Good day, and thank you for standing by. Welcome to the C3 AI Second Quarter Fiscal Year ‘24 Conference Call. At this time, all participants are in a listen-only mode. After the speakers’ presentations, there will be a question-and-answer session. [Operator Instructions] Please be advised that today’s call is being recorded. I would now like to turn the conference over to your host, Mr. Amit Berry. Please begin.

Good afternoon, and welcome to C3 AI’s earnings call for the second quarter of fiscal year 2024, which ended on October 31, 2023. My name is Amit Berry, and I lead Investor Relations at C3 AI. With me on the call today is Tom Siebel, Chairman and Chief Executive Officer; and Juho Parkkinen, Chief Financial Officer. After the market closed today, we issued a press release with details regarding our second quarter results as well as a supplemental to our results, both of which can be accessed through the Investor Relations section of our website at ir.c3.ai. This call is being webcast, and a replay will be available on our IR website following the conclusion of the call. During today’s call, we will make statements related to our business that may be considered forward-looking under federal securities laws. These statements reflect our views only as of today and should not be considered representative of our views as of any subsequent date. We disclaim any obligation to update any forward-looking statements or outlook. These statements are subject to a variety of risks and uncertainties that could cause actual results to differ materially from expectations. For a further discussion of the material risks and other important factors that could affect our actual results, please refer to our filings with the SEC. All figures will be discussed on a non-GAAP basis unless otherwise noted. Also during the course of today’s call, we will refer to certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in our press release. Finally, at times in our prepared remarks, in response to your questions, we may discuss metrics that are incremental to our usual presentation to give greater insight into the dynamics of our business or our quarterly results. Please be advised that we may or may not continue to provide this additional detail in the future. And with that, let me turn the call over to Tom.

