管理層發言
Good afternoon, and welcome to C3.ai's earnings call for the first quarter of fiscal year 2025, which ended on July 31, 2024. My name is Amit Berry, and I lead Investor Relations at C3.ai. Joining me today are Tom Siebel, Chairman and Chief Executive Officer; Ed Abbo, Executive Vice President and Chief Technology Officer; and Hitesh Lath, Chief Financial Officer. After the market closed today, we issued a press release outlining our first quarter results, as well as additional information, both of which can be found in 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 after the call concludes. During today's call, we will share statements about our business that may be seen as forward-looking according to federal securities laws. These statements reflect our views only as of today and shouldn't be seen as indicative of our views on any future date.
We are not obligated to update any forward-looking statements or outlook. These statements carry a range of risks and uncertainties that could result in actual outcomes differing significantly from expectations. For a detailed discussion of the material risks and other factors that could impact our actual results, please consult our filings with the SEC. All figures discussed will be on a non-GAAP basis unless specified otherwise. Additionally, we will reference certain non-GAAP financial measures during today's call. A reconciliation between GAAP and non-GAAP measures is included in our press release. Lastly, in our prepared remarks or in response to your questions, we may cover metrics beyond our usual presentation to provide greater insight into our business dynamics or quarterly outcomes. Please note that we may or may not continue to provide this additional detail in the future. Now, I will turn the call over to Tom.
Thank you, Amit. Good afternoon, everyone, and thank you for joining our call today. We are off to a solid start for fiscal year '25. In Q1, we exceeded all expectations for revenue, cash flow and profitability. This quarter marked our sixth consecutive quarter of accelerating revenue growth, reflecting our high levels of customer satisfaction and increasing demand for enterprise AI applications. Our year-over-year revenue growth has accelerated from 11% in Q1 '24 to 17% in Q2, 18% in Q3, 20% in Q4, and 21% in Q1 year-over-year revenue growth in Q1 of fiscal year '25. Total revenue for the quarter was $87.2 million, exceeding analyst expectations. Subscription revenue was $73.5 million and increased 20% from a year ago. Our non-GAAP gross profit was $60.9 million representing a 70% gross margin. Our GAAP operating loss was $72.6 million. Our non-GAAP operating loss was $16.6 million and substantially better than our guidance for a loss of $22 million to $30 million.
Our non-GAAP net loss per share was $0.05. Our net cash provided by operating activities was $8 million, and we generated free cash flow of $7.1 million in the quarter, both substantially exceeding market expectations. We ended the quarter with over $760 million in cash, cash equivalents, and investments. I'll note that this is the 15th consecutive quarter as a public company in which we have met or exceeded our revenue guidance. In the first quarter, the company closed 71 agreements, including 72 new pilots marking a 117% year-over-year increase in our pilot count. We entered into new agreements with GSK, Eletronbras, Valero, Swift, SmithRx, Sanofi, the U.S. Intelligence Community, the U.S. Department of Defense, Dolce & Gabbana, Ingersoll Rand and others. Additionally, we significantly expanded our footprint across state and local government. In Q1, the company signed 25 agreements with state and local governments with municipal, county and state agencies in Texas, California, New Jersey, Georgia, Washington and Connecticut, Virginia, Rhode Island, Maine, New Mexico and Florida.
State and local government is a large and underserved market that we're rapidly penetrating. Our solutions increase efficiency while maintaining the highest standards for accuracy, transparency and security. They drive substantial cost savings and the combined benefits result in improved public services and improved customer experience. In short, C3.ai is enabling government agencies to do more with less, ultimately benefiting the public. From the Assessor of Riverside County, and I quote, we set the bar high for C3 AI and they deliver with over 90% model accuracy in our property valuations. This technology is letting our staff do the mundane tasks faster and easier so they can concentrate on the very complicated properties. And our employees are starting to see the fruits of this effort translating into better customer service. Overall, we are seeing incredible results with C3 AI. The proven benefits of C3 AI local government suite, combined with our partner-supported and concentrated sales strategy, resulted in accelerated sales cycles within the sector.
