C3.ai ($AI) Stock Hub 2026: Fiscal Q1 2027 Results Due in September, the Fiscal 2026 Revenue Break and the Cost Base Behind It

01 Next scheduled event: fiscal first quarter 2027 results, expected in early September 2026

C3.ai had not announced the date of its fiscal first quarter 2027 earnings release as of August 4, 2026. The company’s newsroom carries no scheduling notice after the June 24, 2026 item, and no Form 8-K since July 17, 2026. Third-party earnings calendars point to Wednesday, September 2, 2026, but that is an estimate rather than a company statement, and readers should treat it as unconfirmed until C3.ai publishes its own notice.

The historical pattern is unusually consistent. C3.ai runs an April 30 fiscal year end, so the first quarter of fiscal 2027 closed on July 31, 2026. Its fiscal first quarter results have landed on a Wednesday in the August 31 to September 6 window in each of the last four years: September 3, 2025 for fiscal Q1 2026, September 4, 2024, September 6, 2023 and August 31, 2022. In 2026 the two Wednesdays in that neighborhood are September 2 and September 9.

The scheduling mechanics are worth knowing because C3.ai publishes a separate short release naming the date, usually one to three weeks ahead: November 10, 2025 for the December 3 report, February 12, 2026 for the February 25 report, and May 28, 2026 for the June 3 report. When that notice appears it carries the registration link and the webcast link. For the fiscal 2026 year-end call on June 3, 2026 the details were a 2:00 p.m. PT start, participant registration at register-conf.media-server.com and a webcast at edge.media-server.com. Those links are specific to that call. The place to look for the fiscal Q1 2027 equivalents is ir.c3.ai and the C3.ai newsroom.

What the September print has to answer is narrow and quantitative. Guidance issued on June 3, 2026 calls for fiscal Q1 2027 total revenue of $50.0 million to $54.0 million and a non-GAAP loss from operations of $40.5 million to $48.5 million. The midpoint of $52.0 million would be 26.0% below the $70.3 million C3.ai reported in the year-ago quarter. The company stated that it could not provide a forward-looking reconciliation to the equivalent GAAP measure, so the GAAP operating loss for the quarter is not part of the guidance.

02 Executive summary

C3.ai sells enterprise AI application software: a development and runtime platform, a catalog of prebuilt industry applications, and generative and agentic AI products layered on top. It has been publicly traded since December 2020 and has never reported a profitable year. What changed in fiscal 2026 is the direction of the top line. Revenue fell from $389.1 million to $250.3 million, a decline of $138.8 million or 35.7%, and the fall was concentrated in the two quarters after a sales reorganization began.

Two numbers frame the rest of the story. The first is the gross margin. GAAP gross margin dropped from 61% in fiscal 2025 to 31% in fiscal 2026, because the cost of subscription revenue rose from $143.8 million to $166.3 million while subscription revenue fell from $327.6 million to $227.1 million. On a non-GAAP basis, which excludes stock compensation and payroll taxes on it, gross margin went from 70% to 46%. Either way, this is not a software gross margin, and it is the clearest evidence that a large part of C3.ai’s cost of revenue is fixed delivery and hosting capacity rather than variable cost of goods.

The second is stock-based compensation. C3.ai recorded $263.7 million of stock compensation expense in fiscal 2026, excluding a further $4.8 million booked inside restructuring. Against $250.3 million of revenue, that is 105.4% of revenue. The gap between the GAAP operating loss of $498.5 million and the non-GAAP operating loss of $217.8 million is almost entirely that item. Investors who read only the non-GAAP figure are reading a number that treats more than a dollar of equity issued for every dollar of revenue collected as if it were free.

On the other side of the ledger, the balance sheet is genuinely uncrowded. There are no borrowings. Cash, cash equivalents and marketable securities stood at $575.4 million on April 30, 2026 and at $673 million on June 3, 2026 after Thomas Siebel exercised a stock option covering 6,166,667 shares at $11.16 and the company collected the exercise proceeds. Free cash flow, a non-GAAP measure the company defines as operating cash flow less purchases of property and equipment, was negative $192.1 million in fiscal 2026. At that rate the June cash position covers roughly three and a half years.