Thank you, Amit. Good afternoon, everyone, and thank you for joining our call today. Results: Bottom line, we continue to accelerate our revenue growth and our customer engagement count and continue to gain traction with C3 Generative AI and our Enterprise AI applications in the second quarter. Total revenue for the second quarter was $73.2 million, an increase of 17% compared to $62.4 million one year ago, and accelerating from an 11% increase in the first quarter. The total number of customer engagements was 404, an increase of 81% compared to 223 last quarter. North American revenue of $61.2 million increased 28% year-over-year, while EMEA revenue of $10.6 million decreased 11% year-over-year, and federal revenue increased 100% year-over-year. Subscription revenue for the quarter was $66.4 million, constituting 91% of total revenue and increasing 12% from a year ago. GAAP gross profit for the quarter was $41.1 million, representing a 56% gross margin. Our non-GAAP gross profit for the quarter was $50.4 million, representing a 69% non-GAAP gross margin. Our GAAP net loss per share was -- the loss was $0.59, and non-GAAP net loss per share was $0.13. We ended the quarter with $762.3 million in cash, cash equivalents and investments. C3 -- AI’s partner ecosystem continues to drive significant growth. In Q2, the company -- the Company closed 40 agreements through our partner network, including AWS, Booz Allen, Baker Hughes, Google Cloud and Microsoft. The qualified opportunity pipeline with partners has increased by 75% in the past year. We’ve signed new and expanded agreements with Nucor Corporation, Roche, Con Edison, Hewlett Packard Enterprise, GSK, formerly SmithKline, the United States Navy, the Administration for Children & Families, a division of Health & Human Services, Indorama and First Bank amongst others. Over the past several months, C3 AI has helped Nucor, the largest steel producer in the United States to better optimize caster production schedules specifically to improve production levels and reduced cost levels in the steel casting process. C3 AI is now helping Nucor scale this across several additional mills. In Q2, C3 AI also kicked off two new additional use cases at Nucor, tackling process optimization and demand forecasting, and we also completed a C3 Generative AI pilot, targeting operational health and safety. GSK, formally GlaxoSmithKline, is now using C3 AI supply chain suite to increase efficiency in its supply chain, using AI to optimize yield and improve demand forecasting processes. Con Edison, a C3 customer since 2017, continues to expand its use of the C3 AI applications, most recently, by adding C3 Generative AI. Con Ed is using C3 Generative AI to help workers quickly find answers to questions and analyses related to smart meters, service levels and infrastructure data. In the second quarter, Con Edison completed two pilots of C3 Generative AI, which have now converted to production. We also continue to expand our footprint in state and local governments with particular interest in C3 AI law enforcement from San Mateo County, California. And C3 AI Residential Property Appraisal from Stark County, Ohio and Charlotte County, Florida. Our federal business continues to show significant strength with bookings up 187% year-over-year. We closed new and expanded deals with the United States Navy, the Intelligence Community, Joint Staff J8, the Defense Logistics Agency, and the Administration for Children & Families. We’ve talked many times about our successes in helping them monetize -- or to modernize, sorry, the Department of Defense, and we’re proud now to say that our products are helping civilian government agencies as well. This quarter, we began work with the Administration for Children & Families, a division of the U.S. Department of Health & Human Services. The agreement with C3 AI was part of their first order under a $90 million blanket purchase agreement. And this part of ACF’s work involves helping unaccompanied children who cross the U.S. border, find temporary shelter and permanent homes. Our platform will be used in complex modeling and predictive analytics at ACF to help them keep track of the number of unaccompanied children in the ACF’s care, staffing needs and determine how long these children are with their case managers amongst other tasks. Our C3 AI continues to leverage its extensive commercial supply chain experience in the federal government and is now applying this experience in the defense sector with the C3 AI contested logistics application for Transcom and for DLA. During the quarter, C3 AI converted two defense logistics agency pilots in the follow-on projects for the Department of Defense. The first project delivers common operating picture of the supply chain for DoD and enables leaders at multiple echelons to see in near real time, their global Class 9 supply posture. The application unifies disparate supply data and provides the Defense Logistics Agency, the ability to identify supply chain inefficiencies, forecast parts consumption and parts shortage and conduct impact assessments and put into place mitigation plans. The second project supports DLA’s energy directorate, leveraging C3 AI’s commercial expertise in the oil and gas sector. The C3 AI’s contested logistics application modernizes and streamlines global fuel distribution for the Department of Defense. Users can see global fuel inventories, anticipate fuel consumption, identify supply network risks and create distribution and transportation plans to prevent disruption and assure supply. These applications promise to significantly impact the efficiency of the Department of Defense Logistic enterprise and improve readiness. Our partnership with AWS deepened with an expanded strategic collaboration agreement in the quarter. And the availability of our new no-code self-service generative AI application, C3 Generative AI now available on the AWS marketplace. I think we announced that last week. This new application allows customers -- users of all technical levels to begin using generative AI within minutes of signing up. And this application, C3 Generative AI, is now available to you on the AWS Marketplace under our 14-day free trial. And so, I encourage you to take a look at it, for those of you who are interested. Under the expanded collaboration agreement with AWS, we’re focusing on offering advanced generative AI