The growth was fueled by a highly collaborative joint sales and marketing campaign with C3 AI and Google Cloud to promote the C3 AI state and local government suite, including C3 AI property appraisal and C3 Generative AI for public benefits. Through these efforts, we saw high adoption with state and local government closing 24 agreements in the quarter. Let me give you a feel for the speed of these sales cycles. In Q1, we closed an agreement with a county in the Northwestern United States. County decision makers got introduced to C3 AI at an industry conference and within 24 hours, we were in contract negotiations. Just 12 hours later, we were in contract for C3 AI property appraisal. Another example is with a county in the Southeast U.S. This customer attended our Annual Users Group Conference, C3 Transform in March of 2024. Then four weeks later, we held an executive briefing for the county leadership.
And now after three months of contract discussion, the county has signed a seven-year, seven-figure, five-year subscription deal for C3 AI property appraisal. Our state and local business has grown more than 500% year-over-year, and we are excited about the traction, the potential for expansion and the customer advocacy for C3 AI. As we now take these public sector solutions to market in Europe, the addressable market more than doubles. Turning to the C3 AI Federal business. This sector continues to experience sustained momentum, representing over 30% of our bookings for the quarter. We entered into new and expansion agreements with the United States Air Force, the U.S. Navy, U.S. Marine Corps, and the U.S. Intelligence Community among others. These agencies trust C3 AI to provide secure and innovative applications that help them modernize. The U.S. Marine Corps and C3 AI continued the successful collaboration to digitally transform the branch's legacy software systems.
The Marine Corps is using the C3 AI Defense and Intelligence suite to improve the efficiency of personnel management systems by accelerating critical processes and time-to-decision support. This work with C3 AI is backed by Manpower IT Systems Modernization program and aligns with the Marine Corps' multiyear roadmap and goals. C3 AI's customer base continues to expand, both within and across industries, while maintaining exceptional levels of customer satisfaction by our continued focus on delivering measurable, significant enterprise value. At Eletronbras, the largest power generation transmission company in Latin America, we partnered to enhance their grid resiliency and availability. Brazil's grid is among the world's most complex due to its diverse generation profile, variability, expansive service territory, and challenging regulatory environment. With C3 AI, Eletronbras can effectively and efficiently process and analyze real-time data servicing low latency AI insights to mitigate network disturbances.
Nucor Corporation is seeing significant success in improving manufacturing outcomes with the C3 AI Supply Chain Suite. This deployment includes three distinct C3 AI applications: C3 AI Demand Forecasting, C3 AI Inventory Optimization, and C3 AI Production Schedule Optimization, which work together to support and optimize daily decision making across multiple facilities. As we enter Q2, we are focused on expanding sales capacity, expanding in North America, expanding in Europe, and expanding in the public sector. Our partner network continues to generate opportunities and open new deal flow. We had a very active first quarter in alliances, working closely with our partners to close 51 new agreements. Net-net 72% of our total agreements were closed with and through our partner ecosystem. This was an increase of 155% year-over-year and 82% quarter-over-quarter. Our partner-supported bookings grew 94% year-over-year, and our activity levels with our partners that include AWS, Booz Allen, Google Cloud, and Microsoft continue to increase substantially.
In the first quarter, we closed 40 agreements with Google Cloud. This is an increase of 300% year-over-year. This growth was largely driven by the joint campaign between C3 AI and Google Cloud that I just discussed earlier, focused on the public sector. C3 AI continues to be an attractive partner for the hyperscalers as our 90 enterprise AI applications rapidly drive substantial workloads in their compute and storage clouds while adding immediate value to our joint customers. As you can see from our supplemental deck, our bookings continue to be increasingly diverse. Our generative AI business is surprisingly diverse with many candidly unanticipated use cases across the board in a wide range of industries. In addition, our total non-Baker Hughes revenue grew 37% year-over-year in Q1.