The governance picture changed twice inside a year. Siebel handed the chief executive role to Stephen Ehikian in September 2025 while remaining Executive Chairman, then took it back on May 8, 2026, with Ehikian moving to President. Siebel’s Class B shares carry 50 votes each; as of June 10, 2026 he and related entities held approximately 49.6% of total voting power. And the market’s positioning is extreme: 32.43% of the float was reported short, and the consensus analyst target price of $8.82 sits below the August 3 close of $9.74.

03 Market Data And Peer Comparison

Price and performance figures below are based on the completed session of Friday, August 7, 2026. Float, ownership, short interest, average volume and the consensus target are from Finviz, pulled on the same date. Company financial figures come from SEC filings and company releases, each carrying its own reference date.

Peer comparison, all figures at the August 7, 2026 close

One row deserves reading twice. The consensus target of $8.82 sits below the August 7 close of $10.22. A sell-side average below the market price is unusual and it is the clearest available summary of how third-party estimates have moved after the fiscal 2026 result. It remains an aggregate of estimates, not a forecast made here.

On analyst coverage the honest position is a narrow one. The consensus target above is a Finviz aggregate of third-party estimates pulled on August 7, 2026. Individual houses, ratings and note dates were not verified for this update, so no coverage table is presented. A consensus figure without named notes behind it is a market-data point, not research, and it is neither a company figure nor a Merlintrader forecast.

04 Verified developments up to August 4, 2026

05 Five quarters of revenue, and where the break happened

Total revenue by quarter, GAAP, in millions of dollars

Source: C3.ai quarterly earnings press releases filed as exhibit 99.1 to Forms 8-K on May 28, 2025, September 3, 2025, December 3, 2025, February 25, 2026 and June 3, 2026. Bars are scaled to the $108.7 million high. The four fiscal 2026 quarters sum to $250.3 million, matching the audited full-year figure.

The shape matters more than any single quarter. Revenue was still growing into the April 2025 year end, then fell 35% sequentially in the first quarter of fiscal 2026 and never recovered. The company’s own explanations, given in the September 3, 2025 release, were a disruptive sales and services reorganization and the founder’s absence from the sales process during a period of health problems. By the February 2026 report a second, larger reorganization was under way. Quarterly revenue has now spent two consecutive quarters near $52 million, which is roughly where fiscal 2027 guidance places every quarter of the coming year.

Full-year revenue, GAAP, in millions of dollars

Source: consolidated statements of operations in the Form 10-K for the fiscal year ended April 30, 2026, filed June 24, 2026, and the fiscal 2027 guidance table in the June 3, 2026 earnings release. Bars are scaled to the $389.1 million fiscal 2025 peak. Guidance is company guidance, not a forecast.

Fiscal 2027 guidance implies another decline. The $210 million to $240 million range is 16.1% to 4.1% below fiscal 2026, with a midpoint of $225 million, or minus 10.1%. That is the first thing to hold in mind when reading any argument about the shares being cheap on a sales multiple: the denominator is still shrinking, and the company itself expects it to shrink for another twelve months.

06 What C3.ai actually sells

C3.ai describes itself as the enterprise AI application software company, and the product line has three layers. The C3 Agentic AI Platform is the development and runtime environment: a model-driven architecture that lets teams build, deploy and operate AI applications across data sources without hand-coding the integration layer. C3 AI Studio is the integrated development environment on top of it, and C3 Code, announced April 6, 2026, extends that with agentic development, so a user can describe a use case in natural language and have the application designed, built and deployed. The platform is cloud-agnostic: Azure, AWS, Google Cloud, on-premises, or FedRAMP-authorized government clouds.

Above the platform sit the C3 AI Applications, prebuilt industry-specific software. The named products include C3 AI Reliability, Production Schedule Optimization, Demand Planning, Process Optimization, Inventory Optimization and Sourcing Optimization. Reliability, the predictive maintenance application, is the flagship and is the product behind the Shell relationship. There is also a set of defense and intelligence applications sold to the U.S. Department of War, including branches such as the Air Force, the Missile Defense Agency and the department’s Chief Digital and Artificial Intelligence Office. C3 Generative AI is a suite of domain-specific generative offerings.