solutions, combined with what they’re doing in Bedrock, okay, and other initiatives for enterprises and for AI applications for customers in multiple verticals, including manufacturing, power and utilities, consumer packaged goods, state and local government and the federal government. C3 AI and AWS’s joint qualified pipeline has more than doubled year-over-year with heightened interest in the C3 Generative AI suite. In Q2, C3 AI has been recognized multiple times for its innovation in the AI space. We’ve been named to the Fortune 50 AI Innovators list, and the list goes on and on. So I’m not going to belabor that. We get recognized all the time. Pilot Growth: This is important. In Q2, we closed 62 agreements, including 36 pilots and trials. Our new pilot count is up 270% from a year ago. Notably, 20 of these were generative AI pilots, 150% increase from Q1. With the lower entry price points of our pilots, we are more easily able to land new accounts. With our pilots, we are engaging customers across a diverse set of industries in this quarter. Our pilots came from manufacturing, federal, defense, aerospace, pharmaceuticals and other industries. Now, we did see sales headwinds in the quarter. While the interest in AI applications and especially generative AI is growing substantially, we’re also seeing, in many cases, lengthening decision cycles. Virtually every company in the last 3 to 6 months has created a new AI governance function as part of its decision-making process. These AI governance functions assess and approve those AI applications that will be allowed to be installed in the enterprise. This has candidly added a step to the decision process in AI. You might have heard it here first, but you will be hearing this from every AI vendor in the next few quarters. Take it to the bank. It has simply -- it has added a step to the process and it is lengthening the normal sales cycle. So this had -- this provided a sales headwind in the quarter. And while the increased scrutiny lengthens the sales process, we believe that this is a healthy process to ensure that companies are adopting safe and appropriate AI solutions. So, we’re all for it. And did it move revenue, a little bit, a click below the center of the range? Yes, it did, okay? But get over it. The world is a better place, people are making very careful well-informed decisions. They have their best people on it, and we will all be happier for this in the long run. So, it did -- that dynamic did provide an unexpected headwind to our Q2 sales revenue performance. In addition, our sales execution in Europe was candidly unacceptable. And since then, we’ve taken -- we’ve been through our planning meetings and we’ve taken appropriate organizational steps to immediately improve sales execution in Europe. Now, let’s take a look at -- this is the big story, this is the top line. And really what this whole story has been about for the last 6 to 7 quarters has been from the transition from subscription-based pricing to consumption-based pricing. And before we switch to consumption-based pricing, you’ll recall the company was growing at a quite a rapid growth rate, like I think, seven quarters ago, order of 38% year-over-year growth rate. So, we were definitely in the top quartile, okay? And we announced the transition to consumption-based pricing that we believe would be and has become the standard in the industry. The consumption-based pricing is based upon per virtual CPU or virtual GPU hour, similar to the pricing at Snowflake, Google Cloud, AWS, Microsoft Azure et cetera. Prior to this, we were doing large enterprise subscription deals of $1 million, $5 million, $20 million, $50 million. And it was a good business. That being said, the downside of that model was lumpiness in bookings, lengthy sales cycles and low levels of revenue predictability. We believe the transition from a primarily subscription-based pricing model to an assumption-based price remodel brought us into line with what we believe are today, the industry standard cloud pricing standards, making it easier and less costly for new customers to acquire our solutions and then increase their spending as their usage and adoption increase. We anticipated and announced when we made that transition that it would have a short to medium negative effect on revenue growth, a long-term drag on RPO as the sales price was significantly reduced and the contracts often lacked a time-certain multi-period commitment. We believed when we made the announcement that the consumption-based pricing model would increase the number of customers and increase the total amount of system consumption, resulting in a return to increased revenue growth, increased customer growth, decreased average selling price and decreased RPO over time. Now while we are still in the process of working through -- completely through this transition to the new pricing model, the preliminary empirical results that we are seeing evidenced by year-over-year growth rates appear to be proving out exactly as expected and exactly as we predicted. Since the transition, revenue growth initially decreased, then it flattened, and now it is increasing as the consumption-based pricing model takes effect. Average selling prices decreased, RPO has decreased, customer engagement has increased substantially. If we look back over the last, say, 1, 2, 3, 4 quarters, 4 quarters ago, our revenue growth was negative 4% and then 0%. The last quarter, it was 11%. Now it’s 17%. Bookings growth, 71% year-over-year. I’m sorry, bookings growth 100% year-over-year. New contract growth 148% year-over-year. Pilot growth, 50% quarter-over-quarter, 170% year-over-year. So, this is basically the beginning, the middle and the end of this story. We announced 6, 7 quarters ago a transition to our consumption-based pricing. We predicted that revenue would decline and then flatten and then increase, and we are now seeing these increases that we predicted. So now, let’s talk about generative AI. Generative AI simply changes everything. I believe that it more than doubles the size of our addressable market overnight. We’ve all seen the predictions from Bloomberg that predicts this is 100 -- this is in excess of $1 trillion -- $1.3 trillion market by 2032. Goldman Sachs predicts that this could increase corporate profits by 30% in the next decade. And the generative AI alone could raise the global GDP by 