Thank you, Tom. Let me first recap where C3 AI fits in the AI tech stack. Starting at the bottom, the four layers include silicon, cloud infrastructure, foundation models. And at the top, harnessing all this innovation to deliver business value are AI applications. This is where C3 AI plays with pre-built AI applications that can be deployed very quickly. In contrast, the offerings you're seeing in the AI software market today fall into two distinct categories. First, legacy software companies that are scrambling to keep up with AI. To try to stay relevant, these companies are rebranding their 20th-century software stacks with AI on the box. The technical depth to rewrite their software to take advantage of a modern, scalable AI tech stack is simply insurmountable. For them, AI is just a bolt-on constrained to a small fraction of the enterprise data managed by their legacy software. Second, we see a few more modern software companies that were designed primarily for data integration, data management, or data engineering lacking the capabilities to build and operate AI applications.
Basing an enterprise AI application plan on these incomplete tech stacks requires customers to undertake significant, protracted software development projects where minimal customer value is realized. This is clearly reflected in the customer satisfaction scores of these vendors. The C3 AI platform is unique in that it was a clean sheet design, providing all the services necessary and sufficient to design, develop, provision, and operate real-world enterprise AI solutions. This provides users with rich workflow-enabled AI applications operating on a platform with advanced data fusion, governance, and scalable AI/ML operations capabilities. We leverage all layers of the AI tech stack, silicon cloud infrastructure services and foundation models. The C3 AI platform was purpose-built and hardened over 15 years to rapidly deliver robust customer AI applications. Now let me take a moment to unpack what we mean by pre-built applications as this is a key differentiator for C3 AI.
As a customer, if you buy a pre-built application, you don't need to define what the application does, design and develop the ontology, the business logic, the AI or GenAI model pipelines or the user interface. We've done all that. C3 AI pre-built applications are tried, tested, and proven in the market, so they can be rapidly configured to the customer data sources. AI/ML models tuned to customers' data and the user interface tuned for the customer. This allows customers to quickly configure, deploy, onboard, and train users, scale out their enterprise and rapidly realize enterprise value. Let me give you an example. Holcim, which is a large global building solutions manufacturer, scaled out C3 AI reliability, our pre-built AI-based predictive maintenance application to most of its plants in just eight months after a successful pilot. The application monitors several hundred critical plant assets such as vertical roller mills and ball mills to accurately predict potential problems well in advance of failure.
It does so by continuously ingesting data from over 9,500 sensors and other systems and analyzes that data using 850 machine learning models. The C3 AI reliability application allows Holcim to avoid significant operational disruptions and high-cost remediation. Today, we have pre-built AI applications with turnkey ontologies that address entire value chains across industry verticals. These applications don't operate in silos. They're designed and built to work in concert, all running on the C3 AI platform. Each new application can deploy faster than the last, thanks to the ability to reuse data integrations, ontologies, AI/ML pipelines, workflows, and user interface components. And as more C3 AI applications are deployed, enterprises accrue business benefits faster and the pace of their AI transformation accelerates. We also offer a suite of pre-built and extensible C3 Generative AI applications that can be used as standalone solutions or deployed alongside other C3 AI applications.
This quarter, we deployed C3 generative AI for industrial asset inspections at a large manufacturer. This application unifies data across unstructured inspection reports and structured data on asset performance, such as risk reports, corrosion analysis, work orders, etc., which makes these data easily and quickly accessible to plant operators. Some of these inspection reports include scans of handwritten text and engineering diagrams that are over 50 years old. Using this application, the company has been able to streamline the day-to-day work of inspectors, reduce costly operational mistakes and risks, and accelerate their turnaround efforts and time. In Q1, we launched C3 Generative AI for government programs, which has immense potential for federal, state or local government agencies by eliminating service delays, reducing wait times, enhancing the effectiveness of contact centers and improving the citizen experience.
