ORCL — Oracle Corporation

The tape, not the sell-side, is bearish. As of 2026-09-13 ORCL was -18.4% over three months, -5.35% over the past week and -1.8% on the day, meaning the market sold the stock into (and immediately after) a genuine beat-and-raise on 2026-09-10; it also sits 10.4% below its 200-day SMA and in the bottom third of a $114.50-$329.50 52-week range. Meanwhile the sell-side is constructive to a fault - 56 Buy ratings, an average target of $236.52, Barclays at $252 post-earnings, some targets up to $400. That gap tells you the market is discounting balance-sheet and capex risk, not doubting the P&L.

Growth is clearly accelerating, not stable: revenue YoY went +12.2% (Aug-2025), +14.2% (Nov-2025), +21.7% (Feb-2026), +20.6% (May-2026), +29.6% (Aug-2026) - a genuine inflection around the Feb-2026 quarter that has compounded since. Operating margin followed the same path, 28.7% (Aug-2025) to 35.3% (Aug-2026), with operating income up 59.5% YoY in the latest quarter, and diluted EPS of $1.56 versus $1.01. Two data caveats: the Aug-2026 gross profit line ($19.14B on $19.35B revenue, a ~99% gross margin) is mis-keyed in the source data and should be ignored, and the Nov-2025 quarter's net income of $6.14B exceeded its operating income of $4.73B, implying a large non-operating gain - so the TTM GAAP EPS of $6.379 is flattered by a one-off, which is why trailing P/E looks cheaper than underlying earnings justify. The offsetting trend is the cash-flow and balance-sheet side: five consecutive quarters of negative free cash flow (-$0.4B, -$10.0B, -$11.5B, -$1.9B, -$5.4B), TTM FCF of roughly -$28.8B, long-term debt up from $82.2B (Aug-2025) to $148.3B (Aug-2026), and stockholders' equity jumping $24.3B sequentially in the Aug-2026 quarter on only $4.76B of net income - i.e. third-party equity funding alongside the debt. The revenue trend is a structural upcycle; the funding trend is a structural risk.

As of 2026-09-13, ORCL's $150.15 price is 23.2x TTM GAAP EPS of $6.379 versus a 5-year average P/E of 34.4x, and 6.03x sales versus a 5-year average of 7.0x - on paper it is cheap versus its own history (the deterministic model's valuation context scores 79.1, 'Cheap vs. its own history'). The forward math is more striking: 18.5x FY27 consensus non-GAAP EPS of $8.114 against 30% EPS growth, with a PEG of 0.50 and EV/EBITDA of 15.2x. But I would call it 'cheap for a reason' rather than mispriced: the trailing base is flattered by the Nov-2025 one-off (ex-it, trailing P/E is closer to 27x), ROIC is only 8.0%, free cash flow is deeply negative, and the equity is now levered 2.34x. The discount is compensation for leverage and capital intensity, and I would not underwrite a return to the 34x historical multiple.

This is the rare case where the stock is not pricing in growth that has already shown up: revenue is up 30% YoY, OCI up 121%, RPO up roughly $209B YoY to $664B, and management raised FY27 guidance to at least $90B revenue and $8.10 non-GAAP EPS - yet the shares are 54% off the high and fell ~5% in the week around that report. What the market is pricing instead is the forward cash-flow statement: capex that consumed $28.5B in a single quarter, $148.3B of long-term debt, and a $24B quarterly equity build that suggests external funding needs. The practical implication: the earnings line is not the swing factor anymore - the FCF inflection is. When capex peaks and RPO conversion shows up as positive FCF, the discount should close; if AI compute demand or the credit markets wobble first, the debt stays and the equity absorbs it. Also worth noting my fair value of $187 sits well below the $236.52 street average - I think consensus is anchored to pre-de-rating multiples rather than to an 8% ROIC infrastructure business.

As of 2026-09-13, RSI(14) was 51.05 - dead neutral, neither oversold nor overbought - with price 0.6% above the 20-day SMA (derived ~$149) and 6.82% above the 50-day (derived ~$141), but 10.36% below the 200-day (derived ~$168). That is a short-term basing attempt inside an intact longer-term downtrend, following a 54% drawdown from the $329.50 high - not a washed-out capitulation (RSI 51 is not an oversold entry signal) and not a confirmed reversal. Actionable levels: a decisive reclaim of the ~$168 200-day would be the first real technical evidence that the bottom is in, while a loss of the ~$141 50-day would signal the base is failing and put the $114.50 52-week low back in play.