The commercial model, and why the numbers move the way they do

Almost every C3.ai relationship starts as an initial production deployment, or IPD, formerly called a pilot: a paid, short, pre-configured deployment lasting roughly one to two quarters. After that, a customer either moves to a monthly fee with consumption charges metered on vCPU and vGPU hours, or signs a multi-period commitment. That structure has two consequences. It makes the top of the funnel measurable, and it makes the conversion rate the single most important operating statistic in the business.

C3.ai executed 71 IPD agreements in fiscal 2026, against 174 in fiscal 2025 and 123 in fiscal 2024. The 10-K attributes the drop to a deliberate shift toward engagements with a higher probability of conversion. That is a defensible strategy and it is also a smaller funnel. In the fourth quarter the company signed 28 agreements in total, including nine new IPDs and seven IPD conversions, and stated plainly that bookings were lower than expected.

Subscription versus professional services, and what happened to the mix

The split needs care, because C3.ai reports a third, blended figure. In fiscal 2026 subscription and prioritized engineering services revenue combined was $245.1 million, or 98% of total revenue. That is the number the company leads with, and it is not wrong, but the two components behave differently. Subscription revenue is licenses, software-as-a-service, standby support from the Center of Excellence, initial production deployments and runtime and hosting fees. Prioritized engineering services is something else: the customer pays a negotiated fee to have C3.ai pull a planned roadmap feature forward, and the resulting code becomes part of the core product available to every subscriber. It is recognized as professional services under ASC 606, over the period the development is completed.

Prioritized engineering fell from $43.0 million to $18.0 million, and in the fourth quarter alone from $17.0 million to $2.1 million. That single line accounts for $25.0 million of the $138.8 million annual revenue decline and for $14.9 million of the $57.1 million fourth-quarter decline. Whether one reads that as customers pausing discretionary roadmap purchases or as a legitimate cleanup of a lumpy revenue category, it is the part of the mix that moved fastest, and it is the reason the subscription share of revenue rose to 91% while total revenue collapsed.

Where the revenue comes from geographically, and how concentrated it is

North America produced $225.8 million of fiscal 2026 revenue, Europe, the Middle East and Africa $22.0 million, Asia Pacific $0.7 million and the rest of the world $1.8 million. The United States alone was 89% of revenue, up from 86% in each of the two prior years, so the international business shrank faster than the domestic one. Customer concentration improved on paper: one customer entity accounted for 14% of fiscal 2026 revenue, against two entities at 19% and 12% in fiscal 2025 and two at 27% and 14% in fiscal 2024. At the accounts receivable level, though, concentration is still high, with two entities at 25% and 11% of receivables on April 30, 2026.

07 Financial position: the fiscal 2026 baseline going into the September print

All figures from the consolidated financial statements in the Form 10-K for the fiscal year ended April 30, 2026 and the reconciliation tables in the June 3, 2026 earnings release. Non-GAAP measures exclude restructuring expense, stock-based compensation and employer payroll taxes on stock-based compensation; free cash flow is operating cash flow less purchases of property and equipment. Fiscal 2024 free cash flow is computed from the cash flow statement as operating cash flow of $(62.362) million less $25.256 million of property and equipment purchases and $2.750 million of capitalized software.

The gross margin is the number to watch on September 2

Cost of subscription revenue was $166.3 million in fiscal 2026, up from $143.8 million, while subscription revenue fell by $100.5 million. As a share of total revenue, cost of subscription went from 37% to 66%. This is the mechanical reason the GAAP gross margin halved, and it says something structural: a meaningful part of C3.ai’s delivery cost does not scale down with revenue in the short run. Part of it is people, and part of it is contracted cloud capacity. The 10-K discloses $379.8 million of remaining purchase commitments for cloud hosting and associated services, plus $61.6 million for professional services, falling due over the next one to four years. The company incurred $104.6 million of costs under those arrangements in fiscal 2026 and $117.1 million in fiscal 2025.