We've already signed a pilot with a state on the U.S. East Coast and look forward to supporting public sector agencies across the board. It's important to understand that the C3 Generative AI applications are unique in the market. We provide the ability to interact with omni-modal data that is structured sensor data, databases, unstructured documents, images, embedded tables, leverage any LLM available in the market. For example, GPT 4.0, Gemini, Claude, LAMA, Mixtral, to name a few, and multiple specialized LLMs in the same application, offering full source traceability for both structured and unstructured data and granular enterprise access controls. We also provide tools to enable data scientists to tune system responses for high accuracy in production with minimal hallucination, perform chain of thought reasoning, author and execute mathematical functions, offer state-of-the-art LLM guardrails and protections, and orchestrate and chain together AI agents to perform high-value tasks.
For example, to initiate workflows or write summaries. We provide out-of-the-box pipelines to synthesize results across AI agents, automatically detect and surfacing consistencies in underlying data, capture engine logs for auditability and traceability, and provide tools for our customers, developers, and data scientists to configure their own custom generative AI pipelines and offer fine-tuning services for embedding models and LLMs. C3 Generative AI is also unique in its enterprise-grade support for operating in highly secure air gap environments. This is a prerequisite for defense intelligence and financial services customers.
Thank you, Ed. I will now provide a recap of our financial results and additional color on our business. All figures are non-GAAP unless otherwise noted. As Tom mentioned, total revenue for the quarter increased 21% year-over-year to $87.2 million. Subscription revenue increased 20% year-over-year to $73.5 million, representing 84% of total revenue. As a reminder, our subscription revenue is comprised primarily of software licenses, Software-as-a-Service offerings, standard support services, pilots and trials of our C3 AI applications or generative AI and consumption-based pricing for which revenue is recognized over time. Our subscription revenue also includes revenue from software licenses for which ongoing maintenance and support is not required. In accordance with ASC 606, the revenue is recognized when the license is made available to the customer. Professional services revenue was $13.7 million.
This represents 16% of total revenue in the first quarter of fiscal '25 as compared to 15% of total revenue in the first quarter of fiscal '24. We expect the professional services revenue to generally stay within 10% to 20% of total revenue for fiscal '25. Gross profit for the quarter was $60.9 million and gross margin was 70%. Gross margin for professional services remained high at over 90%. Operating loss for the quarter was $16.6 million. Our operating loss was better than guidance due to continued focus on expense management. Our net cash provided by operating activities was $8 million. Free cash flow for the quarter was positive $7.1 million. We were able to generate positive free cash flow during the quarter because of continued focus on cash management. We continue to be very well capitalized and closed the quarter with $762.5 million in cash, cash equivalents, and marketable securities, an increase of $12.2 million compared to Q4.
At the end of Q1, our accounts receivable balance was $140.1 million, including unbilled receivables of $81.5 million. Total allowance for bad debt remains de minimis at less than $400,000, and we do not have concerns regarding collections. The general health of our accounts receivable remains strong. During the first quarter, we signed 52 pilots, a 117% increase from last year and up 53% from last quarter. At quarter end, we had cumulatively signed 224 pilots, of which 191 are still active. This means they are either in their original three to six-month term or extended for some duration or converted to subscription or consumption contracts or are currently being negotiated for conversion to subscription or consumption contracts. 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.
We also expect short-term pressure on our operating margin due to additional investments we are making in our business, including in our salesforce, research and development, and marketing spend. As we continue to make significant investments in the business, we expect to be free cash flow negative for Q2 and Q3, but remain on track to be free cash flow positive for Q4 and also for the full fiscal year '25. Now I'll move on to our guidance for the next quarter. Our revenue guidance for Q2 is going to be $88.6 million to $93.6 million. We are maintaining our previous guidance of $370 million to $395 million for fiscal '25. This implies a year-over-year growth rate of 19% to 27%, making C3 AI one of the fastest growing companies in the software public company universe. Our guidance for non-GAAP loss from operations for Q2 is $26.7 million to $34.7 million. We are maintaining our previous guidance of $95 million to $125 million for fiscal '25.
