S-1 Teardown of C3.ai

• LOL the ticker - are you kidding me? AI! ARTIFICIAL INTELLIGENCE!

• No Goldman Sachs as an underwriter - must not be using their software.

• Thomas Siebel opening up with the last 40 years of his career. It's like framing the whole company as a bet on himself

• "We were initially formed in 2009 as C3, LLC, a Delaware limited liability company. In June 2012, we incorporated under the laws of the state of Delaware under the name C3, Inc. In July 2016, we changed our name to C3 IoT, Inc., and, in June 2019, we changed our name to C3.ai, Inc."

  • Company just changing its name and riding the buzz-hype train to funding town

• Not sure how I feel about the best ENTERPRISE AI company being run by a billionaire CEO worth $4B

• "From 1983 through 2006, we saw one wave after another of new technologies: mainframes, minicomputers, personal computers, the internet, relational database technology, enterprise application software, and client-server computing. Each technology breakthrough represented a replacement market for its predecessor, fueling a $1.3 trillion industry by 2006." - Very true

• "I believe we are well-positioned to succeed. The market is large and rapidly growing. We have succeeded at developing a highly differentiated and efficacious AI development platform and an associated family of AI applications. We have manageable competitive risk compared to others, including – (1) companies attempting to build the application from scratch – with little to no success – and (2) a plethora of AI point solutions each of which addresses a small slice of the problem. It boils down to execution risk. Does the C3.ai team have the skills and experience to succeed? Can they manage a rapidly growing business? Can they successfully implement mission critical extraprise application deployments? Can they attract, retain, and motivate the top people in the industry? Can they establish rewarding strategic partnerships with customers and market partners? Can they effectively scale and manage business sales, marketing, and support infrastructures globally? Can they accelerate and maintain technology leadership? I believe that the strength and experience of management and human capital at C3.ai is our strongest asset. This is unquestionably the most talented and experienced team that I have worked with in my career. I believe C3.ai is uniquely qualified to tackle these challenges. But clearly, as an investor, you will need to resolve these questions to your satisfaction."

  • This is WILD - I feel like I'm being sold to - when have you ever seen a letter from the CEO articulate the questions you should be deciding around investing

  • Also could you imagine naming your company after yourself? Cough Siebel Systems cough....

• First let me say, I'm pretty critical of pure AI company's because I think AI solutions have to be highly specific to the domain use cases - I'm unsure what horizontal AI applicability really is although that is certainly the way they are pitching it

Market

• Market size: "We serve a large and rapidly growing market, estimated to be $174 billion in 2020, growing to $271 billion in 2024, a 12% compound annual growth rate, or CAGR.

  • Our solutions address use cases across:

  • Enterprise AI Software. $18 billion in 2020, $44 billion in 2024, a 24% CAGR.1

  • Enterprise Infrastructure Software. Application Development, Infrastructure, and Middleware; Data Integration and Quality Tools, and Master Data Management Products: $63 billion in 2020, $82 billion in 2024, a 7% CAGR.2

  • Enterprise Applications. Analytics, Business Intelligence and Customer Relationship Management, or CRM: $93 billion in 2020, $145 billion in 2024, a 12% CAGR.3

  • 1 Source: IDC, Worldwide Artificial Intelligence Systems Spending Guide, September 2019

  • 2 Source: Gartner, Forecast: Enterprise Infrastructure Software, Worldwide, 2018-2024, 3Q20 Update

  • 3 Source: Garter, Forecast: Enterprise Application Software, Worldwide, 2018-2024, 3Q20 Update"

• Massive market for sure - but the company only has 29 customers? Interesting. More to come on that later

• I'm surprised Enterprise AI software is growing that fast

• "Enterprise AI is a huge addressable market."

  • This just screams VC - we have a massive market opportunity, don't worry about us only having 29 customers and a weird sales motion / contracting / customization focus

PPP BRUH

• "In May 2020, given the uncertainty caused by COVID-19 and related events we applied for and received proceeds of approximately $6.3 million from a loan under the Paycheck Protection Program, or the PPP Loan, of the Coronavirus Aid, Relief, and Economic Security Act, or the CARES Act. The PPP Loan had a term of two years, was unsecured, and was guaranteed by the U. S. Small Business Administration, or the SBA. The PPP Loan carried a fixed interest rate of 1.00% per annum, with the first six months of interest deferred. Under the CARES Act, we may have been eligible to apply for forgiveness of all loan proceeds used to pay payroll costs, rent, utilities, and other qualifying expenses, provided that we retained a certain number of employees and maintain compensation within certain regulatory parameters of the Paycheck Protection Program. However, we repaid the entire balance of the PPP Loan in August 2020."