That commitment is the single most under-discussed line in the filing. It is not debt, it does not appear on the balance sheet, and it does not go away if revenue falls further. The 30% reduction in annualized non-employee costs the company is targeting by the second half of fiscal 2027 is where any relief has to come from, and the June quarter is too early to show much of it.

What to watch in the fiscal Q1 2027 print

• Revenue against the $50.0 million to $54.0 million guidance. A print at or above the top of the range would be the first time in five quarters the trend has bent upward.
• GAAP gross margin. It was 22% in the fourth quarter and 17% in the third. Anything below 30% means the cost base has not yet responded to the restructuring.
• Non-GAAP loss from operations against the $40.5 million to $48.5 million guidance, and the GAAP figure alongside it, which the company does not guide.
• Remaining performance obligations. $203.1 million at year end, with $129.3 million expected to convert within twelve months. A further decline would mean the restructuring has not stopped the bookings slide.
• Cash and investments, which will be the first quarter to include the June option exercise proceeds.
• Stock-based compensation, which needs to fall in absolute terms for the equity story to change, not just as a percentage of a recovering revenue line.
• Initial production deployment count and conversions, the leading indicator that turns into subscription revenue two to four quarters later.

08 Stock-based compensation is the central accounting fact of this company

Stock-based compensation expense as a percentage of total revenue

Computed from the stock-based compensation table in the Form 10-K for the fiscal year ended April 30, 2026: $215.761 million on $310.582 million of revenue in fiscal 2024, $230.988 million on $389.056 million in fiscal 2025 and $263.715 million on $250.268 million in fiscal 2026. Figures exclude the additional $4.817 million of stock compensation recorded inside the fiscal 2026 restructuring charge; including it, total stock compensation was $268.532 million, or 107.3% of revenue. Bars are scaled to the 105.4% high.

The absolute expense rose 14.2% in a year when revenue fell 35.7%. Broken down by line, fiscal 2026 stock compensation was $95.5 million in sales and marketing, $79.2 million in research and development, $51.5 million in general and administrative, $35.6 million in cost of subscription and $1.9 million in cost of professional services. It is not concentrated in one place; it is embedded in every operating line, which is why the non-GAAP gross margin of 46% is fifteen points above the GAAP figure of 31%.

The commitment is also forward-looking. As of April 30, 2026 there was $383.7 million of unrecognized stock compensation on outstanding restricted stock units, to be recognized over a weighted-average 3.3 years, plus $73.3 million on stock options over 2.7 years and $6.8 million on the employee stock purchase plan. That is roughly $464 million of expense already granted and still to run through the income statement, against a company guiding to $210 million to $240 million of revenue in fiscal 2027. Even a complete halt to new grants would leave the GAAP loss heavily inflated for three more years.

One nuance cuts the other way. The company paid its fiscal 2025 annual bonuses in fully vested restricted stock units rather than cash, issuing 951,779 Class A shares during fiscal 2026 for that purpose, after 751,927 shares the year before. That preserves cash at the cost of shares. Readers deciding how much weight to give the non-GAAP figures should decide first whether they regard that trade as free.

09 Cash, burn and the runway arithmetic

Cash, cash equivalents and marketable securities, in millions of dollars

Source: the April 30, 2025 and April 30, 2026 balance sheets in the Form 10-K, the cash balance disclosed in each quarterly earnings release, and the June 3, 2026 figure stated in the fiscal 2026 results release. Bars are scaled to the $742.7 million April 2025 high. The June 3 step-up reflects the proceeds of Thomas Siebel’s exercise of a stock option over 6,166,667 shares at $11.16, which was cash paid to the company.

Read left to right, the four quarter-end bars show $167.2 million of cash and investments consumed over fiscal 2026, an average of $41.8 million a quarter, with the largest single-quarter drawdown of $53.1 million in the January quarter. Free cash flow, the non-GAAP measure, was worse than the balance movement at negative $192.1 million for the year, because interest income and the maturity of securities offset part of the operating outflow. In the fourth quarter alone free cash flow was negative $54.8 million.