Thank you, Hitesh. Let me address the path to profitability. Our cost of goods sold is substantially less than our cost of generating revenue. As such, C3 AI is a structurally profitable business. Our year-over-year revenue growth in Q1 was 21% and accelerating. Our year-over-year expense growth rate was 12%. In the coming years, we expect our revenue to generally grow at a greater rate than our expense growth rate. It follows that non-GAAP profitability is now simply a function of scale. While we continue to invest in market share, we believe our revenue growth rates would generally exceed our expense growth rates. The expense and revenue lines will converge, crossing over to consistent non-GAAP profitability. At this time, we expect to be cash flow positive in Q4 fiscal year '25 and for the entire year of fiscal year '25. Now let's talk about the general area of the enterprise AI market and customer success.
In C3 AI, we really are partners to our customers, often on speed dial, engaging with them multiple times a day to make them successful and self-sufficient. We're not working from home; we are out there on-site with our customers every day, every week. We sit shoulder to shoulder, and we get into the details of their business to make sure that everything is running smoothly as it relates to their enterprise AI applications. We hold weekly executive reviews to track progress at the highest levels. This proactive, boots-on-the-ground approach is what drives the customer satisfaction reflected in our net promoter scores. This commitment to close hands-on collaboration is demonstrated in our partnership with Shell. Shell has over 100 C3 AI applications in development and deployment for everything from asset integrity to production optimization. We started working with Shell initially on predictive maintenance back in 2018.
Now they monitor over 15,000 pieces of equipment with C3 AI. They estimate that this program alone generates, this whole C3 AI program that they have written large as they call Shell AI, generates an annual benefit to Shell of $2 billion and we're not done. We continue to collaborate expanding into new use cases, including some service reservoir management and oil condition monitoring. These applications we expect to be scaled out across multiple assets. This approach is also evident in our work with Con Edison. We began our partnership with Con Ed in 2017, initially focusing on the advanced metering infrastructure project. This is a project that enhances the operational efficiency, public safety, and customer satisfaction of their smart grid infrastructure, and it continues to scale. Annual cost reduction uncovered is over $45 million. The project is exceeding expectations with a projected benefit of more than $3.2 billion over 20 years, $500 million more than originally projected.
Now Con Ed is deploying generative AI for meter management and asset mapping, which addresses billing discrepancies and reduces missed mappings on the electric grid, thereby reducing cost and increasing customer satisfaction. Customer success is at the core of why we built this company. Now let's take a moment here and think back. For those of you who've been around for a few decades and are in touch with what's going on with people. C3 AI is the original enterprise AI company, hard stop. We began C3 AI in January of 2009 with the vision to develop a software platform and enterprise applications that allow organizations to exploit what we believe would be the computing platform of the future. This included elastic cloud computing, the Internet of Things, big data, and predictive analytics. We did this during the depths of the AI winter, well before the advent of the GPU, before Azure existed, before GCP existed, and when AWS was nascent.
We invested in thousands of personal users over a decade, building the C3 AI platform, the first reference architecture for enterprise AI, hard stop. Today, we have over 90 C3 AI enterprise applications that address the value chains of energy, government, defense, manufacturing, financial services, agribusiness, pharma, and others. We deploy these applications today in Europe, Asia, South America, North America, and in governments around the world. Now, 15 years later, the Elastic Cloud, the Internet of Things, big data, and predictive analytics have become ubiquitous. Enterprise AI is broadly recognized as one of the largest and fastest-growing markets in the history of enterprise software. It is a mandate for every corporate and government leader today to harness the power of enterprise AI, and C3 AI is extraordinarily well-positioned to serve this large and rapidly growing market demand.