  • WTF - a company with 100M in cash decided to apply for PPP money?

• 
  

CUSTOMER / customer ?

• "We commonly enter into enterprise-wide agreements with entities that include multiple operating entities or divisions. We define a Customer as each such buying entity that has an active contract to deploy the C3 AI Suite or one or more C3 AI Applications. We often provide our software to a distinct department, business unit, or group within a Customer, and use customer to mean each distinct department, unit, or group within a Customer. As of September 30, 2020, we had 29 Customers and 59 customers."

  • WHAAAAAT? Customer = actual customer, customer = separate operating division of a big Company

  • How can the SEC be like yeah this makes sense and is not confusing to the average investor???

    • LOL the SEC made them change this to "Entity" and "customer" - LOL

• Literally LOLOLOLING: "As of September 30, 2020, we had 29 Customers and 59 customers."

  • Always NOT a good sign when the company is playing games with customer count

Baker Hughes with a weird contract

• "In June 2019, we entered into a three-year arrangement with Baker Hughes as both a leading customer and as a partner in the oil and gas industry. This arrangement included a subscription to our AI Suite for their own operations (which we refer to below as direct subscription fees), the exclusive right for Baker Hughes to resell our offerings worldwide in the oil and gas industry, and the non-exclusive right to resell our offerings in other industries. Under the arrangement, Baker Hughes made minimum, non-cancelable, total revenue commitments to us of $50.0 million, $100.0 million, and $170.0 million, which are inclusive of their direct subscription fees of $39.5 million per year, for each of the fiscal years ending April 30, 2020, 2021, and 2022, respectively, with the remainder to be generated from the resale of our solutions by the Baker Hughes sales organization. During the fiscal year ended April 30, 2020, we recognized as revenue the full value of the first year of the direct subscription agreement and the value of deals brought in by Baker Hughes through the reseller arrangement. This arrangement was revised in June 2020 to extend the term by an additional two years, for a total of five years, with an expiration date in the fiscal year ending April 30, 2024 and to modify the annual amount of Baker Hughes’ commitments to $53.3 million, $75.0 million, $125.0 million, and $150.0 million, which are inclusive of their revised direct subscription fees of $27.2 million per year over the fiscal years ending April 30, 2021, 2022, 2023, and 2024, respectively. Any shortfalls against the total annual revenue commitment made to us by Baker Hughes will be assessed and recorded by us at the end of the fourth quarter of each fiscal year. We are obligated to pay Baker Hughes a sales commission on subscriptions to our products and services offerings it resells in excess of these minimum revenue commitments."

  • This is confusing but basically they have a resale agreement with Baker Hughes (not a great acquisition btw) which is weird because does Baker Hughes sell Enterprise AI?

  • Anyway, the contract was supposed to be 50, 100, 170 and now its 53.3, 75, 125, 150 and their subscription fee went down by 10M?

  • This seems like an odd relationship but I frankly don't understand Baker Hughes business model

    • I do know GE recently wrote off some of the Baker Hughes acquisition

• "We experienced a $1.1 million decrease to subscription revenue in the three months ended July 31, 2020 as compared to the three months April 30, 2020, primarily driven by a decrease in subscription revenue of $3.6 million due to the Baker Hughes contract modification that occurred during the three months ended July 31, 2020. "

Quarterly P&L

• Wow so revenue actually decreased QoQ from April to July - that is not a great sign - they are growing 16% YoY and 75% YoY the previous quarter

  • ACCOUNTING RULES WTF - like how is this possible? How can I view this as an accurate picture of the business

• Overall though, revenue has these weird jumps and so do sales and marketing and R&D - that seems weird are they throwing some custom services work in R&D? Like why is it 24M in Oct 2019 quarter and then 12M the very next quarter?

• Also this tidbit is pretty funny: "Having established the scalability of our product offerings and their utility across a wide range of AI use cases at large-enterprise scale, we will now (1) expand those same use cases across similar companies, (2) establish middle market sales organizations, including telesales and online sales, and (3) leverage our marketing partners as enterprise and mid-market distribution channels."

  • So the plan is go down-market with your $15M ACV contracts? Will be interesting to see how they setup a whole selling organization and product to match

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

  • AN S-1 IS A CLEARLY A MARKETING DOCUMENT

Risk Factors

• "Our sales cycles can be long and unpredictable, particularly with respect to large subscriptions, and our sales efforts require considerable time and expense."