7%. People, this is a big deal. It is difficult to overestimate the levels of interest that we’re seeing in the category of generative AI. Now, by combining our multibillion dollar, say, 14-year investment in the C3 AI platform, with the recent developments in large language models and retrieval augmented generation, C3 AI is unique in the market and that we are able to solve the disqualifying hobgoblins that are preventing the adoption of generative AI in government, in defense, intelligence, in the private sector. What are those hobgoblins? Those are the facts that -- the answers that come out of these large language models are sarcastic. They’re random. They’re not traceable. We have this hallucination problem, which is extraordinarily problematic. We have research -- none of our data access controls it, be it DoD or Bank of America are enforced. We have these problems with LLM cause data exfiltration, LLM cause cyber threats and IP liability. In addition, all the solutions that are out there, almost all those solutions, I would say, with the exception of AWS Bedrock, tend to be LLM specific. And I don’t think anybody wants to be LLM -- hook their wagon on to any given LLM today with all the innovation that’s going on in the market and to be dependent on any LLM provider that could make some announcement on Friday and be gone on Monday, so you’ll put AI for detail. So this LLM agnostic -- the bottom line is our solution in the market addresses every one of those hobgoblins that prevent the installation of generative AI in the enterprise. And so, this is really unique and it took 14 years and $2 billion of software engineering for us to be ready for this. This is why we could solve it. So while the rest of the world is playing catch-up, we’re -- how about multimodal. I mean, we completely nail multimodal. We’ve been doing it for 14 years. Multimodal, what does this mean? Rather than all these LLM solutions basically handle text, we handle text, we handle telemetry, we handle images, we handle signals, we handle enterprise data, we have the structure data, we have unstructured data. We are unique in the market, and the result is quite exciting. So while the rest of the world is playing catch-up and we have scores of start-ups with three guys, four girls and two cats in apartment in San Francisco, being -- they’re getting $1 billion funding and multibillion dollar market valuation, see PitchBook for details. We have -- I don’t know how many customers, we have order of 1,000 employees. I don’t know how many countries, and we’re delivering these solutions today. And so, while the rest of the world is playing catch-up, we’re working closely with our customers and new customers to install high value generative AI solutions that rapidly realize value to their organizations. We believe that our strategic decision to invest in generative AI could address our addressable market opportunity. Our suite of 28, now I think 29 generative AI products wins on reliability, flexibility, adaptability, accuracy and security, all of the same qualities that are inherent in our Enterprise AI platform. Our vision to expand our customer base is working. The idea, and this is very much idea about the work that we’re doing on the AWS Marketplace is to go from 8 customers to 80 customers to 8,000 customers, to 80,000 customers. So what we’re dealing with now is kind of a new game with massive market leverage, and we are the first to market. And so I think we have the opportunity here through our innovation, through our applications that will proliferate across the business. C3 Generative AI has enabled us to land high-caliber new customers and expand agreements with the current customers. The surge of interest led to our C3 Generative AI qualified pipeline increasing new opportunities, increasing 55% sequentially quarter-over-quarter in the second quarter, representing the most rapid acceleration of all our product offerings. We expect this momentum to grow as we continue to innovate and build the increasingly exciting products. Our November announcement of the self-service C3 Generative AI on the AWS Marketplace plays a big part in this story, potentially expanding our addressable customer pool and our user base exponentially. This new application allows users of all technical levels to enroll in the application and begin productively using generative AI in minutes. Again, this product is available today on the AWS Marketplace, should you have interest. As I introduced last quarter, we made a well-considered decision to seize the immediate and candidly staggering market opportunity that we see in generative AI. As such, we are making and increasing, a sizable and timely investment in application development, model engineering, lead generation, branding and market awareness to seize market share in generative AI as rapidly as possible. This will put short-term downward pressure on free cash flow and profitability. Closing thoughts: The generative AI opportunity is staggering. We believe that it is in the best interest of our shareholders, to further accelerate our investment in generative AI, deepening our investments in lead generation, branding market awareness and customer success. Given our substantial cash balance, we believe it is a strategic imperative to invest further in the generative AI opportunity at this time. Separately, now with the release of our platform version of our 8.3 product line, which is really quite remarkable in terms of the benefits that it brings to our customers and the increase in performance that it brings to our customers, we have decided to further invest in our customer base to accelerate their upgrade from version 7 to version 8.3, which we believe will further increase our customer satisfaction levels that are already quite high. That being said, we continue to expect positive cash flow in Q4. And while we’re not giving fiscal year ‘25 guidance yet, we continue to expect positive cash flow for full year fiscal year ‘25. C3 AI remains focused. We are one of the few AI software pure plays that has established relationships, a tried test and proven technology platform, add reputational equity to capitalize on this generative AI market opportunity. Now I’ll turn the call over to Juho Parkkinen , our Chief Financial Officer, to talk more about our financial performance and provide guidance for the remainder of the fiscal year. Juho?

