Why Oracle’s Record Earnings Sent the Stock Down 12%

Oracle just reported 21% revenue growth, a record $638 billion order backlog, and 93% growth in its cloud infrastructure business. Then the stock fell as much as 12% in a single session.

The paradox has left investors searching for answers, and the resolution sits beneath the headline numbers. Understanding what drove the Oracle stock drop requires moving past top-line revenue and examining what it actually costs Oracle to generate that growth, how the company plans to finance those costs, and what that financing does to the value of each existing share.

What follows explains each layer of the story in sequence: what the earnings showed, why capital intensity matters, how equity dilution mechanics work, what the debt load means, and what the broader lesson is for anyone evaluating high-growth companies in the AI buildout era.

What Oracle actually reported, and why the headline numbers looked great

Oracle’s Q4 FY2026 results were, by any conventional measure, strong. Quarterly revenue reached $19.2 billion, representing 21% growth in USD and 20% in constant currency. Full-year FY2026 revenue totalled $67.4 billion.

The order backlog was the standout figure. Remaining performance obligations (RPO), a measure of contracted future revenue, reached $638 billion, an $85 billion sequential increase from the prior quarter’s $553 billion. Cloud infrastructure (IaaS) grew at approximately 93%, roughly 98% of AI data centre capacity is already under contract, and Oracle holds major contracts with companies including OpenAI. Cloud applications (SaaS) grew at roughly 10%, while legacy software revenue declined approximately 2%, reflecting a deliberate shift away from on-premise licensing rather than demand deterioration.

The Oracle Q4 FY2026 earnings release confirmed quarterly revenue of $19.2 billion, IaaS growth of approximately 93%, and a record $638 billion in remaining performance obligations, the primary source figures underpinning the analyst and investor reactions that followed.

$638 billion in contracted future revenue, nearly 10 times Oracle’s annual revenue, is sitting in the order book.

The stock still fell as much as 12% intraday following the release.

Comparing Oracle’s results against hyperscaler earnings from AWS, Google Cloud, and Azure in the same period reveals a consistent pattern: record backlogs and accelerating cloud growth are appearing alongside compressed free cash flow and rising capital commitments across every major AI infrastructure player.

From software company to data centre builder, what changed and why it costs so much

Oracle’s historical business model was asset-light. Selling software licences and database subscriptions required relatively little physical infrastructure, which is why the company historically generated strong free cash flow. The transition to AI cloud infrastructure is a fundamentally different proposition, requiring the construction and operation of physical data centres at scale, with substantial upfront capital deployed before a dollar of revenue is collected.

The contrast between the two models explains the financial strain:

• Historical model (asset-light): Software licences and database subscriptions; minimal physical infrastructure; strong free cash flow generation; low capital requirements relative to revenue
• Current model (capital-intensive): Physical data centre construction; approximately 1.2 gigawatts of new capacity additions planned in FY2026; $55.66 billion in net capex during FY2026 (exceeding the company’s own $50 billion target); FY2027 guided capex of approximately $70 billion

The scale of spending is not a temporary anomaly. It reflects a structural shift in what Oracle is as a business.

Why the full $70 billion capex figure overstates Oracle’s solo burden

Not all of the capital burden falls on Oracle’s balance sheet alone. Customer prepayments across large AI contracts total $75 billion, and customer-supplied GPUs further reduce the capital Oracle must independently raise. Management has indicated that rising input costs, including memory, power, and personnel, will be passed through to customers via contractual mechanisms rather than absorbed into Oracle’s margins.

These mitigating factors are material. They do not, however, resolve the central investor concern: the overall debt trajectory and negative free cash flow remain the dominant risk.

How Oracle is paying for the buildout, and what equity issuance does to your shares

During FY2026, Oracle raised $43 billion in debt and $5 billion in equity to fund the infrastructure buildout. For FY2027, the company plans to raise approximately $40 billion through a combination of roughly $20 billion in new debt and a $20 billion at-the-market (ATM) equity issuance programme.

The equity component is what rattled shareholders. Equity dilution, the reduction in each existing shareholder’s proportional ownership when new shares are created, is a straightforward mechanic that explains why announced share issuance can drive immediate stock repricing even when operations are strong.

Here is how the arithmetic works:

1. A company earns $100 in total profit with 100 shares outstanding. Earnings per share (EPS) is $1.00.
2. The company issues 20 new shares to fund capital expenditure. Total shares outstanding rise to 120.
3. Total earnings remain unchanged at $100.
4. EPS falls to $0.83, a 17% decline per share, with no change in the underlying business.
5. The stock reprices to reflect the lower per-share economics.

The principle is critical: the market does not wait for dilution to materialise. It reprices the stock the moment equity issuance is announced or signalled.

The mechanics of dilution at the individual company level connect to a broader market dynamic: rising net equity supply from AI-driven capital programmes across multiple large technology companies is removing a structural price support that has underpinned US equity valuations for over a decade.

“The market does not wait for dilution to materialise. It reprices the stock the moment equity issuance is announced.”

This is precisely the dynamic that played out with Oracle. The stock sold off not because demand was disappointing, but because investors recalibrated the per-share economics of a growth story that requires very expensive financing.

What “at-the-market” equity issuance means in practice

An ATM programme allows Oracle to issue new shares gradually into the open market rather than in a single large offering. This provides flexibility, but it creates ongoing dilution overhang throughout FY2027. The $20 billion ATM programme introduces uncertainty because the pace and scale of issuance is not predetermined, making future EPS harder to model.