  • "The length of our sales cycle, from initial evaluation to payment for our subscriptions is generally six to nine months but can vary substantially from customer to customer and can extend over a number of years for some customers."

    • How do you measure ROI on a massive ACV deal that takes YEARS to close - its just a crazy selling motion you don't normally see

• "Historically, a limited number of customers have accounted for a substantial portion of our revenue. If existing customers do not renew their contracts with us, or if our relationships with our largest customers are impaired or terminated, our revenue could decline, and our results of operations would be adversely impacted."

  • Our top three Entities together accounted for 34% and 44% of our revenue for the years ended April 30, 2019 and 2020, respectively. Our top three Entities by revenue for the year ended April 30, 2020, have been with us for an average of 3.3 years. Each of ENGIE Information et Technologie, or Engie, and Caterpillar, Inc. accounted for greater than 10% of our revenue for the year ended April 30, 2019, and each of Baker Hughes Company, or Baker Hughes, and Engie each accounted for greater than 10% of our revenue for the year ended April 30, 2020.

    • Wow 44% customer concentration from top 3 customers is pretty wild. Also ENGIE is a massive French multinational Utility, Baker Hughes was acquired and divested from GE in less than three years and Caterpillar is a massive construction and equipment manufacturer. Clearly a focus on big, complex deployments

• "We have business and customer relationships with certain entities who are stockholders or are affiliated with our directors, or both, and conflicts of interest may arise because of such relationships. Some of our customers and other business partners are affiliated with certain of our directors or hold shares of our capital stock, or both. For example, in June 2019, we entered into a strategic collaboration agreement with Baker Hughes whereby Baker Hughes had a right to appoint a director. Our director, Lorenzo Simonelli, is an employee of Baker Hughes, and Baker Hughes is a stockholder."

  • Ahhhhh this Baker Hughes deal makes sense now - not incredibly uncommon to see big companies invest in smaller ones around tech relationships - see Visa, Mastercard, Salesforce, Cisco

The AI Suite for Digital Transformation

Portfolio Of Cross Industry Applications

• So one of the biggest critiques with enterprise AI is that its not context specific and ends up lacking real measurable results

• Its interesting that they've verticalized (they definitely started in O&G and Utilities)

• The tough/good thing here is that each of these could be their own multi-billion dollar product/industry - to do all of these well would be super compelling but I wonder about spreading themselves too thin already

  • What is "Data Fusion" and why does it only apply to Aerospace and Defense

• The last product in every single thing is C3 CRM - is this a requirement of adopting the solution? Why do you need a CRM to do AI?

• "Our production footprint across the C3.ai customer base today includes: 770 unique enterprise and extraprise source data integrations; integrated data from 622 million sensors; 4.8 million machine learning models in production use; 1.1 billion predictions per day generated by customers; and 50 million businesses and customers touched daily."

  • This is no joke though - wonder how distributed it is - how many production models come from one customer vs. the rest?

Awash in "AI" Platforms

• First of all I wonder why some vendors are bigger than others - like Databricks, Spark, Tensorflow, Cassandra seem much bigger than everything else?

• "Today the market is awash in “AI Platforms” that purport to be solutions sufficient to design, develop, provision, and operate Enterprise AI applications, including Cassandra, Cloudera, DataStax, AWS IoT, and Hadoop. AWS, Azure, IBM, and Google, each of which offer an elastic cloud computing platform. The point is: these utilities are all useful, but we believe none is sufficient by itself. Each addresses only a part of the problem required to develop and deploy an AI or IoT application."

