Oracle's Stock Crash Has Cost Larry Ellison $213 Billion in 10 Months. Should Investors Buy the Dip?

Oracle’s shares are down over 50% from peak levels, wiping roughly $213B from Larry Ellison’s fortune, as investors question the returns on aggressive AI data center spending. Capital expenditures rose to $21B in fiscal 2026 (from ~ $7B prior year) with guidance for more than $25B in fiscal 2027, while S&P Global downgraded Oracle to BBB- (one notch above junk), likely increasing borrowing costs. Despite a record $638B in remaining performance obligations and sustained AI cloud demand, the credit hit and capex intensity are driving risk-off sentiment around near-term profitability.

Analysis

This is less a fundamental break than a financing-quality reset: the market is now forcing Oracle to prove that AI infra can earn back capital faster than its debt cost. That matters because in capex-heavy software/cloud models, the first derivative to watch is not revenue growth but spread behavior — if the bond market keeps pricing ORCL as a quasi-high-yield credit, equity upside gets capped even if bookings stay strong. In the next few weeks, the stock can stay weak simply because levered growth names often de-rate before any operating data changes.

Relative winners are the hyperscalers with stronger balance sheets and cheaper incremental funding (AMZN, MSFT, GOOGL). They can absorb AI demand without forcing the market to underwrite balance-sheet risk, which should widen the valuation gap between “platforms” and “build-at-all-costs” cloud names. NVDA is a mixed read: Oracle’s spend still supports near-term hardware demand, but if creditors start leaning on capex discipline across the sector, the market may rotate from infrastructure-beta into quality cash-flow compounders.

The contrarian miss is that backlog alone does not solve returns on capital. If utilization ramps faster than feared, ORCL could stabilize sharply over 1-3 months, but the falsifier is simple: another credit-negative update, wider spreads, or any sign capex stays above cash generation through the next print. Over 6-18 months, the real risk is that Oracle ends up with expensive capacity and only average utilization, which would pressure both equity multiple and debt terms.

Key Decisions for Investors

• Short ORCL on rallies or buy 3-6 month put spreads; thesis is valuation de-rating continues until management proves capex converts to FCF. Best risk/reward if bond spreads keep widening.
• Pair trade: long MSFT or AMZN vs short ORCL over 1-3 months. Favor the names with lower funding risk and less downside if AI monetization slows.
• Avoid chasing NVDA off this headline; ORCL capex still supports GPU demand near term, but any sector-wide capex discipline would be a headwind. Reassess only if ORCL guides down data-center spend.
• Watch ORCL credit spreads and next capex guide as the key falsifier. If spreads tighten materially and utilization commentary improves, cover shorts quickly.