C3.ai
The Enterprise-AI Pure-Play, After the Collapse

Snapshot: Executive Summary

C3.ai, Inc. is the cleanest small/mid-cap public pure-play on "enterprise generative AI" software, a positioning that once drove the stock above $30. FY26 (May 2025–April 2026) was a reckoning: revenue fell 36% to $250.3M (from $389M in FY25), the GAAP operating loss was $(498.5)M, and the non-GAAP operating loss was $(217.8)M¹. Q4 FY26 revenue was $51.6M. In response, founder Tom Siebel resumed the CEO role (May 8, 2026) and launched a restructuring targeting ~$135M of annualized cost savings. The balance sheet is the cushion: ~$575M of cash and no debt (~$3.70/share, ~40% of the market cap).

Why we downgrade Sell to Hold:

• The bear case has largely played out. The stock is down ~70% from its highs to ~$9, near its 52-week low, and now trades essentially at the Street's average target. The easy short, priced-for-perfection into a de-rating, is over.
• The cash floor is real. ~$575M of cash and no debt anchors the stock: even if the market ascribes almost no value to the operating business, the equity is worth roughly the cash. That is why the bear case ($5) sits well above zero.
• Turnaround optionality. Siebel's return plus a $135M cost-reduction plan creates a credible path to slower cash burn and, potentially, revenue stabilization, real optionality that a pure Sell no longer captures at this price.

The offset, and why this is a Hold rather than a Buy: revenue is still shrinking, the company remains deeply unprofitable, and hyperscaler-native AI (Azure, AWS Bedrock, Vertex) keeps pressuring the enterprise-AI budget. At ~3x NTM EV/Sales on a declining revenue base, the stock is not obviously cheap ex-cash. We move to Hold, 12-month PT $9 via EV/Sales scenarios (Bull $14 / Bear $5). Probability-weighted blended fair value ~$8.8, essentially at spot and at the Street average (~$8.82). Revenue stabilization and the pace of cash burn are the two data points that would move the thesis either way.

Tactical: AI is trading at ~$9, roughly at our $9 12-month target and the Street average (~$8.82). The ~$575M cash floor limits downside, but a still-shrinking, deeply unprofitable business limits upside. Rating Hold; we would turn negative again on an accelerating revenue decline, or constructive on two prints showing revenue stabilization and lower cash burn.

Investment Thesis

Rating: Hold (downgraded from Sell). Probability weights: 20% Bull / 50% Base / 30% Bear → blended fair value ~$8.8, essentially at spot and at the Street average (~$8.82). The weighting stays bearish-leaning because the fundamental trajectory is still down, but the ~$575M cash floor and the Siebel-led turnaround change the risk/reward enough that a fresh short is no longer warranted. The tell for an upgrade is two consecutive prints showing revenue stabilization; the tell for a re-downgrade is an accelerating decline.

Business Overview

C3.ai, Inc. (NYSE: AI) sells enterprise AI software, both pre-built industry applications (predictive maintenance, supply-chain optimization, customer churn prediction) and a development platform (C3 AI Suite, more recently C3 Generative AI Platform). The company is headquartered in Redwood City, California; the founder, Chairman, and CEO is Tom Siebel, who previously founded Siebel Systems (acquired by Oracle, 2006). The fiscal year ends approximately April 30.

Revenue mix and trajectory

Subscription / professional services mix shift

RPO trajectory: the leading indicator

The Siebel factor

Tom Siebel is the single most important person at C3.ai. He is the founder, chairman, CEO, and largest individual shareholder. He is highly involved in customer relationships (especially Shell and the federal/DoD book) and is the company's most credible spokesperson for the enterprise-AI thesis. Siebel-founded companies have a mixed history: Siebel Systems (acquired by Oracle for $5.85B in 2006) is widely viewed as a CRM-software triumph. But Siebel-led C3.ai has not repeated that financial outcome, revenue scale today is roughly 1/3 of what Siebel Systems achieved at peak. Key-person concentration is itself a risk to monitor.

Customer Concentration & Profitability Path

The most important section. Combining the two structurally bearish facts of C3.ai, high customer concentration and persistent negative operating margin, into a single analytical view because they reinforce each other. Concentration limits the operating-leverage story; absent leverage, the path-to-profitability cannot deliver; absent that path, the multiple compresses.

Pillar 1: Customer concentration

The Shell relationship in detail

• Origin: Long-running enterprise AI partnership starting ~2018. Shell uses C3.ai applications for predictive maintenance, energy operations optimization, supply-chain analytics.
• Commercial structure: Consumption-based pricing (announced shift from subscription, early 2024), Shell pays per-deployment, per-usage. Multi-year commitment, but the specific terms are not publicly disclosed.
• Renewal cycle: Not disclosed; estimates suggest Shell's current commitment extends through FY27 with re-negotiation expected late FY27 / early FY28.
• The single binary risk: If Shell ends the partnership, C3.ai revenue compresses ~30% nearly overnight. If Shell expands the partnership, the bull case materially strengthens. The asymmetry of this single relationship dominates the bear thesis.