C3 AI is today one of the fastest growing companies in the public software universe. Most importantly, in this current cacophony of AI market hype, C3 AI is achieving among the highest levels of customer satisfaction for value realized in the enterprise software world. I refer you to our earnings supplemental deck today, where we're showcasing C3 AI's Net Promoter Scores, benchmarking C3 AI against other leading enterprise software companies, many of which prefer to be, about all of which purport to be AI companies. This report is published by a third party called Comparable. The bottom line is our enterprise software companies are the most satisfied in the software industry, consistently driving unmatched value from our solutions. Achieving this level of customer satisfaction doesn't just happen; it is the result of relentless dedication, unwavering commitment, a thorough understanding of our customer needs, superior software technology, and deep expertise in enterprise AI, honed over many years of excellence.
In the final analysis, customer value realized will be the only criterion that will determine the leader in enterprise AI, and C3 AI is exactly on track. Now, I'd like to turn this over to the operator to begin our Q&A session. Thank you for all. Thank you all.
分析師問答
Thank you. Our first question comes from Patrick Walravens with Citizens JMP. Your line is open.
Great. Thank you very much and congratulations. Hey, Tom. Can you start out by just sort of characterizing what the tone of business was like for you guys in Q1? And specifically, I'm looking at your deal band chart, where you had 71 deals and the average TCV is $700. So if you multiply those together, you get like almost $50 million, which is up a lot. So just if you could just comment on the tone of business, that would be great.
It's quite chaotic in the market right now, especially with the emergence of generative AI. It's challenging for people evaluating these companies to grasp the complexities of the AI market. There are numerous variations of enterprise AI, and generative AI applications vary widely in scale and pricing, making it hard to predict future performance. This isn't a straightforward business. As we navigate this new AI landscape, the situation is quite intricate. We can't just project future results based on past sales volumes. The growth rates and profit margins are unique to this space. We continue to be surprised by the diverse applications we're discovering for enterprise AI, particularly in generative AI, many of which we hadn't foreseen. For example, we didn’t expect to find significant opportunities in the public sector, but that turned out to be highly profitable. Law firms and medical diagnostics are also proving to be intriguing areas.
You're talking about the guidance for the Q2, revenue guidance.
No, for the year, you kept it the same.
Yes, that still represents a 19% to 27% revenue growth. It still makes us one of the fastest-growing companies in this public software company universe.
Thank you. Please stand by for our next questions. Our next question comes from the line of Timothy Horan with Oppenheimer. Your line is open.
Thanks, guys. On the subscription revenue has been a little lumpy here. On professional services, I know, Tom, you basically just said it's very difficult to model out. But can you give us some color on the trends? Like is this a good jump-off point for the rest of the year? And then maybe just on the expenses, where should we be modeling and the expenses increased almost in the next quarter or two? Thanks.
Thanks. Let me address subscription versus services under ASC-606, it's a little complex. Certain things that used to be called software are now called services under the new guidance, and we comply with what the guidance is. That being said, we've guided that our services revenue would be 10% to 20% of revenue in any given quarter. Our services revenue was 16% in this quarter. So we're well within our guidance there. Before we kind of bemoan the fact services were a little higher than they expected to be, let's remember what services margins are here at C3 AI. While our software certification margins are quite high at 66%, our services margin, you'll recall, is in excess of 90%. So services is a pretty darn good business. And while we continue to be focused, I mean, we're a licensing company, okay? We're not one of these services companies pretending to be a software company. There's one or two of them out there. We're not bad, but we continue to be focused. For the next year or two, you can expect our services to bounce around in the 10% to 20%. It will probably average about 15%. But remember, our services margins are greater than 90%. I think 93% is the number I got. So it's good work if you can get it.
No, no reclassification.
Okay. Great. And expense line items, anything to focus on the next quarter or two?
Yeah. So in terms of our expenses, we plan to continue to make investments in our sales force, R&D, as well as marketing efforts.
Okay. Great. Pretty much across the board.
Well, Ed, what is the competitive environment? You're closer to it than I.
Yeah. I think the default or de facto competition is the information technology, the IT organization, the CIO trying to build these applications themselves. These are our best prospects, the ones that have tried to do data science and scale that up across a large enterprise, typically figure out how difficult that is without the right data and AI platform. So that is the de facto competition in the market for us.