• AI infrastructure capex cycle: Oracle's growth is now a derivative of hyperscaler and enterprise AI compute demand; any evidence of overbuild or slower GPU/accelerator deployment hits the $664B RPO conversion narrative directly.
• Interest rates and credit spreads: $148.3B of long-term debt and a business that must keep issuing paper to fund construction makes Oracle unusually rate-sensitive for a software company - easing conditions help, a spread widening hurts EPS and the multiple at once.
• Data-center physical constraints: power availability, grid interconnection, land and cooling. Oracle added 850MW of capacity in Q1 FY27 - the pace of power procurement, not sales demand, may be the limit on how fast the backlog converts.
• Sector rotation within technology: capital has been moving toward cash-generative software and away from AI-infrastructure names with negative free cash flow; with a beta of 1.73, ORCL is a high-amplitude expression of that rotation.
• Regulation and sovereign/enterprise procurement: multicloud arrangements (Azure, AWS, GCP) are a tailwind, but EU data-sovereignty rules and scrutiny of large AI contracts add compliance cost and can slow deal timing.

• Q2 FY27 earnings on 2026-12-14 (after close): the market will test whether OCI's triple-digit growth holds, whether capex is peaking, and whether the $664B RPO is converting into recognized revenue fast enough to justify the build.
• First sign of positive free cash flow: a capex plateau or backlog-conversion inflection would flip the narrative from 'levered capex hole' to 'operating leverage' faster than any other single data point.
• Financing events: new debt issuance, credit-rating actions, or further equity raises (equity rose $24.3B sequentially in the Aug-2026 quarter) will signal how much external capital the buildout still needs.
• New or expanded AI/cloud contracts (or cancellations) among the large customers underpinning RPO, plus any shift in the rate environment that changes the cost of funding the build.

• Balance-sheet leverage: $148.3B long-term debt against $66.8B of equity (D/E 2.34), a 1.17 current ratio, and five straight quarters of negative FCF totaling roughly -$28.8B on a trailing basis. This is a real capital-structure risk, not a cosmetic one.
• RPO counterparty quality: the $664B backlog is only worth what the customers behind it can pay; concentration in a handful of capital-hungry AI customers means a single counterparty problem would reprice the whole thesis.
• Low returns on incremental capital: ROIC of 8.0% means growth funded with debt and equity is being deployed at roughly the cost of capital, and the depreciation burden from ~$75B of trailing capex has not yet hit the income statement in full.
• Dilution and funding dependence: the unexplained ~$24B sequential equity increase in the Aug-2026 quarter implies Oracle is now partially equity-funding the build, and the 1.42% dividend plus buybacks compete directly with capex for a business burning cash.
• Competitive and mix pressure: AWS, Azure and GCP compete hard on price, and the revenue mix is shifting toward lower-margin IaaS even as reported operating margin rises - if IaaS pricing compresses, the margin trend reverses.
• Technicals and sentiment: below the 200-day, 54% off the high, beta 1.73 - if the AI-infrastructure trade de-rates further, ORCL has no multiple cushion near term despite looking cheap on trailing numbers.

The business is accelerating while the stock is priced for trouble - revenue +30% YoY, OCI +121%, RPO of $664B, raised guidance, forward P/E of 18.5x and PEG of 0.5 at $150.15 as of 2026-09-13 - but the business Oracle is becoming is a highly levered infrastructure builder, with $148.3B of long-term debt, negative FCF in five straight quarters, an 8% ROIC, and equity issuance now sharing the funding burden. My calculated fair value is about $187, implying roughly 25% upside, so the risk/reward is favorable for a patient investor who believes the backlog converts - but the dispersion around that number is unusually wide because the swing factor is the AI capex cycle and the credit markets, not the next two quarters of EPS. Own it small, and watch the capex-to-FCF inflection rather than the earnings beat.

I applied 23x to FY27 consensus/guided non-GAAP EPS of $8.11 (management guided $8.10; consensus $8.114), giving $186.5, rounded to $187 - a deliberate discount to Oracle's 34.4x 5-year average trailing P/E to reflect 8% ROIC, negative free cash flow, $148.3B of long-term debt and an equity base of only $66.8B, but still a premium to the 18.5x the stock traded at on 2026-09-13. Cross-check: ~15.5-16x FY27 EBITDA of roughly $39-41B on an EV basis (net debt ~$112B, ~2.88B shares) brackets $165-$190, so $187 sits at the constructive end of a range I regard as genuinely wide.