The runway calculation is simple and should be stated plainly. Against the $673 million disclosed on June 3, 2026, a continuation of the fiscal 2026 free cash outflow of $192.1 million a year implies roughly three and a half years before the balance is exhausted, with no borrowings to service in the meantime. If the restructuring delivers and the burn narrows toward the guided non-GAAP operating loss range of $128 million to $160 million, the runway lengthens. If revenue undershoots and the cloud commitments continue at the fiscal 2026 rate of $104.6 million a year, it shortens. This is not a company facing an imminent financing decision, and that is the strongest single fact in its favor.

10 Share count, dilution and the option overhang

Shares outstanding at each fiscal year end, Class A and Class B combined, in millions

Source: the consolidated statements of stockholders’ equity in the Form 10-K for the fiscal year ended April 30, 2026 for the four year-end figures, and the Form 10-K cover page for the June 10, 2026 count of 151,949,285 Class A and 3,499,992 Class B shares. Bars are scaled to the 155.4 million high.

Share count has risen 36.4% in three years, from 113.9 million to 155.4 million, with no acquisition and no capital raise. All of it is equity compensation and option exercises. Fiscal 2026 alone added 14.1 million shares at the year-end measuring point: 10.6 million from vesting restricted stock units, 2.6 million from option exercises and 0.9 million from the employee stock purchase plan. The additional 7.4 million shares between April 30 and June 10, 2026 are mostly Siebel’s single 6,166,667-share option exercise.

There is a genuine offset here that is easy to miss. With the shares at $9.74 and the option pool struck at a weighted average of $15.44, the entire 32.4 million share option book is under water, and the 10-K states the aggregate intrinsic value as zero. Those options will not be exercised at current prices, and 11.6 million restricted stock units were forfeited in fiscal 2026, more than half the number granted the year before. Dilution from this pool is therefore price-dependent rather than automatic: it accelerates if the shares recover and stalls if they do not.

11 Backlog, guidance and the difference between contracted and expected revenue

C3.ai does not report backlog in the defense-contractor sense. Its disclosed forward metric is remaining performance obligations: contracted future revenue not yet recognized, comprising deferred revenue plus non-cancellable committed amounts. Contracts a customer can cancel without significant penalty are excluded, and monthly usage-based runtime and hosting charges are excluded entirely.

Two qualifications belong next to that headline. First, $50.7 million of the $203.1 million consists of non-cancellable commitments where the actual product selection and quantities are still to be determined; the money is committed, the deliverable is not. Second, the company itself warns in the 10-K that RPO “may not accurately reflect future revenue growth,” because of pay-as-you-go pricing, renewal timing and the timing of IPD conversions.

Setting RPO against guidance gives a useful sanity check. Fiscal 2027 guidance is $210 million to $240 million. Of that, $129.3 million is already contracted and expected to convert within twelve months. That leaves roughly $81 million to $111 million to come from new business, renewals and usage during the year, out of a company that signed 28 agreements in its most recent quarter. The guidance is not obviously heroic, and it is not obviously conservative either.

The restructuring is the bridge between those two lines. The February 24, 2026 plan cut 26% of the global workforce and targets an approximately 30% reduction in annualized non-employee costs, and on May 12, 2026 the company quantified the whole program at approximately $135 million in annualized non-GAAP cost savings, with a similar reduction in cash burn. Workforce actions are substantially complete; the non-employee reductions are expected to be substantially realized starting in the second half of fiscal 2027, which means the November 2026 and March 2027 quarters, not the September print.

12 The partner network, the Baker Hughes shift and the Shell renewal

C3.ai sells through four partner categories, and the 10-K is explicit that the partner ecosystem “serves as a primary channel for customer acquisition and expansion.” In the second quarter of fiscal 2026 the company disclosed that 89% of total bookings were driven with or through partners. For a company with 764 employees and a direct sales force that has been reorganized twice in eighteen months, that dependency is the business model, not a detail.

