Siebel keeps exercising $3.90 options into C3.ai's lows because a 2024 plan tells him toC3.ai, Inc. · NYSE: AI
- What he did
- 13 Oct 2026: exercised 435,363 options at $3.90 (code M) and sold the same 435,363 shares (code S) at a $11.8179 weighted average, range $11.565–$12.0123. Gross proceeds about $5.15m; exercise cost about $1.70m. Direct holding after: 722,362. Remaining options: 1,533,585 at $3.90, fully vested, expiring 27 Nov 2028. Indirect holdings unchanged at 8,447,956, of which 7,105,832 sit in the Siebel Living Trust and 1,237,115 in Siebel Legacy Investments II LLC. Both legs made under a Rule 10b5-1 plan dated 20 September 2024.
- Price context
- C3.ai closed $11.67 on 14 Oct 2026, having set 52-week lows at $13.18 and then $12.56 during the year.
Consensus read
The founder-CEO is selling at the lows, which is either capitulation or, more charitably, irrelevant: it is a 10b5-1 plan, so it carries no information at all and should be ignored.
Contrarian read
Both halves of that are wrong. It is not capitulation, and it is not informationless. The plan is destroying real money in a way its author plainly did not intend, and that is itself the finding: a two-year-old share-count plan is force-exercising options with two more years of life at the worst price in the stock's history.
Evidence that discriminates
The gap between the option's expiry and the exercise date. These options expire 27 November 2028 — twenty-five months after this exercise — and are already fully vested. Exercising a vested, deep-in-the-money option two years early throws away its entire remaining time value for nothing; no discretionary holder does this, and it is the single clearest tell that no human made this decision on 13 October. The second discriminator is the plan's date: 20 September 2024. A plan written then was written against a materially higher share price, and plans of this kind specify share counts, not dollar proceeds — so as the stock fell, the plan kept selling the same number of shares for steadily less money, and kept consuming option inventory at the same rate. That combination distinguishes the mechanical reading from both the capitulation reading (which needs a recent decision) and the "ignore it" reading (which needs the mechanism to be costless, and it is not).
Inference: I do not have the AI share price on 20 Sep 2024 from a source checked here, so the "written against a higher price" claim rests on the stock having set 52-week lows during 2026 at $13.18 and $12.56 and now trading below both. Treat the direction as established and the magnitude as unverified.
What would falsify me
A Form 4 or 8-K showing the plan was amended or re-adopted in 2026, or that it contains a price floor — either would mean the exercises reflect a current decision. Equally: Siebel terminating the plan before the next scheduled tranche, which would show discretion was available and used.
Pattern vs. his own historyHis indirect position — 8.45m shares across the Living Trust, First Virtual Holdings, HS ET DE Investments and two Legacy LLCs — has not moved. Only the option inventory is draining, and only on the plan's schedule.