Thank you, Tom. I will now provide a recap of our Q2 financial results and some additional color on our consumption-based revenue model, which we introduced five quarters ago. Then, I’ll discuss factors that would drive our financials in the back half of the year. All figures are non-GAAP unless otherwise noted. Total revenue for the second quarter increased 17.3% year-over-year to $73.2 million. Subscription revenue increased to 11.7% year-over-year to $66.4 million and represented 90.7% of total revenue. Professional services revenue was $6.8 million and represented 9.3% of total revenue. Gross profit for the second quarter was $50.4 million, and gross margin was 68.8%. As a reminder, we continue to expect short-term pressure on our gross margins due to a higher mix of pilots, which carry a greater cost of revenue during the pilot phase of the customer life cycle. Operating loss for the quarter was negative $25 million compared to our guidance range of negative $27 million to negative $40 million. The improvement in operating loss versus guidance was driven by timing and amounts of the generative AI-related investments we made to capture market share as well as our team’s ongoing focus on disciplined expense management. At the end of Q2, our accounts receivable was $143.2 million, including unbilled receivables of $104.8 million The general health of our accounts receivable remains strong. Now turning to RPO and bookings. Reflecting our transition to consumption-based contracts, we reported second quarter GAAP RPO of $303.6 million which is down 27.3% from last year and current GAAP RPO of $170.2 million, which is up 3.5% from last year. We continue to see positive trends in the diversity of our pilot bookings with 10 industry segments represented in Q2 pilots as compared to 8 in Q1. Free cash flow for the quarter was negative $55.1 million. We continue to be very well capitalized and closed the quarter with $762.3 million in cash, cash equivalents and marketable securities. Now I’ll provide an update on our consumption business model for the second quarter. During the quarter, we started 36 pilots, a 50% increase from last quarter. We are pleased to report that the actual vCPU consumption data that we’re seeing from pilot activity has validated the assumptions we made when we transitioned to the consumption-based pricing model five quarters ago. Our pilot conversion rates are trending upwards or are getting close to our target of 70%. At quarter-end, we had cumulatively signed 109 pilots, of which 103 are still active. This means they are still in their original 3- to 6-month term, extended for 1 to 2 months, converted to consumption or a license contract or are currently being negotiated for a production license. Finally, our customer engagement count for the quarter was 404, an 81% increase from 223 a year ago. Turning to guidance. As Tom mentioned, we expect Q3 revenue to range from $74 million to $78 million, and non-GAAP loss from operations to range from negative $40 million to negative $46 million. We remain committed to delivering positive cash flow in Q4 FY24 and for the full year of fiscal year ‘25 and non-GAAP profitability in the second half of fiscal ‘25. For the full fiscal year ‘24, we are maintaining our previous revenue guidance in the range of $295 million to $320 million. We are increasing our non-GAAP loss from operations guidance to a range of negative $115 million to negative $135 million. I’d like to turn the call over to the operator to begin the Q&A session. Operator?

