment on how we're, like, hemorrhaging cash, when every quarter, at the end of the quarter, I look, we seem to have about $750 million left in the bank. But, you know, these sell-side guys need something to write. Present company excluded. But I think, you know, we're well-capitalized. We're in a position now where our 90 applications. We're growing our application footprint, okay?
We're making the application much more accessible. The applications are now becoming available on, and are available on the Azure marketplace, the AWS marketplace, and the Google marketplace. So we can see our business going in the years ahead from scores of users, to hundreds of users, to thousands and tens of thousands of users.
Well, Tom, that's a great point. I mean, now that they're on the marketplace or, like, I guess, how portable are these applications or how easy is it for the enterprises to adopt these applications now?
Well, you can, you know, any one of you right now can sign on to the AWS or Google marketplace today, download, you know, a Generative AI for the enterprise, and you'll be running it in 10 minutes. Do it. I think it costs you $1.
Right.
Please do it, and you know, go sign on to that. I think it cost $1 to get started. It might be $10, but it's you know not... It's not significantly non-zero. But the applications are available there today, and then we sell also our enterprise applications through the marketplaces also. Those are, of course, larger app, larger you know, the amount of money involved is greater, but we're very much concentrated on making it so that people can get the training online, you know, be up and running in an application in you know it kind of very quickly.
Guys, we take these applications, we take these enterprise applications, like supply chain optimization, production optimization for supply chain optimization for companies that deliver $100 billion in protein, process optimization for organizations that deliver, you know, tens of billions of dollars' worth of polyethylene, oil, energy. We optimize these applications, and our pitch to the people who run these businesses is, we will bring this application live, okay, in six months for $500,000. $500,000, guys, that's nothing. They don't need an RFP, they don't need a budget. I mean, the person we're dealing with just writes the check. If you like it, keep it. Okay, and, and-
What, what's happening?
That's the value proposition.
[crosstalk]
The alternative is, go to any other vendor or go to Accenture, and they'll build this application for you in three years for like, you know, $60 million. In three years, maybe, then you get to the change orders. So we have a very attractive... This switch to consumption-based pricing gives us a very attractive value proposition, and you've seen in our results, you know, the number of transactions that we're doing in every quarter has been increasing dramatically, as has our top-line growth rate.
What type of productivity, like the supply chain examples you use, what, what type of productivity improvements are they seeing, and, and what's the payback period?
Well, just in the inventory optimization, I mean, it's easy to reduce the amount of inventory, okay, in the supply chain, aka in the whole supply chain by 30%-40%. That's a no, that's a no-brainer, okay? Now, the other, more importantly, though, is supply network risk, where we're dealing with large, complex global supply chains. And, you know, everybody today is dealing with contested logistics. All logistics are contested. And, you know, we can identify disruptions in the supply chain, okay, before they occur, and then put mitigations in place so the customer can, you know, deliver the product, you know, on time, in full, and the economic benefit is in $ billions. I mean, if you deal with, you know, Shell.
Shell has said, you know, in public, the economic benefit that they're getting from C3 AI is in excess of $2 billion a year. $2 billion a year, okay? This is across predictive maintenance, this is across production optimization in wells, process optimization and refining, hydrocarbon loss accounting, integration of renewables at a global scale. Koch Industries has said that they're out to get—this would be, Koch would be Georgia-Pacific, Flint Hills Resources, and what have you. I mean, they're looking for, you know, $1 billion in economic benefit a year. Guys, you mean—I've been in the enterprise application software a while, okay? I can tell you, no Oracle customer, no SAP customer, no Siebel customer, no Salesforce customer has ever stood up on stage and said, "We're getting an...
Okay, we're getting economic benefit, $1 billion a year." Okay, it never happened, okay, but it has happened. It does happen to us, and if you go on our site, go on our website, look at what our customers say. There's hundreds of videos of our customers on our website, from our users conference and others, attesting to the economic value they're getting, attesting to the rapid time to value that they're seeing from these applications. This is not AI hype, guys. This is value, and let's remember, we began this in 2009, okay? From 2009 to 2022, we're the only company in the world talking about enterprise AI, and now the world has kind of come our way.
So I do have a question from the audience here, and if anyone wants to ask one, just chat to me. The question is: Any additional information on the 50,000 inquiries C3 AI received last quarter? Those, like, are you seeing any of them convert to sales? And, the second part of the question is: Can you keep up with the demand?
Yeah, good question. Okay, first of all, I think I said, again, my recollection, I said that last quarter we received 34,000, okay? But I could be wrong. Whatever, what did I say, Amit? Oh, I said 50? I think we expected, I think I announced that we expected to see in the current quarter that just ended, order of 100,000 inquiries. Okay, I'm not gonna comment on whether that happened or not, but again, I would tell you the demand for what we're seeing is growing very, very rapidly. Okay, what was the second part of the question?
Can you keep up with the demand?
Can we keep up with the demand? I mean, keeping up with the demand is basically, guys, it's a function of sales and service capacity, okay? And the sales and service capacity translates into market share, okay. Sales and service capacity also translates into, you know, spending money on people in advance of revenue. So if, you know, the market right now is very sensitive about every company being cash, cash positive profitable, I think people forget, you know, how many quarters it took for Amazon to be profitable. I think that was, if my memory serves me correct, 29 quarters, okay?
Years.
Oh, I'm sorry.
Years.
