Oracle: The $28B CapEx Illusion And The Reality Of Software Cannibalization

Oracle reported strong cloud growth and raised its revenue outlook, but overall underlying revenue growth remained minimal as declining on-premises software sales offset cloud gains. Annual capex of $90-95 billion has driven free cash flow negative and compressed margins as high-margin software is replaced by capital-intensive infrastructure. The investment program is being partly funded through a $20 billion secondary share offering, creating dilution risk for shareholders.

Analysis

The key equity issue is not cloud growth but the deterioration in incremental economics: Oracle is exchanging recurring, high-contribution license/support revenue for infrastructure revenue whose returns depend on utilization, power availability, GPU procurement, and customer contract duration. That raises the probability that consensus is capitalizing a temporary AI capacity shortage as a durable software-like earnings stream. The financing choice also creates a reflexive problem: dilution and weaker free-cash-flow conversion reduce the valuation support historically provided by buybacks and capital returns.

Over the next 1-3 months, the likely catalyst is a reset to estimates if management cannot demonstrate that new capacity is generating attractive returns on invested capital rather than simply backlog conversion. Investors should focus on cloud gross margin, capex-to-revenue, deferred revenue/backlog conversion, and whether remaining performance obligations are tied to concentrated counterparties with cancellation or ramp-risk provisions. A further equity raise, slower-than-expected data-center energization, or AI customer renegotiations would pressure both EPS estimates and the multiple.

The second-order beneficiary is hyperscale competition: MSFT, AMZN, and GOOGL can spread AI infrastructure investment over materially broader customer ecosystems and subsidize cloud pricing with other profit pools. Oracle's most vulnerable position is between those platforms and lower-cost specialized GPU capacity; if AI compute becomes less scarce in 6-18 months, pricing competition could expose the weaker economics quickly. The contrarian case is that contracted capacity is sufficiently long dated and customer-funded to support a rapid utilization ramp; that thesis is falsified if cloud growth decelerates while capex remains elevated for two consecutive reporting periods.

Key Decisions for Investors

• Initiate a 1-3 month relative-value position: short ORCL / long MSFT in equal dollar amounts. The thesis is widening return-on-capital and balance-sheet-quality dispersion rather than a broad AI unwind; cover the ORCL leg if Oracle reports cloud gross-margin expansion alongside a credible reduction in forward capex intensity.
• Use ORCL earnings as the primary catalyst window; maintain a short bias only if implied volatility is below the expected post-results move. If options are inexpensive versus event risk, prefer 3-6 month ORCL put spreads rather than an unhedged short, targeting 2:1 upside-to-premium risk and limiting exposure to a backlog-driven upside surprise.
• Avoid treating reported cloud growth as sufficient confirmation. Set an alert for evidence that operating cash flow covers a rising share of investment spending, plus stable or improving on-premises renewal trends; either development would undermine the negative free-cash-flow and mix-deterioration thesis.
• For a 6-18 month structural expression, favor diversified AI infrastructure beneficiaries with less customer-concentration risk, such as long GOOGL or AMZN versus ORCL. Reassess if Oracle discloses contract terms demonstrating take-or-pay commitments, customer prepayments, and returns above its cost of capital on new data-center cohorts.