The federal / DoD book

U.S. federal government (estimated ~22% of revenue) includes Air Force predictive maintenance, intelligence community analytics, and various smaller programs. The DoD book is structurally stickier than the Shell relationship (multiple contracts, slower decision-making) but is subject to budget cycles, contracting rules, and the small-business competition introduced by 8(a) procurement preferences. The bull view treats federal as durable; the bear view notes federal AI procurement is increasingly steered toward Microsoft Azure Government and AWS GovCloud rather than third-party application vendors.

Pillar 2: The profitability path

Why the path keeps getting pushed

• The consumption pricing transition. Announced early 2024, this shifted revenue recognition from subscription (recognized over contract life) to consumption (recognized as used). The transition added near-term revenue volatility and structurally reduced margin recognition pace.
• Generative AI R&D investment. Building the C3 Generative AI platform required substantial 2023-2024 R&D investment that has not yet generated proportional revenue.
• Customer-acquisition costs. Sales & marketing remains ~50%+ of revenue, a heavy spend for what is effectively a relationship-led enterprise sales motion (Tom Siebel personally involved in major accounts).
• Professional services dilution. Mentioned above, services are lower-margin than subscription and have grown faster.

What an inflection would look like

For the profitability path to inflect, we would need to see: (a) operating margin improving in two consecutive quarterly prints (not just one); (b) S&M ratio declining as a percentage of revenue (currently flat); (c) RPO bookings re-accelerating; (d) revenue growth accelerating beyond the FY26 ~9% guide. None of these are present today. The bull case requires all four to begin appearing across FY27. Without them, the multiple should rest at unprofitable-SaaS lows.

Competitive Landscape

C3.ai's market position is structurally challenged from two directions: hyperscalers building native enterprise-AI offerings, and profitable enterprise-AI peers commanding higher multiples that effectively limit C3.ai's strategic options. The competitive dynamic is the medium-term existential question.

Hyperscaler-native enterprise AI

Enterprise-AI public peers

The closest public-market comparable is Palantir (PLTR), which has demonstrated that profitable enterprise-AI software at scale can command a 60x+ EV/Sales multiple. The comparison is unflattering for C3.ai and got worse in FY26: PLTR runs multi-billion revenue growing ~35% at a positive operating margin, while C3.ai's revenue fell 36% to $250M with deeply negative margins. The enterprise-AI thesis is structurally similar but financially divergent, which is exactly why one trades at 60x sales and the other near its cash.

The disintermediation question

C3.ai's value proposition depends on enterprises wanting an industry-specific AI application layer rather than building on hyperscaler primitives directly. That proposition was strong when LLMs were exotic and integration was hard. By 2025-2026, hyperscalers have made native integration cheap and well-documented. The category of "third-party enterprise AI application platform" is increasingly questionable, the work is being done either by hyperscalers (bottom-up) or by Palantir-style high-touch deployment shops (top-down). C3.ai's mid-market enterprise software positioning is being squeezed from both ends.

Financial Health & Trends

Q4 / FY26 results (preliminary, reported May 12, 2026)

FY26 was a reckoning: revenue fell 36% and the GAAP operating loss reached $(498.5)M. The bright spots are relative, operating losses came in better than guidance, and the ~$575M cash balance leaves ample runway. The founder's return and the restructuring are the company's answer; whether they can stabilize revenue is the open question that defines the turnaround.

Three-year P&L summary

Negatives shown in (parentheses) per house convention. The story in one table: the FY26 revenue collapse (down 36%) drove operating losses sharply wider as costs lagged. FY27E assumes the restructuring narrows the loss and revenue stabilizes, the base case for the Hold, but the trajectory into FY26 is why this is not a Buy.

Capital Allocation & Runway

The runway clock

• Current balance: ~$575M cash and investments, no debt (FY26 end).
• The restructuring: ~$135M of annualized non-GAAP cost savings, initiated in Q4 FY26, aimed squarely at reducing cash burn.
• Implied runway: years, even at an elevated burn, and longer if the savings land. Solvency is not the near-term question; revenue stabilization is.
• The branches: (a) restructuring + revenue stabilization narrows the burn (the bull path); (b) continued decline consumes the cushion more slowly than a going-concern risk but caps the multiple (the bear path).

No buyback. No dividend.

C3.ai has no buyback program or dividend; the cash funds operations. Stock-based compensation remains meaningful and dilutive, but with ~$575M of cash and no debt, the balance sheet is a genuine asset, roughly 40% of the market cap is cash, which is the core of why we moved from Sell to Hold at this price.

Insider & leadership signal

Founder Tom Siebel resumed the CEO role on May 8, 2026, a strong signal of founder commitment to the turnaround, with Stephen Ehikian continuing as President. Siebel holds a substantial position; significant accelerated insider selling beyond scheduled 10b5-1 plans would be a meaningful negative tell on confidence. Monitor Form 4 filings.