29 years, okay, for Amazon to be profitable. How, how'd that work out for their shareholders, okay? And Salesforce, 25 years till they became profitable. 25 years before they got profitable. How'd that work out for Mark's shareholders? Worked out pretty well. I was a nominal shareholder at Siebel, running a cash-positive, profitable business. Mark just spent money like crazy. It worked out pretty well for his shareholders. Took Apple 21 years, okay, to become profitable. 21 years. People forget this. Not 21 quarters, 21 years. Now, okay, right now, the investor community is intolerant of companies that, you know, are investing in market share, you know, because you need to be cash positive, you need to be cash positive right now.
So that kind of gates the rate at which we- which we're kind of investing in sales capacity and service capacity, and gates and gates the, the, our growth, our top-line growth rates. I think if I had a brain in the head, in my head, I would take the $750 million that we have right now, I'd invest it in sales capacity, and I'd invest it in service capacity, and I'd go out and raise another $2 billion. But I don't think the market will tolerate it, so we're doing, you know, so we're going for lower growth rates, although they're accelerating now quite rapidly, going through... I think we were predicting 23% for this year, which is, I think, one of the top five public software companies that exists, if I'm not mistaken, okay?
The, you know, that's the trade-off.
I do have another question here. I haven't read it, but I'm just gonna read it. Can you tell me how you're thinking about how to price your services? And if you were achieving a $2 billion impact at Shell, it seems they should be paying you at least $100 million if you're saving them $2 billion. I don't know if you can answer that one or either of them, but how do you price your services, I guess?
I agree. They should be. Today, we're priced basically on consumption, where people are paying us based on volume between, I think it's $0.22-$0.50 per vCPU hour or vGPU hour. So the amount that they pay us is associated with the amount that they use the application. Let's just say it's working well for C3. Our revenue is growing, you know, quite well. I think in the last quarter that we announced, I ran a cash-positive business, right?
Mm-hmm, $18.8 million.
How about out, out-
$18.8 million.
Okay, so we're $18.8 million. So we finished the quarter with roughly $20 million greater than when we began it. I think I've announced that we expect this year, fiscal year 2025, to be cash positive. So, I mean, look, this is what our customers say. It's working for them, it's working for us, everybody's happy. I think their pricing model is great.
Your federal business, another question, grew really, really well last year. Can you talk about the opportunity in federal? Can you, you know, keep growing at those type of rates?
I think the federal opportunity is huge, guys. I mean, you know, the I would say we've been primarily focused on the defense and intelligence markets. And the defense and intelligence, I mean, this is roughly $900 billion of a $7 trillion dollar, I mean, it wasn't that long ago the federal spending was, like, $4 trillion, $3 trillion. You know, now it's like you know, that's just a few years ago, okay? Now it's, like, four years ago. Okay, now it's, like, $7 trillion. So the, you know, defense and intelligence is almost kind of rounding error in the overall, in the overall federal budget. This is where we've expanded the bulk of our effort in the last few years.
It's, you know, I could- I am quite confident the defense community and the intelligence community will be investing massive amounts in AI associated with prosecuting the next war or being prepared to prosecute the war- next war, whether that be in INDOPACOM or the Middle East. The, you know, this- these AI capabilities are, you know, critical to the kill chain technologies that will be in effect at that time, and we're kind of very active in those discussions, as you know. I think the real opportunity that is before us, where we haven't invested enough yet, is in the civilian sector. I mean, my recollection is, Health and Human Services, HHS, has a, you know, almost a $2 trillion annual budget. But here we have, you know, whether, you know, veterans, CMS, health-...
HHS, at least I think, Internal Revenue Service, Justice, there are enormous applications of AI to allow government to provide, you know, higher quality services at lower cost to more satisfied constituents, and we're kind of very active in that. One of the areas, and we're seeing this at the state level and the local level, the state level, for example, with benefits programs like Affordable Care Act, Medicare, you can think about, when you think about the Affordable Care Act, as you can imagine, the amount of regulations associated with that are biblical in scale. It's incomprehensible, and these are administered by the states. So there we have the, the issues associated with all of the insurance regulations, say, for the state of New Jersey or the state of Maryland or the state of Texas.
Okay, and then we have to combine that with all of the rules and regulations that's sort of the Affordable Care Act. What is that national application of Generative AI, where we load that corpus of data into a learning model, and then this is a hugely successful application for us, which is C3 Generative AI for public benefits. And then with a Mosaic browser, people can ask, you know, any question in 133 languages and get the answer in any one of 133 whatever language they ask, about how to enroll, how to get their daughter enrolled, how to get their grandmother enrolled, what's in scope, what's out of scope, what have you. And so we're gonna see.
I think the budget for, in Health and Human Services, Veterans and Social Security to increase customer service in the current budget cycle is $250 billion. $250 billion just to improve customer service, guys. I mean, compared to this, to the whole defense budget of, like, I think my recollection is that was $860 billion. So these numbers are big, and we see... Well, you can see us putting much bigger emphasis going forward on the civilian sector, while we continue to advance our technologies as they become programs of record in the defense community with, you know, readiness, contested logistics, surveillance technologies, law enforcement, what have you.
Well, Tom, I can't wait to listen to your earnings call. I think it's gonna be one of the better ones of the season here. Thanks a lot for spending time with us today, and we'll talk soon.
Thank you so much. Everybody, have a great day!