Oracle's $988M of CEO Options Went Underwater on Day One — That's the Real Story

The grant dates are the giveaway. Magouyrk and Sicilia's options were struck at $308.46 on Sep 24, 2025 — two days after their promotion was announced and the same week Oracle's stock was still hovering near $300 on the back of the Sep 10 OpenAI Stargate deal disclosure. Ellison's package, granted in October, carried a $280 strike. The 52-week range since has been $114.50 to $322.54, meaning the options were priced within a hair of the cycle high.

Magouyrk's full FY2026 reported pay was $627.5 million — 99% in equity — with a1:94,740 pay ratio against Oracle's median global employee. The package has not yet vested (four-year time vesting on 80%, three-year performance vesting ending May 31, 2028 on the remaining 20%), so the options still have time-value. But as a skin-in-the-game signal, the structure now argues the opposite of its intent: management is more exposed to the downside of a further re-rating than it is positioned for the upside of contract conversion.

The bigger issue is what Oracle had to do to keep the AI capex story funded while the stock collapsed. In FY2026 (ended May 31, 2026), Oracle spent $55.7 billion on capital expenditures — up 163% year-on-year — and posted negative free cash flow of -$23.7 billion against $32.0 billion of operating cash flow. Capex now exceeds revenue on a trailing basis (TTM capex/revenue of 105.4%), and FCF yield sits at -7.3%.

Funding the build has required both debt and equity issuance. In FY2026, Oracle raised roughly $43 billion of debt and $5 billion of equity. In Q1 FY2027 alone, it completed a $20 billion at-the-market (ATM) equity offering. Management has stated it does not expect additional debt issuance in calendar 2026, but with capex guidance for FY2027 in the $90–95 billion range, the financing burden is plainly structural — not transitional. Net debt/EBITDA stands at 3.17x, and debt/equity at 2.34x on a TTM basis.

The compensation reset matters because it lands on a balance sheet where Oracle is the only hyperscaler-scale player with negative free cash flow, the only one with debt-to-equity above 2x, and the only one funding the build primarily with new debt and equity rather than operating cash flow. Microsoft's Azure grew 31% in Q3 FY2026 with a35% capex/revenue ratio; Amazon's AWS grew 28% in Q1 2026 with a 22% capex/revenue ratio. Oracle's Cloud Infrastructure (IaaS) grew 121% in Q1 FY2027 — but capex/revenue of 105% means each incremental dollar of IaaS revenue is being underwritten by more than a dollar of physical infrastructure build.

In parallel, supply-chain partners are in different positions. NVIDIA — Oracle's GPU supplier — runs on capex of just 2% of revenue and a 2.3% FCF yield; the company's economics are best when AI build-out accelerates, regardless of which hyperscaler wins. Dell servers and Cisco networking gear similarly capture Oracle's spend without absorbing Oracle's leverage. Equinix data-center REIT revenue grows on third-party occupancy, not Oracle's build alone.

The Sep 12, 2026 disclosure that Larry Ellison cancelled his 10b5-1 plan to sell Oracle stock — combined with the proxy's note that he had pledged 413 million shares (worth roughly $57 billion at $137) as collateral for personal term loans as of Sep 21, 2026 — points to a related, parallel signal. Ellison is doubling down rather than distributing; the share pledges have grown about 19% since 2025. That keeps the founder's exposure aligned with shareholders, but it also means any further draw-down in the share price is now a personal balance-sheet event for him as well.

The Sep 2025 OpenAI Stargate deal — $300 billion of compute purchases over five years starting in 2027, across 4.5 GW of capacity — is the single largest order behind Oracle's RPO surge. That deal and the related backlog are what the $90B FY2027 capex plan is built around. The 1–3 year question is binary: does the contracted revenue convert into cash at margins consistent with Oracle's current 63.9% gross margin and 32.4% operating margin, or does the financing burden compound faster than the conversion?