C3.ai S1 Teardown
5 min read

C3.ai S1 Teardown

C3.ai S1 Teardown

Overview

C3.ai is an enterprise AI software company and founded in 2009 by Silicon Valley technology legend Tom Siebel of Oracle and Sibel Systems fame.

The applications built by C3 enable the rapid deployment of highly complex, large scale projects.

The core suite of technology is an application development and runtime environment that allows enterprises to rapidly design, develop and deploy AI applications at scale.

The C3 AI Suite, with its model-driven architecture, addresses the requirements for the digital transformation software stack, providing a low-code/no-code AI and Internet of Things platform that accelerates software development, reduces cost and risk, and delivers applications that are flexible enough to meet evolving needs

C3 AI Applications is an app store-like product that includes a large section of turnkey AI solutions that are industry or application specific and can be immediately deployed.

C3 estimates that their approach speeds development time within their customer set by 26x and reduces the amount of code that needs to be written by 99%.

SaaS Benchmarks

C3's last valuation was $3.3B and it will no doubt be targeting significantly more than that in the IPO.

C3 v. BVPs Cloud Index (B2B)

To reach that number, C3 needs to be valued at ~21x LTM revenues. As you can see above, that might be a stretch given C3's performance to date.

While C3's growth has been solid, especially for a company of its size, it is only average when benchmarked against the fastest growing enterprise SaaS equities.

I don't put as much stock into gross margins when it comes to complex applications like C3. Additionally, it is not uncommon to see a services or consulting component when working with laggard industries.

Given the size and complexity of integration, I am slightly surprised that C3's revenue numbers aren't higher as a result of a higher ACV and low likelihood of churn.

Cap Table

C3 has raised over $350M to-date from investors including BlackRock, TPG, Baker Hughes, Shell, and Constellation Technology Ventures.

Multi-time entrepreneur and founder Tom Sibel owns 34% of C3.ai, but more importantly, owns 76% of voting power through the company's Class B shares.

TPG owns 22.6% and Baker Hughes around 15%.

Business Model

The bulk of C3's revenue comes from software subscriptions - roughly 86% as of this quarter.  The revenue is broken down into 4 primary categories.

  • 3-year subscriptions for the C3 AI Suite
  • 3-year subscriptions for C3 AI Applications
  • Usage-based CPU-hour consumption fees
  • Services fees associated with training and assisting customers

As you might expect, since C3 works in large, complex industries like energy and defense the average contract value (ACV) is massive. In 2016, 2017, 2018, 2019, and 2020 ACVs were $1.2 million, $11.7 million, $10.8 million, $16.2 million, and $12.1 million, respectively.

More on the decrease from 2019 to 2020 later in this post.

C3 is phenomenal at land and expand within its largest customers. The average initial contract value with the largest 15 customers was $12.8 million. On average, each of those customers purchased an additional $26.1 million in product subscriptions and services.

Performance

C3's revenues grew from $91.6M to 156.7M year-over-year. Like most high-growth startups, C3 appears to lose more money as it grows.

Fiscal 2019 saw a loss of $33.1M and that number more than doubled in 2020 to $69M.

Gross margins continue to improve as the services component becomes less important to the business. GM's have peaked as high as 81% and are frequently in the mid to upper 70% range.

C3's quarter over quarter numbers actually declined in the most recent quarter. The company attributes this decline to re-working its contract with Baker Hughes. Given the troubles of the oil and gas market of late, it won't be a surprise if C3 sees a decline in revenues from the sector - at least temporarily.

Go-to-Market

C3.ai has a traditional go-to-market strategy with sales teams divided into regions and sectors. In addition to this traditional approach, C3 deploys a partnership approach within industries where the barrier to trust has been traditionally high.

In oil and gas, C3.ai has a partnership with its investor and oil field services giant Baker Hughes. They use the same playbook in defense with Raytheon as their partner of choice.

Risks

The company identifies the following as key risks outside of the normal S1 jargon.

  • Long, unpredictable sales cycles
  • Revenue concentration
  • The market for AI dries up (unlikely)
  • Increasing competition (highly-likely)

Revenue concentration is the elephant in the room from where I sit. While C3 doesn't give us the exact numbers, we know that the largest 15 customers likely account for ~500-600M of annual revenue or roughly 50-60%.

What we don't know is how many of those customers are in oil and gas. We also don't know how oil and gas reacts to margin pressure as a result of stagnant prices.

Do they try to optimize and spend on software? Do they tighten budgets all together? Do they use excess capital to buy depressed assets or renewable assets? All of these questions remain unanswered.

An additional long-term fringe risk is the proliferation of AI /ML / data science talent in the coming years. It appears C3 has enjoyed a substantial talent advantage over its customers and their industries. Will that continue over time? Or will AI / ML remain a continued hotbed of talent and become more ubiquitous?

Analysis

C3 is a first-mover in a space that is certain to become larger with time.  AI and machine learning will permeate every industry moving forward.  I have little doubt the market can support multiple winners much less one and competition is heating up here.

The short-term question for C3 is revenue concentration, not only by customer but by segment. What % of revenue is tied up in the oil and gas sector and how does that commodity risk play out over time?

Given the company's cash position and high ACV, there's little concern over the length of sales cycles even though they are long and uneven. They should also shorten over time as even laggard industries adopt increasingly complex software.

Finally, C3 seems to have taken advantage of a talent gap in AI/ML engineering. In this way, I'm actually reminded of Crowdstrike and cybersecurity. Will talent become more accessible and ubiquitous in the next 5-10 years?

The next 2-3 years should be exciting for C3, the energy transition is in full force and secular tailwinds behind the technology should increase its visibility within every sector. Can C3 overcome the near-term challenge and what is sure to be an increasingly competitive space? Only time will tell.