Our first question comes from the line of Timothy Horan of Oppenheimer.

Thanks a lot, guys. Really appreciate the time. Can you give us a sense of what you’re seeing with AI in terms of productivity improvements? And what is the major bottleneck that you think customers need to overcome to really start implementing services? Thanks.

I’m sorry. The question related to gen AI?

Yes, specifically on gen AI. What type of productivity improvements do you think customers can see on specific applications? And what is the major bottleneck for them adopting G AI?

The major bottleneck as it relates to generative AI relates to the problems that are inherent in these large language models, and they’re very real. I mean, as you know, if you ChatGPT or Google Bard, both of which are like excellent products, but the answers tend to be sarcastic. So every time you ask a question, get an answer, the -- if it doesn’t know the answer, it hallucinates. The data access controls are not in force. So the CEO and the person on the factory floor get access to the same information. We’re -- Carnegie Mellon and others are now identifying huge cybersecurity risks that are associated with these large language models to corporations and government entities. We have IP liability problems that people are concerned about because these large language models are trained on and have access to all the data into the internet. This is weather, stock prices, what have you. Those -- somebody has the copyright to all those data, be it the weather company or Bloomberg, and they want to get money. So, the quintamenials [ph] of the world are going to build big businesses litigating these issues in the next 10 years. We have -- so there’s very real issues. The other issue relates to almost all the solutions that are being offered are LLM specific and in, say, December of 2023, to hook your wagon on any specific LLM is kind of crazy because next week, some is going to leave rocket by a factor of 10. You need to be able to switch, you need to be LLM agnostic. So I think those are really the hobgoblins, cybersecurity, hallucination, information security that are basically making -- so many organizations will not allow any generative AI application to be installed. What’s unique about the C3 AI solution is -- we can talk about this some other time or you can look it up on the internet. But by combining with the 14 years of work that we did with the C3 AI platform, we’ve addressed all those problems, cybersecurity, data security, hallucination, what have you. So I think that’s the hobgoblin. That’s what slows things down. And people need to be need to be satisfied to those issues resolved and if they’re not resolved, you’re not being installed at any reasonable organization like General Motors or JPMorgan Chase or you name it. Now as it relates to productivity increases, holy moly, they’re going to be staggering, whether you’re a lawyer, whether you’re a realtor, whether you’re a physician or whether you’re running a paper machine or whether you’re operating the infantry or the space command. I mean you -- if you do not have or not being supercharged by generative AI, your competition will be, okay? And if they are, and you’re not, they win, you lose hard stock.

So specific to your customers, what do you think the bottleneck is for adoption? If you have -- it sounds like you have all these problems pretty much resolved for them. What do you think they require at this point to really start adopting?

Well, we just -- our sales cycles are pretty fast. Our sales cycles for generative AI has been close as 24 hours. And basically, our offering is, okay, we’ll bring the application live in 1 or 2 months, if you like it for, I don’t know, $0.25 million or something. And if you like it, keep it. So this has to do with people evaluating bond portfolios, people running paper machines, people running steel mills, the intelligence community, missile defense agency, others. So, we just -- many of them are existing C3 customers, although increasingly, we will be serving -- 9 out of 10 will not be existing C3 customers. But we have to address their concerns that identified. We seem to be able to address those. After that, we just bring the application live, we get it live in 4 to 8 weeks and if they like it, keep it. And so it’s a pretty short sales cycle for us, and you’re seeing a very substantial increase in the pilots that we’re deploying. You can expect -- we’re expecting a pilot to production conversion rate of -- it looks like about 70%. And so it does look like a big opportunity.

[Operator Instructions] Our next question comes from the line of Mike Cikos of Needham.

I wanted to ask first about the subscription gross margins, and this probably goes back to Juho’s prepared remarks. But it was good to see gross margins actually increased sequentially despite the increased pilot count. And I know that you guys are calling out the short-term pressure just based on the growing mix of pilots. And so, can you help us think about like what was it that actually went better for you guys? Because I think we were expecting a little bit more degradation in the subscription gross margins versus how you guys -- how the quarter actually came through?

Thanks, Mike. This is Juho. So yes, in the big picture, as we announced five quarters ago, as we’re seeing, we are expecting the gross margin degradation for the subscription to continue. Now, in the -- in this particular quarter, we were very pleased to see some improvement on a sequential basis, but I think we would expect flattening to down again on the next quarter as the pilot count increases and is going to put pressure before the consumption amount start picking up and offsetting that.

