rewrite this title Can AI Agents Be an Agent of Revenue Growth for C3 AI? – Nanalyze

Last month, we wrote about how ServiceNow (NOW) may be the best company for investing in AI agents. These intelligent algorithms are more than the sum of their code, able to act autonomously to gather data, make decisions, and perform tasks. As we noted in the article, AI agents are like robotic process automation on steroids. The big question is whether the technology will be powerful enough to pump up the enterprise software company’s revenue growth, which has been strong and steady at between 22% and 24% the last three years.

The early signs are promising. As we reported, the fastest-growing product line in ServiceNow’s history is an agentic AI tool called “Now Assist” that currently generates about $250 million in annual contract value (ACV). It sounds impressive, but that’s just 2% of the company’s total ACV. Management claims the number will quadruple next year to $1 billion thanks to a new pricing model based on usage. 

It’s estimated that more than half of software-as-a–service (SaaS) firms now employ at least some hybrid version of a consumption-based model that charges based on clicks or client seats. We did a deeper dive into this trend last month in our article about Confluent (CFLT) and its pivot into a cloud-based consumption model. Go read that one and then come back to learn about another SaaS company making the transition from a pure subscription model to a consumption-based business, while also pursuing the potentially lucrative AI agentic market. It’s all going to work out perfectly, right? Right?

The Evolution of C3 AI

C3 AI (AI) is an enterprise software company that has reinvented itself a few times since it was founded back in 2009. It originally focused on harnessing the power of cloud computing, big data, and the emerging Internet of Things (IoT) to deliver predictive analytics and energy management solutions for enterprises and utilities. C3 eventually jumped on the bandwagon expanded its platform to include machine learning and other artificial intelligence (AI) technologies to automate business processes. C3 continued to rebrand and retool over the years, and today is all about AI. Hence, the (latest) official name – C3 AI. At a recent tech conference, C3 CEO and founder Tom Siebel claimed his company has spent more than $3 billion building out its platform.

However, the last couple of articles we’ve written about C3 AI stock questioned how well all that money was spent. We first covered the slew of accounting irregularities, general criticisms, and short reports back in 2023. When we followed up last year, C3 had tried to address the controversies and called upon the unholy power of generative AI to entice and impress both customers and investors. At the time, we concluded that it was not a business that we would likely invest in despite decent revenue growth. The lack of strong SaaS metrics and a general sense that management was blowing (or inhaling) a bunch of smoke put us off. There are also customer concentration concerns and doubts about their ability to generate alpha from generative AI. 

Now we’re two years further down the road from the clouds of controversy: Are there enough compelling reasons to hold shares of C3 AI stock?

Breaking Down the C3 AI Business Model 

Let’s start with revenue growth. Since the company stumbled to single-digit revenue growth in 2023, it has come back to post double-digit increases over the last two years, including a 25% year-over-year rise to nearly $390 million in 2025. Accounting for 84% of total revenue, subscription income grew 18% for the year. 

However, in Q4-2025, revenue from subscriptions represented just 80% of total revenue, with professional services accounting for the rest. Subscription revenue accounted for 90% of total revenues in 2024 but 86% back in 2023. What is going on? 

Consumption-Based or Employee-Based?

The shift to a consumption-based model, which began in Q1-2023, aims to lower barriers to entry for new clients but also introduces revenue volatility, which may partly explain why revenue periodically skews higher toward professional services (more on that momentarily). Strangely enough, gross margins appear to be much better for professional services than software subscriptions. This is perplexing. Professional services, which rely on overpaid software engineers shotgunning expensive ketamine cocktails, should theoretically have lower margins than scalable software subscriptions.

A few factors appear to be in play here. C3 AI’s subscription revenue currently includes a significant share of pilot projects and trials. These pilots are resource-intensive, requiring substantial support, customization, and engineering effort to demonstrate value to potential customers. The transition to consumption-based revenue is compounding the issue, because these also require short-term pilots or smaller contracts with higher upfront costs to onboard and support new customers. Management says it has closed more than 100 pilots on the consumption model since its introduction, leaving us a bit in the dark on its actual success. 

This kind of begs the question: If C3’s $3 billion platform is so great, why does it require so much time and money (including nearly a quarter of a billion dollars – 62% of total revenues – in sales and marketing in 2025) to convince customers of its value? 

In addition, a significant portion of C3 AI’s professional services revenue (up to 80%) comes from what it calls Prioritized Engineering Services (PES). PES involves customers paying C3 AI to accelerate development of new software features, which then become part of the core product. These projects are negotiated at fixed fees and result in production-level software, allowing C3 AI to recognize revenue for work that (again, theoretically) also benefits its broader product line. This model can potentially drive higher margins than standard consulting, because the company can (yes, theoretically) monetize new innovations across their entire customer base.

Partners for Life?