The debt load Oracle is carrying, and what $124 billion in borrowings means for the business

Oracle’s long-term notes payable stood at approximately $124.7 billion as of the most recent report. Total gross debt, depending on classification methodology and whether short-term obligations are included, is estimated at $134-$156 billion.

The debt burden is already showing up in the financial statements. Interest expense rose 32% to approximately $1.18 billion in a recent quarter, and FY2026 free cash flow was negative $23.7 billion. The net debt-to-EBITDA ratio (a measure of leverage relative to operating earnings) is estimated at roughly 3-4 times based on management guidance, a threshold at which institutional investors typically begin applying valuation discounts.

High leverage creates three specific channels of investor concern:

• Higher future interest expense compresses net income even as revenue grows, reducing the earnings available to shareholders
• Reduced financial flexibility limits the company’s ability to respond to business changes or economic stress without additional dilutive financing
• Valuation multiple compression occurs as investors apply a discount for financial risk, meaning the stock trades at a lower multiple of earnings than a less-leveraged peer with identical growth

EBITDA growth is expected to help moderate the leverage ratio over time as the buildout matures and revenue scales. Whether that moderation arrives quickly enough is one of the open questions the market is pricing.

The institutional market for AI debt financing is itself under structural pressure: a $15 trillion corporate refinancing wall expected between 2026 and 2028 is pushing institutional capital toward hybrid AI credit instruments, creating demand conditions that simultaneously support Oracle’s ability to raise debt and introduce new complexity into how that debt is priced.

Thin margins on AI cloud, why Oracle’s biggest growth segment earns less than it looks

The fastest-growing part of Oracle’s business is currently its least profitable. Internal Oracle documents reported by The Information (via CNBC) showed gross margins of approximately 14-16% in certain periods on the AI infrastructure business, materially below prior analyst assumptions.

That figure sits far below management’s stated target of 30-40% gross margins on AI data centre projects over time.

Oracle’s fastest-growing segment currently earns gross margins of 14-16%, compared to a management target of 30-40% and the much higher margins on its legacy software business.

This margin dynamic compounds the free cash flow problem. Even with a $638 billion backlog and 93% IaaS growth, thin incremental margins combined with heavy upfront capex produce near-term free cash flow that looks substantially weaker than revenue growth alone would suggest. The negative $23.7 billion FY2026 free cash flow is the tangible result of this combination.

What it would take for margins to improve

Management expects margin improvement as the business mix shifts, with more software and services layered on top of the infrastructure. Scale benefits are expected to reduce per-unit costs, and customer contracts should mature into more profitable phases over time.

The pace of this improvement is a key variable. If margins do not move toward the guided 30-40% range within two to three years, the free cash flow inflection point investors are pricing in would be pushed further out, and the valuation would face additional pressure.

For investors wanting to stress-test the assumption that large cloud backlogs reliably translate into margin expansion, our deep-dive into Alphabet’s cloud backlog and margin dynamics examines how Google Cloud moved from 18% to 32.9% operating margins in a single quarter while simultaneously absorbing $35.7 billion in quarterly capex, offering a direct comparison case for Oracle’s own guided margin trajectory.

The metrics that will tell investors whether Oracle’s bet pays off

The core investor question is not whether AI demand is real. Oracle’s $638 billion backlog removes that doubt. The question is how much of that demand’s value will accrue to existing shareholders after accounting for capital costs, financing dilution, and margin dynamics.

Five forward-looking metrics will determine the answer:

1. Free cash flow trajectory: When does negative FCF (negative $23.7 billion in FY2026) inflect positive as the buildout matures? Positive FCF is the clearest signal that growth is generating more cash than it consumes.
2. Gross margin progression: Does the 14-16% current level on AI cloud move toward the guided 30-40%? This indicates whether scale benefits and business mix improvements are materialising.
3. Equity issuance pace: How quickly is the $20 billion ATM programme being utilised, and does the programme expand? Faster or larger issuance means more dilution than currently modelled.
4. Debt and interest coverage: Does EBITDA growth outpace debt accumulation, improving the net-debt-to-EBITDA ratio from the current estimated 3-4 times? A declining ratio signals the leverage risk is receding.
5. RPO conversion rate: Does the $638 billion backlog convert to revenue on the expected timeline, validating the capex investment? Delayed conversion would raise questions about the pace of return on deployed capital.

These metrics apply beyond Oracle to any company in the AI infrastructure buildout. The framework separates revenue growth, which captures demand, from shareholder value creation, which captures what each investor actually receives.

Strong revenue growth and a falling stock price can both be true at once

The paradox resolves cleanly. 21% revenue growth and a 12% stock decline are not contradictory once the full picture is visible: capital intensity, dilution mechanics, and margin realities each explain a portion of the repricing.

Three factors determine whether growth creates or destroys per-share shareholder value:

• Capital intensity: How much must be spent upfront to generate each dollar of future revenue. Oracle’s FY2027 guided capex of $70 billion places it at the extreme end of this spectrum.
• Financing mix: Whether that capital comes from debt (which raises interest costs and reduces flexibility) or equity (which dilutes existing shareholders). Oracle is using both, with a $20 billion equity programme and approximately $20 billion in new debt planned for FY2027.
• Margin profile: Whether the incremental revenue comes with adequate profitability to justify the capital deployed. Current AI cloud margins of 14-16% fall well short of the 30-40% management target.

Oracle’s $638 billion backlog represents genuine demand and a potentially strong multi-year business. The debate is not about the company’s customers. It is about the economics of serving them at this scale of investment, and whether the value created by that investment will flow to existing shareholders or be consumed by the capital required to capture it.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Past performance does not guarantee future results. Financial projections are subject to market conditions and various risk factors.