Got it. Thank you. And if I just shift down to OpEx for a second as well, I guess two quarters here. So first, I know that you guys are increasing the anticipated operating losses here. Last quarter, we had cited increased investment in like branding and lead gen and awareness, right? So can you help us think through where you guys are doubling down? And then, the second piece there, there was obviously that article that came out in Bloomberg, I think it was in mid- to late-November citing headcount costs -- headcount cutting, I’m sorry. So can you just comment on the validity of the Bloomberg article? Just because I think people are trying to see if you did make those headcount cuts, how much are we doubling down on these investments, or if that article proved to be false?

Hi Mike, it’s Tom. Doubling down, we’re doubling down on data scientists, we’re doubling down on large language model engineers, we’re doubling down -- a lot of it is going into engineering, but also candidly in lead generation. I mean, there’s an opportunity now as we move to these marketplaces to be dealing transactions in hundreds to thousands to tens of thousands of units rather than scores. And that I can assure you is the plan that we have. As it relates to -- I’m not familiar with Bloomberg article that you talked about. It sounds like somebody mentioned something that we did some layoffs in the quarter. Mike, we do performance-related layoffs every quarter, okay? And the -- so we -- I think last quarter, we had 42,000 job applicants. We -- how many people did we hire, Juho? Order of 100. And these people, yes, they went to MAT. Yes, they worked at Bank of America. Yes, they went to Chicago TSB and they command an F1-8 squadron. And so, we’re constantly upgrading our human capital, and we move underperformers out regularly. So if somebody said that in a Bloomberg article, I don’t know what they said. What I told you is the truth.

Our next question comes from the line of Kingsley Crane of Canaccord Genuity.

I wanted to touch on the pilot program. You mentioned that you’d move to a lower entry price point for pilots. Could you give us a sense of the magnitude of that change? And then has the minimum fee post pilot also changed? I’m curious what kind of upsell you’re seeing upon conversion, if any?

I think the standard pilot that we have at generative AI and the enterprise is like $250,000. But that being said, you can get the AWS -- generative AI for AWS, which basically handles documents like every other LLM, handles text, it’s not really multimodal, but that’s free for 14 days. So, that would be pretty available. Is there a question that you asked that I didn’t answer?

Okay. Yes. Thank you. That’s helpful. And I just want to touch on OpEx as well. So, I think it makes sense that you want to invest more in both, LLM engineers and lead gen. And it looks like that’s particularly hitting harder in Q4 of this year. But as we think about fiscal ‘25, it seems like some of the nature of those investments would naturally continue as you scale in the some large opportunities. So, is it about timing in this year, or are you expecting those to continue next year?

Kingsley, I expect them to continue next year. But if you look at the guidance that we gave you in terms -- about six quarters ago, what we see is the consumption over the first 12 quarters in terms of CPU seconds per new customer. We just did an analysis of, Juho, I don’t know, about 30 customers -- or 12 customers. And those data that we predicted, I think 6 or 7 quarters ago and provided you, it’s uncanny in how accurate it is. It’s basically plus or minus 10%. And so if you look, as these things kick in, in quarter 5, 6, 7 and 8, the consumption numbers get pretty big. So you can expect that -- we don’t really need to cut back on the investments to get to the point of cash positive and non-GAAP profitable. So, the top line kind of takes care of that.

Our next question comes from the line of Sanjit Singh of Morgan Stanley.

Great. This is Steve on for Sanjit. Tom, maybe starting with you. I mean, with a couple of quarters of the consumption model now under your belt, clearly, you’re seeing a lot of sort of quantity of deals and pilots. Is there any way that you can frame or give us a sense of the quality of those customers that went with the consumption model early on? I guess, any sort of scale in terms of spending or growth profile that they’re hitting now that you can kind of shed up some light and give us the quality piece where you’ve given us, I think, a lot on kind of the quantity piece of those yields? And then for Juho maybe, could you just give us some color on the subscription revenue versus the services revenue this quarter, and then also maybe the partner impact and sort of what that looks like on a go-forward basis? Thank you.