Baker Hughes: from exclusive channel to one partner among several

The Baker Hughes relationship began in June 2019 as a strategic collaboration under which Baker Hughes was the exclusive channel partner and reseller of C3 AI software in oil and gas, and a non-exclusive reseller elsewhere. It was revised in January 2023 and again in April 2025. The 10-K now states the position in one sentence: “Baker Hughes is no longer C3 AI’s exclusive reseller.” Baker Hughes still appears in the list of industry partners and in the alliance program, but it sits beside Booz Allen Hamilton rather than above everyone.

This matters historically as well as commercially. The single sentence still at issue in the Reckstin securities class action, after the March 12, 2026 dismissal of the other claims, concerns exactly this arrangement: the statement in the December 2020 IPO registration statement that the company 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.” Every other allegation about access to the Baker Hughes sales force and the wider strategic partnership representations was dismissed and is no longer part of the case.

Shell: the largest disclosed commercial expansion of 2026

On June 4, 2026 C3.ai announced that Shell Information Technology International B.V. is extending its collaboration under a new multi-year agreement. The relationship dates to 2018 and covers an enterprise-scale predictive maintenance program monitoring more than 13,000 pieces of equipment. The extension broadens C3 AI Reliability beyond equipment anomaly detection into AI agent-based root cause analysis and remediation. It runs on the C3 Agentic AI Platform deployed on Microsoft Azure.

13 Federal and defense: the fastest-growing part of the book

The 10-K states that the U.S. federal government, including defense, the intelligence community and civilian agencies, “has emerged as a significant and growing segment of our business.” C3.ai does not break out federal revenue as a reported segment, so the only quantified evidence is in the bookings disclosures, and those are the figures to hold on to.

The credentials behind that pipeline are real and dated. C3.ai achieved FedRAMP authorization in December 2025, and the 10-K lists certifications under ISO 27001, ISO 27017, SOC 1 Type II, SOC 2 Type II, HIPAA and NIST 800-171. In December 2025 the U.S. Department of Health and Human Services selected C3.ai as an enterprise AI platform, and the Army selected the company to deliver a contested-logistics solution for combat operations. In May 2025 the ceiling on the Air Force Rapid Sustainment Office award was increased to $450 million, and in June 2025 a $13 million task order expanded the predictive maintenance program across the Air Force fleet. In June 2025 C3.ai listed its solutions in the AWS Marketplace for the AWS Secret Region.

The distinction that matters here is the same one that matters in every defense-adjacent story. A contract ceiling is not revenue. The $450 million Air Force figure is the maximum value of an award vehicle; the $13 million task order is a funded order against it. Ceilings are announced, funded orders are what shows up in the accounts, and C3.ai’s federal business is disclosed in growth percentages on an undisclosed base. A doubling of a small number is still a small number, and the fourth quarter, in which the company chose not to publish any bookings detail at all, is the reason to hold that thought.

There is also a specific fragility to a federal-weighted pipeline. The 10-K names U.S. federal government shutdowns as a macroeconomic risk twice, and government customers can reduce, delay or terminate for convenience. The upside case for the shares leans heavily on federal, and so does one of the clearest downside scenarios.

14 Management, governance and control

Two chief executive changes in nine months is a lot of turbulence for any company, and the sequence is public. In July 2025 Siebel and the board announced a search for a successor; on September 3, 2025, the same day as the fiscal Q1 2026 results, Ehikian was appointed; on May 8, 2026 Siebel took the role back. In the May 12 release Siebel wrote that his vision “remains impaired” but is improving and that the health issues encountered during 2025 are largely resolved. In the June 3 release he described the recent sales performance as “entirely unacceptable, to the point of surreal.” The candor is unusual. It is also, for a reader trying to model the business, a description of an execution problem that management has not claimed to have solved yet.

Voting control

C3.ai has a dual-class structure. Class A carries one vote per share; Class B carries 50 votes per share. There are 3,499,992 Class B shares outstanding out of 3,500,000 authorized. As of June 10, 2026, the 10-K discloses that Siebel and related entities beneficially owned approximately 87.8% of the Class B stock and approximately 14.9% of the outstanding Class A stock, giving him beneficial ownership of capital stock representing approximately 49.6% of total voting power. In practice, no matter affecting the company passes without him. That is a governance fact to be weighed rather than judged: it removes activist and takeover pressure, and it also removes the mechanism by which shareholders would ordinarily force a change of course.