Those are potential opportunities a few years ahead because they often fall apart. We have a CIO overseeing 10,000 employees in Bangalore attempting to construct this from components provided by hyperscalers, and it typically fails around them. So we place them in our pipeline for two or three years down the line, and then they return.
Thank you. Please stand by for our next question. Our next question comes from the line of Mike Latimore with Northland Capital Markets. Your line is open.
Hi, team. Thanks. The pilot growth was very strong sequentially, year-over-year. Are the sizes of the pilots, the value of the pilots for standard and generative AI kind of as expected?
The value of the pilots, on average, is approximately $0.5 million for enterprise AI applications and around $0.25 million for generative AI applications. A generative AI pilot typically requires three months to complete, while an enterprise AI pilot takes about six months. After that, we enter contract negotiations before they go live. We previously indicated that we expect around 70% of our pilots to transition into production contracts, and I believe that figure is accurate. The 30% that do not convert is not due to unsuccessful projects; rather, it often happens because a decision-maker chooses to attempt building the solution independently using components from a hyperscaler. They will likely return in the future.
Got it. Okay. So it sounds like there will be no change in pricing or value regarding pilots. As for the second quarter guidance, the midpoint indicates approximately 24% revenue growth, which is a significant increase from this quarter's growth. I believe this is the largest increase in the last four quarters. What gives you confidence in that improvement? Is it the pilots you've recently secured, the growth in professional services, or substantial deals being deployed? What contributes to your confidence in that acceleration?
Using our best professional judgment, the confidence we express reflects our expectations for the future. Having been public for 15 quarters, we have accurately predicted outcomes each time. I recognize that many of you are trying to model our business. If you can succeed at that, you may have an advantage over us, as there are many variables involved. The most reliable indicator of our performance is the guidance we provide, which has proven to be accurate for the last 15 quarters. Currently, this is our expectation, and we will be working hard to meet or exceed it.
Thank you. Ladies and gentlemen, we will take our last question from the line of Kingsley Crane with Canaccord. Your line is open.
All right. Thanks for taking the question. 40 agreements with GCP, that's spectacular. 51 agreements with the partner network overall. So could you speak to the partner efforts outside of GCP? What's working well? What would you like to improve? And then is the GCP partnership as dominant from a bookings contribution perspective as it is in deal frequency?
I'm not sure, Kingsley; I don't think I have the bookings contribution data here. So I can't correctly answer that one. I would say that the relationship with GCP is great. The relationship with AWS is great. We're doing a lot of good work with AWS, with Microsoft Azure. I mean, we're great partners for these guys. What we do when they partner with us, we have the application up fast, the customer gets value; they don't make money when companies are building applications using their platforms; they make money when people are running their applications. So we're consuming CPU resources or consuming GPU resources or consuming storage. They're all great partners, they're great companies, they're wonderful to work with, and it's a privilege for us to be able to partner with them.
All right. Okay. That's very helpful. And then last one, just to clarify. For the state and local agreements, were most of those pilots?
They all kind of started as pilots, yes. I'm not sure which number you're referring to. Did we refer to how many pilots that were and how many converted in the quarter? We said this?
No. We don't talk about the conversions. We have the number for how many pilots closed in the quarter, 52...
Kingsley, they almost all began as pilots. They all begin as pilots. They go from, let's say, one to three months, and then the great majority of them convert to production. Thank you, everybody. I think this is the end of our call. We really appreciate your attention. This is quite an adventure. I think we're breaking ground in enterprise AI. I know we're breaking ground in enterprise AI. And I could just tell you, for those of you who have visited us or those of you who have a chance to visit, it's so exciting, and it's just palpable. We're going to continue to get after it, and we thank you for your time. We thank you for your attention, and wish you all a good day.
Ladies and gentlemen, this concludes today's conference call. Thank you for participating. You may now disconnect.