Another key part of C3 AI’s go-to-market strategy is its partnership network. For instance, about 73% of agreements in Q4-2025 and 193 deals for the full year (a 68% increase year-over-year) were secured through collaborations with major partners such as Microsoft, AWS, and Google Cloud. Partner-supported bookings shot up almost 420% in the last quarter and demo licenses accounted for nearly 30% of the company’s Q4 revenues. Microsoft is an especially important ally: Together, the two companies closed 28 joint agreements in a single quarter, and they are jointly targeting more than 600 accounts globally.

On one hand, these partnerships certainly help accelerate the adoption of C3 AI software products to market by selling, implementing, and supporting solutions across a diverse range of industries and customers. And, in fact, C3 is diversifying its revenues streams beyond its historical reliance on the oil and gas sector. Revenues outside this sector surged 48% year-over-year. The company closed 71 agreements in government sectors alone, driving more than 100% growth (of what?) in that vertical. 

On the other hand, one has to assume that these deals are another factor eroding gross margins in subscription sales. After all, Microsoft is not flogging C3 AI software out of the goodness of its microscopic heart. The approach comes with other potential drawbacks. For instance, these partnerships, particularly with companies like Microsoft, potentially represent significant customer concentration risk (not to mention relying on potential competitors with shifting priorities). How much exposure? We don’t know because C3 does not offer up any details, except conceding facts such as three-quarters of Q4-2025 deals came via partners. Tell us again how you spent $240 million on sales and marketing?

These concerns are outside of the long-standing relationship that C3 AI has with Baker Hughes, an oilfield industrial services firm that was (still is?) one of the biggest shareholders in the enterprise AI company. Over the last six years, Baker Hughes has committed $495 million to C3 AI – or about 30% of the $1.56 billion in total revenue generated during that time. In April 2025, the companies renewed and expanded their partnership through a “multi-year agreement” without disclosing any financial terms.

All In on AI

So, we are not entirely sold on the business model or go-to-market strategy. But what about the solution itself? As of today, C3 offers three broad AI-driven platforms with plenty of AI agents thrown in: 

C3 AI Applications: a portfolio of industry-specific Enterprise AI applications that enable the digital transformation of organizations globally

C3 Generative AI: a library of agentic AI applications to retrieve data, analyze information, surface insights, and orchestrate workflows to drive business value

C3 Agentic AI Platform: an end-to-end platform for developing, deploying, and operating Enterprise AI applications

The company’s technology strategy hinges on its ability to offer hyper-specialized AI applications to specific industries, including 28 domain-specific generative AI models. Mega companies like Amazon and Google are similarly investing in agentic AI frameworks through their cloud services to automate complex business processes. However, some analysts see C3 AI’s focus on pre-built, industry-tailored solutions in sectors like defense, healthcare, and utilities as a potentially winning strategy.

CEO Siebel would certainly make that argument. During the Q4-2025 earnings call, he said his company has more than 100 agentic AI solutions deployed in those and related sectors. He claimed the agentic AI business has annualized recurring revenues of about $60 million.

“If we were to spin that business out … and take it to an Andreessen Horowitz or a Bessemer or Nvidia or whatever it is, that business alone would be valued at multiples of where C3.ai, Inc. is today, and we all know that’s a true statement. Where is our growth going to come from? Our growth will come from additional applications.”

Remember how at the top of the article we said ServiceNow is already generating about $250 million in ACV from its agentic AI tool – or about 65% of C3’s total revenues? We think it might be a hard sell, Tom. Or is Siebel ready to sell out to one of its suitors partners a la Oracle-Siebel Systems?

Out on AI Stock

Some of our other reservations about C3 AI stock continue to persist. Since its IPO, C3 has experienced significant stock dilution. As of early 2025, the number of shares outstanding has increased to nearly 133 million from 99 million in December 2020. In other words, total share count has grown by roughly one-third over this period, representing a 33% dilution. A primary culprit is stock-based compensation (SBC), another red flag. In 2025, SBC was about $231 million, consistent with the prior year’s figure of about $215 million. 

The argument is that sky-high SBC is necessary to retain the best and brightest. But at 60% of total revenue? For comparison, Palantir Technologies spent $721 million in SBC, which was less than a quarter of its total revenues. ServiceNow paid out a staggering $1.75 billion in SBC but that’s less than 15% of total revenues. The company did become cash flow positive in Q4-2025 but appears far from becoming profitable based on R&D and sales and marketing spend alone. The lack of other SaaS metrics, such as net revenue retention rate, suggests company growth is focused on new customers (involving more expensive upfront costs) rather than cross-selling solutions to existing clients.

Conclusion

C3 AI management is guiding between 15% and 25% revenue growth, underscoring the ongoing uncertainty inherent in the company’s business model. Unlike shares of Palantir (PLTR), which are trading at astronomical valuations, C3 AI stock sports a simple valuation ratio ($3.2 billion market cap/$436 million in annualized revenue) of about 7. That’s at our catalog average, but we have no intentions of making any trades on C3 AI stock anytime soon.  At best, C3 AI stock is a hold thanks to continued revenue growth and our ongoing belief in the agentic AI and broader enterprise AI market opportunity thesis.