Litigation, stated precisely

• Reckstin Family Trust v. C3.ai, filed March 2022. On March 12, 2026 the court dismissed three of five causes of action, two of them with prejudice, including all Section 10(b) and 20(a) claims and all Section 20A insider-trading claims. What remains is a Sections 11 and 15 claim against the company and three individuals about one sentence in the December 2020 IPO registration statement.
• Liggett v. C3.ai, filed August 2025, alleging misleading statements between February and July 2025 about Siebel’s health. Dismissed in its entirety on July 14, 2026.
• Nine shareholder derivative actions across two groups, all stayed pending the outcome of the underlying securities cases. The company is a nominal defendant in each.
• C3.ai v. Enel Global Services, filed February 27, 2024 in the Court of Rome, alleging misappropriation of trade secrets under the Italian Industrial Property Code and breach of contract, seeking compensatory damages of 2.1 billion euros. C3.ai also filed a report of criminal misconduct with Italian law enforcement. On December 23, 2024 Enel initiated two suits in Rome claiming ownership of two C3.ai patent applications at the European Patent Office; motions to stay prosecution of those applications were granted. The 10-K treats any recovery as a gain contingency, to be recognized only when realized.

The Enel claim is the item most often misread. A 2.1 billion euro demand against a company with a $1.5 billion market capitalization is arithmetically eye-catching, but it is a plaintiff’s demand in a foreign court, it is unquantifiable as to timing and outcome, and under U.S. accounting rules none of it can be recognized until an award is realized or realizable. It is not an asset on the balance sheet and it should not be treated as one.

15 Ownership, Short Interest And Retail Sentiment

A short interest above 30% of float in a company with a $1.5 billion market capitalization and 6.3 million shares of daily volume creates a mechanical feature of the tape that has nothing to do with C3.ai’s software. It takes about seven full sessions of average volume to cover the position. Any news that forces a reappraisal, positive or negative, moves the price further and faster than the news alone would justify. The reverse is also true: a large short base is a standing bid under the shares, and it means bad news is, to a degree, pre-positioned.

Insider activity in 2026 has been mixed in a way that resists a simple reading. Siebel’s June 1 transaction was an exercise of a stock option over 6,166,667 shares at $11.16, an option that would have expired in August 2030, followed two days later by a gift of 6,182,053 shares to the Siebel Living Trust, of which he is trustee. The company received the exercise proceeds, which is why the June 3 cash figure stepped up. A stock option exercise well ahead of expiry is not the same signal as an open-market purchase, and the company’s own release describes it as a purchase; both descriptions are defensible, but the Form 4 transaction code is the precise record.

Retail sentiment around $AI is loud and polarized, and it should be read as what it is: non-professional opinion, not research. The bull thread on retail boards runs on the federal contract announcements, the FedRAMP authorization and the founder’s return. The bear thread runs on the revenue decline, the stock compensation and a long record of announced partnerships that did not visibly convert into revenue. Neither side, as a rule, works through the filings. The consensus analyst target price of $8.82, below the market, is the closest thing to a professional aggregate view, and it is itself an average of estimates that have been cut repeatedly.

The block below is a snapshot of the Stocktwits flow, with its date. These are opinions of retail traders and non-professional investors, not analyst research, and they measure attention and how one-sided positioning has become rather than anything about the business.

