Oracle beats on earnings and revenue, adds $20 billion to planned capital raise

Oracle beat fiscal Q4 expectations with adjusted EPS of $2.11 versus $1.96 consensus and revenue of $19.18 billion versus $19.10 billion, while revenue rose 21% year over year. The company raised its adjusted EPS forecast for fiscal 2027 to $8.05, but maintained $90 billion revenue guidance and said it plans to raise $40 billion more through debt and equity to fund its AI buildout, pressuring the stock 5% in extended trading. Remaining performance obligations surged 363% to $638 billion, underscoring strong AI-related demand despite concerns over heavy capital needs and negative free cash flow of $23.7 billion for the year.

Analysis

The market is telling you the equity is no longer being valued on near-term earnings beats; it is being discounted as a financing story with AI capex as the dominant variable. That matters because when a software name starts funding data-center expansion like an infrastructure company, the multiple should migrate from growth/quality toward balance-sheet and cash-conversion scrutiny. The immediate second-order effect is that Oracle's cost of capital is now part of the competitive landscape: if it can still sign large AI workloads while issuing equity/debt, then rivals without similar backlog visibility may be forced to choose between growth and dilution.

The bigger signal is not the quarter, but the concentration risk embedded in the backlog. If a meaningful share is tied to one hyperscale AI customer base, the market will begin to handicap a renegotiation/refinancing cycle over the next 12-24 months rather than a simple revenue ramp. That creates a paradox: reported demand can look exceptional while equity holders are effectively underwriting long-duration capex with low visibility on incremental returns, especially if power, land, and GPU supply constraints delay monetization.

For the financing partners, the read-through is constructive but nuanced. Banks and private-capital providers can earn fees and senior claims on a high-profile buildout, but the equity component signals management may not want to overleverage into a potentially lumpy demand curve. The contrarian view is that the selloff may be overdone tactically if the next 1-2 quarters confirm backlog conversion and free cash flow inflects less negatively than feared; however, if financing headlines persist, the stock can de-rate further even on beats because the market is treating capital needs as a proxy for execution risk, not opportunity.

Over the next 30-90 days, the key catalyst is guidance credibility: if remaining performance obligations translate into booked revenue without another equity raise, the stock can stabilize quickly. If not, every incremental AI contract announcement may be interpreted as more dilution rather than more value, which is a dangerous narrative loop for holders.

Key Decisions for Investors

• Short ORCL on strength over the next 2-6 weeks if the shares rebound toward pre-print levels; risk/reward favors fading rallies until management proves financing needs are bounded. Use a tight stop above the post-earnings gap high.
• For event-driven exposure, buy ORCL put spreads 1-3 months out to express downside from dilution/financing overhang while limiting premium burn; target a move back toward the low-90s if capital markets concern stays front and center.
• Long BAC vs short ORCL as a relative value pair over 1-2 quarters: BAC benefits from underwriting/financing activity and has less single-name execution risk, while ORCL bears the balance-sheet and capex uncertainty.
• Watch BX for a tactical long only if the Michigan data-center financing closes on attractive terms; the setup is fee-positive, but size it small because any sign of weak economics would pressure private-capital multiples broadly.
• Do not chase ORCL until first-quarter commentary confirms capex discipline and backlog monetization; if free cash flow improves and financing cadence slows, revisit a long only after the next print, not on headline AI demand alone.