Continued poor performance by C3.ai's business has begun to weigh on the stock in recent months, despite a large amount of hype still surrounding the company.
The company's solution for enterprise search using generative AI is not unique and is likely to face soft demand relative to investor expectations.
C3.ai's business model transition is maturing, which should support growth and profitability in the coming quarters.
Despite this, C3.ai's stock still has significant downside risk due to a combination of valuation, inefficiency and weak demand for its products.
High Speed Light Streaks internet data lines
Some shine has come off $C3.ai, Inc.(AI)$
C3.ai
C3.ai continues to be something of a battleground stock given wildly diverging opinions about the company. While it is widely recognized that C3.ai has a fairly generic technology platform and its business skews towards consulting, many also consider it a leader in AI.
C3.ai was founded at a time when cloud computing was still fairly nascent and user-friendly tools for managing data at scale were limited. In response, the company developed a scalable platform for developing applications and has been trying to find a product-market fit ever since. AI is just the latest in a long line of end markets that C3.ai has tried to address, and the company subsequently has little claim to expertise in this area.
In particular, the rapid adoption of ChatGPT has caused investor excitement regarding the potential of AI in enterprise software. C3.ai is focused on applying generative AI to enterprise search. This differentiates C3.ai from many companies that are using LLMs as assistants, although this makes sense given that C3.ai primarily offers applications rather than a platform.
There are a number of problems that need to be overcome for LLMs to be effective in enterprise search through:
LLM answers are frequently random.
LLM answers are not traceable to ground truth.
LLMs can lead to data exfiltration.
Enterprise data access controls may go unenforced.
LLM-specific solutions prevent enterprises from taking advantage of the rapid innovation in LLM development.
C3.ai believes that its solution addresses these issues in a number of ways, although its solution is neither unique nor difficult to replicate. C3.ai's solution combines LLMs with a vector embedding of organizational knowledge and access controls to provide deterministic results that are traceable and do not leak IP.
While C3.ai has positioned this as a unique offering, it appears to be essentially the same approach that $Elastic N.V.(ESTC)$ has taken with its solution. Elastic brings a long history in search though, and expertise in traditional search, semantic search and hybrid approaches. C3.ai's solution may still be competitive, but it should be viewed as a fairly standard turnkey solution for customers who can't or don't want to develop their own solutions rather than some revolutionary technology.
C3.ai's Generative AI Architecture
C3.ai has reportedly seen a large increase in demand since the release of ChatGPT, but there is currently only limited evidence of this in the company's financials. C3.ai's generative AI product was released around March/April and hence it would be expected to begin impacting financials in coming quarters. The company's pipeline has reportedly doubled in the last year and the CEO has suggested the company is swamped.
C3.ai closed eight new agreements for Generative AI in the first quarter of FY2024, bringing the total to 12. The company also has a pipeline of 140 qualified opportunities, which exceeds the demand of any prior product release. The deployment of C3 Generative AI applications appears to be low-touch relative to C3.ai's other solutions, meaning that timelines are shorter (12 weeks). The initial cost is 250,000 USD, after which customers pay based on compute usage.
Investors must recognize that it is still early days for generative AI applications and that it will take time for the true costs and benefits to become apparent in many cases. LLMs as a productivity tool for developers is one of the earliest and most obvious use cases, but data indicates that Microsoft's (MSFT) Copilot product is losing a large amount of money. Users are currently only being charged 10 USD per month for the service, but it is reportedly costing Microsoft around 30 USD per user. While many users would likely be willing to pay significantly more for the product, it is not clear how much demand there would be at a 50-100 USD per month price point.
Outside generative AI, C3.ai's new business model is leading to more customer wins, with the company closing 32 agreements during the first quarter of FY2024, of which 60% were with partners. C3.ai's qualified partner opportunity increased by over 100% in the past year and its qualified pipeline with cloud providers grew by 61% QoQ.
At the end of the first quarter C3.ai had signed a total of 73 pilots, with 70 of these either moving into production, extending the pilot or in the process of negotiating a production license. C3.ai has been assuming a 70% conversion rate from pilots to production projects.
C3.ai's business model transition is leading to more deals being signed, albeit with much smaller contract values. This is helping to reduce sales cycle lengths, but the ultimate impact won't be known until more data is obtained regarding C3.ai's ability to retain customers and expand their spending.
C3.ai Deals Closed by Quarter
C3.ai Sales Cycle Length
Government, defense and aerospace continue to be areas of strength for C3.ai, along with its energy/utility business. The US Air Force is using C3.ai to conserve fuel, adding to C3.ai's existing predictive analytics for maintenance and AI-generated missile trajectory projects.
C3.ai's federal business is showing significant strength with federal bookings up 39% compared with the year ago quarter. C3.ai's Booz Allen partnership resulted in 10 closed agreements during Q1 FY2024 in the Federal Defense market. The company continues to expand its work with the US Department of Defense with new and expanded projects with the Chief Digital and AI Office, CDAO, the US Marine Corps, US Air Force, the Missile Defense Agency, and the Defense Counterintelligence Security Agency.
Financial Analysis
C3.ai's revenue has been fairly flat for almost two years now, which can be attributed both to a difficult macro environment and C3.ai's business model transition. While the macro environment is likely to remain tough going forward, the maturing business model transition should lead to a growth reacceleration in coming quarters. AI hype has increased investor expectations though, which probably means any growth reacceleration is already priced into the stock.
C3.ai Revenue
C3.ai Business Model Transition
C3.ai Annualized Growth Rate from Prior to Business Model Transition
Pilot projects continue to be a gross profit margin headwind for C3.ai and this situation is likely to persist for some time. This margin compression is likely to prove temporary though and could be considered an investment in customer acquisition rather than a drop in margins. It does raise questions about C3.ai's ability to land customers on merit alone though, as customers are currently being highly incentivized to trial the platform.
C3.ai Gross Profit Margin
C3.ai Pilots
C3.ai's business model transition, coupled with heavy investments in sales and marketing and R&D, is also weighing on operating profit margins. This situation should correct somewhat if growth reaccelerates, but this will not change the fact that C3.ai is an inefficient organization that struggles to land new customers and expand within existing customers.
C3.ai Operating Expenses
Valuation
While C3.ai's stock has pulled back significantly in recent months, the company still looks significantly overvalued based on its likely prospects. Growth should accelerate somewhat, and margins improve, as the business model transformation matures, but C3.ai still has a weak value proposition which isn't resonating with potential customers.
C3.ai Relative Valuation
Comments