16 Catalysts to monitor

17 The two cases, stated as fairly as possible

• No borrowings and $673 million of cash and investments as of June 3, 2026, against a market capitalization of $1.51 billion. Enterprise value is roughly $840 million, or about 3.4 times fiscal 2026 revenue.
• A cost program already executed on the employee side: 26% of the workforce gone, headcount at 764, and roughly $135 million of annualized non-GAAP savings targeted, most of the non-employee half still to come.
• A federal and defense pipeline that grew 89% then 134% year over year in consecutive quarters, backed by FedRAMP authorization, NIST 800-171, an Air Force award vehicle raised to a $450 million ceiling, and named wins at HHS, the Army, USDA, the Department of Energy and NATO.
• Partner leverage that is unusual for a company this size: 89% of second-quarter bookings came with or through partners including Microsoft, AWS, Google Cloud, McKinsey and Booz Allen Hamilton.
• A flagship commercial reference renewed in June 2026: Shell, running since 2018 across more than 13,000 pieces of equipment, extending into agentic root cause analysis.
• The founder, who built and sold Siebel Systems, is back in operational control with 49.6% of the vote and every incentive to fix the sales motion himself.
• The option pool is entirely out of the money and 11.6 million restricted units were forfeited in fiscal 2026, so near-term dilution from equity awards is running slower than the grant tables suggest.

• Revenue fell 35.7% in fiscal 2026 and the company guides to a further decline of 4% to 16% in fiscal 2027. Five years after the IPO, the growth thesis is being defended with a shrinking top line.
• GAAP gross margin of 31%, and 22% in the fourth quarter, is not a software margin. Cost of subscription rose in absolute terms while subscription revenue fell by $100.5 million.
• Stock compensation of $263.7 million was 105.4% of revenue, with a further $464 million of unrecognized expense already granted and running for up to 3.3 years.
• The GAAP operating loss was 199% of revenue. The non-GAAP loss of $217.8 million is the number management guides against, and the difference is almost entirely equity issued to employees.
• $379.8 million of non-cancellable cloud hosting commitments plus $61.6 million of professional services commitments fall due over one to four years, off balance sheet and independent of how revenue develops.
• RPO fell 13.6% to $203.1 million, initial production deployments fell from 174 to 71, and the fourth quarter release contained no bookings detail at all.
• Two chief executive changes in nine months, a director on leave, and a founder who has publicly described his own vision as impaired and recent sales performance as surreal.
• 32.43% short interest, a consensus target below the market price, and a share count up 36.4% in three years with no acquisition to show for it.

18 Scenario framework

The table below is an analytical framework for organizing what would have to be true in each case. It is not a forecast, not a target price and not advice.

19 Merlintrader bottom line

C3.ai enters the September report as a company with a clean balance sheet and a broken income statement. The cash is real: $673 million as of June 3, 2026, no borrowings, no maturity wall, roughly three and a half years of runway at the fiscal 2026 burn rate. The revenue is real too, and it is going the wrong way: down 35.7% in fiscal 2026, guided down again in fiscal 2027, with a GAAP gross margin of 31% that does not look like software and stock compensation at 105.4% of revenue that makes every non-GAAP figure a partial account.

The evidence that something is changing is concentrated in two places. The federal and defense bookings grew 89% and then 134% year over year in consecutive quarters, on an undisclosed base, and the restructuring has already removed 26% of the workforce with roughly $135 million of annualized savings targeted. The evidence that it is not yet changing is concentrated in one place: the fourth quarter, when revenue landed at $51.6 million, bookings came in below plan, and the company published no business highlights at all.

The fiscal first quarter 2027 report, expected in early September and not yet dated by the company, is where those two threads meet. Guidance of $50.0 million to $54.0 million sets a low bar, and the specific things that would show a change of direction are narrow: revenue at the top of the range, GAAP gross margin above 30%, RPO no longer falling, and stock compensation declining in absolute dollars rather than as a percentage of a moving denominator. Until then, the numbers in the filings are the only ones that exist.

Related Research On Merlintrader

• BigBear.ai Holdings (NYSE: $BBAI) stock hub – defense AI, backlog and the authorized-share overhang, a useful contrast in how a much smaller AI vendor reports federal work.
• Palladyne AI Corp. (Nasdaq: $PDYN) stock hub – a pre-revenue-scale AI software company with the same partner-led, defense-weighted go-to-market.
• Small-cap AI valuation reset 2026 – the KPIs that separate AI companies with revenue from AI companies with announcements.
• Top 10 AI mid and small cap stocks for 2026 – where $AI sits in the wider small and mid cap AI universe.
• Free catalyst calendar – every dated catalyst tracked in one filterable sheet.