Regarding quality, I think there’s only two ways to look at pilot quality. It’s going to be what’s the conversion rate and what’s -- and what are they going to consume. Based upon our best guess at this time, based upon looking at every pilot we have out there, going to look at what actually has converted and what we think we will convert, we think our guesstimate that we gave you 6 or 7 quarters ago, 70% is about right. So, there’s one indication of quality. The other indication of quality is how many CPU seconds are they consuming over -- as you go from quarter 0 to quarter 12. And it’s tracking right in line. I mean, it varies a little bit from one quarter to another, but it’s basically right in line with what we told you. The quality is pretty high. Now that being said, as we move now to mass markets and start dealing with hundreds of thousands of people just either kind of ordering this online and playing with it, you can expect that conversion rate from that level of pilot to be, I would say -- I mean, the quality there will be much lower. And I think we need to measure quality by conversion rate and consumption levels. A lot of those people will try it for 5 minutes and drop off. And that’s just the way that it is with free stuff. Now the rest of the question, I think, goes to you.

Yes, right. So your second part about subscription versus services. So we were 9.3% professional services this period, which is a little bit lighter than our expected long-term model of 10% to 20% on professional services. We continue to expect that we will be at that range on a go-forward basis. And then, I think you were asking about how we feel about the partners in a go-forward basis. And partners are hugely important for us. And we continue to believe that they’re the key part of our go-to-market approach going forward.

It looks like we have time for one last question. Our last question will be from Pat Walravens of JMP Securities.

This is Owen Hobbs, on for Pat. I guess first one for Tom. What would you say are the top one or two federal use cases for generative AI that you’re seeing with those new -- those five new federal degenerative AI deals this quarter?

Our largest federal use case, as you know, is predictive maintenance of the United States Air Force. This was chosen by the Chief of Staff. And we now are doing -- this is the PANDA system, which is the only AI system of record that we’re aware of in all of DoD. So, this is a system record for the Air Force for predictive maintenance for all assets. So far, we have loaded the data, I believe, from 22 weapon systems. F-15, F-16, F-18, F-35, KC-135, F-22, et cetera, into unified federated image. This is 100 terabytes of data. Some of it is maintenance data, sorted data, inventory data, flight data, flight history, telemetry and one aircraft like B-50 -- each B-1 bomber has 42,000 sensors on it, admitting telemetry, and I’m not sure what hurt cycles but pretty fast. So, this is a stack of data. I will be there on Monday -- that is by next Monday. I will be in Washington, D.C., showing this to our customers with a generative AI front end. So, think about this as a Mosaic browser front end where a general officer can ask any question about -- this a 100-terabyte production system. This is one of the largest production enterprise AI applications in existence. Okay? And that person will be able to ask on Monday, be able to ask any question that you could ask of the weapon system, for example, where aircraft are operative at Travis Air Force Base now? What is my cost of operating the B-1 bomber program in the last year? What is the -- as it relates to F-35, where are my largest part shortages? And rather than going through some cold war era, menu-based SAP or even -- I don’t want to take shots at SAP enterprise information system user interface that looks like your Bloomberg terminal, which is -- I have one I guess, it’s unusable. It’ll just be a Mosaic browser, you could ask any questions and get the answer related to any one of these weapon systems in the United States Air Force, I guess, their production data. And we will show it on Monday, on Tuesday, on Wednesday. And I am telling you we expect some light bulbs to flash.

And if I could sneak one last one in for you Juho. Can you please explain the dynamics between the increase in accounts receivable from last quarter to this quarter, despite revenues kind of staying flattish?

Well, accounts receivable is timing of invoicing. So obviously, when we drop an invoice, it shows up in the accounts receivable. So it’s just timing of invoicing.

Ladies and gentlemen, I think we’re at end of program. We appreciate your time and your attention. And thank you very much. And we look forward to talking with you next quarter. Standby, it does appear to be a game-on in the AI industry at global scale, and I can assure you we are very much in the game. So, thank you all. And we’re signing off.

Thank you. Ladies and gentlemen, this does conclude today’s conference. Thank you all for participating, and have a good night. You may now disconnect.