The acquisition optionality

An upside branch worth flagging: C3.ai could be acquired by a larger enterprise-software incumbent (Microsoft, Oracle, SAP) or a strategic buyer, especially given the cash-rich, beaten-down valuation. A takeout premium would be an upside surprise to the Hold. We do not model it as the base case (M&A is hard to time and price), but at this valuation the downside is partly protected by both the cash floor and strategic-buyer optionality.

Valuation Overview

Why EV/Sales over P/E for C3.ai

P/E is not usable: non-GAAP EPS is negative through FY27E and GAAP EPS through at least FY29E. Forward P/E does not exist as a useful anchor. EV/Sales is the right lens because (a) revenue is the only currently-measurable scale metric, (b) net cash position is meaningful relative to enterprise value, (c) the peer set (enterprise-AI software) is uniformly valued on EV/Sales by sell-side and buy-side.

Why the multiple is what it is, and why so much is cash

C3.ai's ~3x EV/Sales is what a shrinking, deeply unprofitable business earns once the AI-narrative premium erodes. But the more important number is the cash: ~$575M against a ~$1.39B market cap. Enterprise value is only ~$0.82B, so the market already values the operating business at close to one turn of sales. The bull re-rates the multiple on a credible turnaround; the bear compresses it toward cash.

What a "fair" multiple would look like

For a business whose revenue just fell 36% with op margins deeply negative, a fair EV/Sales multiple on the operating business is low-single-digits, and the equity is anchored by the balance sheet. We use ~3.2x as the base-case multiple, which reconciles to the $9 PT once net cash is added. The bull case (revenue stabilizes, costs come out) supports ~5.7x; the bear (continued decline) collapses toward ~0.9x, i.e. the stock trading near its cash.

EV/Sales Scenario Model

FY27E revenue × EV/Sales multiple = enterprise value, plus ~$575M net cash, divided by ~155M diluted shares = 12-month price target (the forward fair value; no separate discount). Note how much of the equity value is cash: in the bear case the operating business is worth almost nothing and the stock trades toward the balance-sheet cash.

Sensitivity grid: FY27E revenue × EV/Sales multiple (PT in $)

Quick PT calculator

Risk / Reward calculator

Upcoming Catalysts

Risk Factors (Including Upside Cases)

For a Hold, the risk taxonomy is symmetric. We list both downside risks (which would push the rating back toward Sell) and upside cases (which would push it toward Buy). Given the ~$575M cash floor and the ongoing turnaround, both directions are genuinely live.

Downside risks (would re-downgrade to Sell)

• The decline accelerates. The largest single risk. Another double-digit revenue drop would signal the FY26 collapse has not bottomed and push the operating multiple toward net cash (the $5 bear case).
• The restructuring under-delivers. If the ~$135M of cost savings do not land and cash burn stays elevated, the runway shortens and the turnaround narrative loses credibility.
• Customer concentration bites. Loss or material reduction of a large customer would compress revenue further from an already-shrunken base.
• Hyperscaler squeeze intensifies. Federal and enterprise AI spend steered to Azure/AWS/Vertex/Palantir leaves less for a third-party application vendor.
• Key-person risk. Siebel's return is a positive, but concentrates the turnaround on one executive; any renewed health or succession issue would be a setback.

Upside cases (would upgrade to Buy)

• Revenue stabilizes. Two consecutive flat-to-up prints would confirm the bottom and, with the cost savings, materially narrow the loss, the clearest path to an upgrade.
• Operating loss narrows faster than modeled. If the restructuring drives the non-GAAP loss well below the FY27E path, the market re-rates the turnaround.
• Acquisition. A strategic buyer (Microsoft, Oracle, SAP) could take out a cash-rich, beaten-down name at a premium. Low probability, but a real upside branch that caps downside.
• A named new large customer. One meaningful new logo would re-establish credibility and lift the multiple off the near-cash floor.
• Product differentiation. If the C3 platform finds clear differentiation against hyperscaler-native offerings (industry-specific data + governance), the revenue mix could re-accelerate.

Bull vs Bear Debate

Technical Analysis

Trader's view

• Wide 52-week range ($7.68-$30.11); the stock sits near the low at ~$9 after the FY26 collapse.
• Key support: ~$8, close to the net-cash value (~$3.70/share cash), which sets a rough floor. A close below would imply the market pricing in going-concern-type risk that the balance sheet does not support.
• Key resistance: ~$16 (prior 2026 shelf). A move above would require evidence the turnaround is working.
• Short interest is elevated; any turnaround-progress headline can trigger a multi-day squeeze, which is part of why we no longer favor a short here.
• R/R from ~$9 to the $14 bull target with a ~$5 (near-cash) downside frames a two-sided, event-driven setup, consistent with a Hold.
• Volatility note: realized vol is high; AI is a high-beta name with binary catalyst sensitivity (earnings, restructuring updates, customer announcements).