• “Do It Yourself” AI? Software innovation cycles follow a typical pattern. Early in the cycle, companies often take a “do it yourself” approach and try building the new technology themselves. In the 1980s, for example, when Oracle first introduced relational database management system, or RDBMS, software to the market, interest was high. In our experience, RDBMS technology offered dramatic cost economies and productivity gains in application development and maintenance. We believe it proved an enabling technology for the next generation of enterprise applications that followed, including material requirements planning, or MRP, enterprise resource planning, or ERP, customer relationship management, or CRM, manufacturing automation, and others. The early competitors in the RDBMS market included Oracle, IBM (DB2), Relational Technology (Ingres), and Sperry (Mapper). But the primary competitor to Oracle was not any of these companies. In our experience, it was in many cases the CIO, who attempted to build the organization’s own RDBMS with IT personnel, offshore personnel, or the help of a systems integrator. When those efforts failed, the CIO was replaced and the organization installed a commercial RDBMS. When enterprise applications including ERP and CRM were introduced to the market in the 1990s, the primary competitors included Oracle, SAP, and Siebel Systems. But in the early years of that innovation cycle, many CIOs attempted to develop these complex enterprise applications internally. Hundreds of person-years and hundreds of millions of dollars were spent on those projects. A few years later, a new CIO would install a working commercial system. In our experience, some of the most technologically astute companies—including Hewlett-Packard, IBM, Microsoft, and Compaq—repeatedly failed at internally developed CRM projects. All ultimately became successful Siebel Systems CRM customers. Just as happened with the introduction of RDBMS, ERP, and CRM software in prior innovation cycles, the initial reaction of many IT organizations is to try to internally develop a general-purpose AI and IoT platform, using open source software with a combination of microservices from cloud providers like AWS and Google."

  • I think this is an interesting case study but I'm not actually sure whether this is reality

  • Does Shell really want to develop AI and IoT internally?

DIY / Point / Cloud Solutions

Cash Flow and Business Metric

Team, Board, and Compensation

Key Considerations

• The ticker will be AI and this thing will ride the hype train like Palantir before it

• Anyone who digs in will see a company with only 30 actual customers whatever their weird Entity/Customer/customer designations are

  • Just wild that this company has raised $360M at a $3.3B valuation and only has 30 customers - is the market really there for this solution or are they forcing custom dev work into big organizations by throwing money at the problem?

  • Probably won't matter - people will see AI in blinking lights accompanied by No-code, Low-code, Machine Learning, IoT, Big Data, and Digital Transformation and buy the stock

• Performance here has been pretty muted - they've turned a profit - likely only from these weird three year deals that they get to recognize most of the revenue upfront - going from 75% YoY growth to 15% in one quarter is WILD

  • Subscription revenue decreased and I actually don't believe this is real "subscription revenue" similar to how Palantir is just doing government projects for 5 years and calling it "Subscription"

• The sales and marketing go-forward plan is middle market and partner? It just seems odd because the whole pitch is big ACV, enterprise customers and Land and Expand - is that still available at lower points in the market? Also the consolidation pitch of "We are your AI vendor" doesn't really work if you are selling piecemeal products to mid-market customers?

  • Wonder what the pitch really is for those new adopters

IPO Demand / Outcome

• I would expect it to get a ton of retail demand given people think AI is cool and saw how Palantir got this massive bump

• C3 raised at a $3.3B valuation in September 2019 which was about 25x Subscription Run-rate revenue

• I'm going to bet it goes out around a 15-20x Subscription Run-rate - $36M*4 = 144M * 15-20x = $2.1B - $2.9B valuation

• It is entirely possible that it gets a 25x multiple on some much bigger forward multiple because of their odd revenue recognition policies

• Either way I expect this to get blown up on Twitter and in the hedge fund world and pushed up pretty quickly

• I am staying away though - don't like the lack of governance, the weird rev rec, only having 30 CUSTOMERS and the likely sky-high valuation

Bull Case

• They actually deliver on this multi-industry consolidated enterprise approach to AI and become the defacto abstraction layer for AI/ML above the public clouds

• They successfully adopt mid-market, partner, and suped-up enterprise sales motions to acquire more customers and reduce risk of having so few customers

• They productize their new CRM offering into a defacto CRM and competitor to salesforce (hilariously Siebel and Oracle were super anti-salesforce and now they are all playing catchup)

• Self-driving cars, compute to the edge, 5G coalesce into this insanely difficult enterprise challenge that is ACTUALLY VALUABLE to solve - C3.ai is they only group that has the experience to do it

Bear Case

• They lose big customers at high ACV values and the stock tanks and they don't get expansions from their existing customers after the multi-year deals are complete

• They stay at a really small number of customers for a long time and spend a lot of Sales and marketing to just reacquire the same customers over and over and over

• AWS, Azure, GCP develop some standard abstraction layer for ML (kinda the way they went all in on Kubernetes) and render C3.ai not useful

  • Subsequently C3.ai probably can only get acquired by IBM because of anti-trust considerations with AI/ML tech

• The company fails to develop multiple sales motions and spends lots of money on inefficient sales

• IoT and 5G don't really change the landscape really - there are lots of ways to solve the IoT devices talking to eachother problem and C3.ai doesn't stand out