Oracle (ORCL): fair value

Fair-value band

The normal range for this name runs from $177 (cheap) to $384 (expensive); it is now priced at $130.

That is 2.2σ below this name's own normal premium.

Standing premium not backed by fundamentals: nothing in this company's health score has changed to earn the rating the market normally gives it.

Is Oracle undervalued?

Cheap, relative to its own history. Oracle trades at $130, 16% below the fair value this model estimates, but it has habitually traded 81% above fair value, and today's gap is 2.2σ below that norm, wider than the range it normally trades in, so the verdict is Cheap: priced further below its own norm than usual, not just below the model's fair value. The health score behind that pricing hasn't moved to earn it, though: read Cheap here as “priced the way this name normally is”, not as a claim the price is justified.

Diligence checklist

6 pass · 3 fail · 2 no data

• PASSCash actually collected from profit, last 12 months: 172%XBRL · Oracle gets $1.72 of real cash for every $1 of reported profit which means profits are real and not just numbers on paper. That's a good sign.
• FAILGap between paper profit and real cash: -11.6%XBRL · Oracle collected more cash than it booked as profit, by 11.6% of everything it owns. That is the conservative direction, but it is far enough past my 5% line to flag: a gap this wide usually means something large and non-cash is running through the earnings, and I want to know what it is before I trust the profit figure.
• PASSDebt compared to one year of cash earnings: 2.59xXBRL · It would take 2.59 years of earnings to clear what Oracle owes, net of the cash it holds. My limit for a name I would buy is three years, so this passes, though the balance sheet is doing more of the work than it does at the top of this list.
• PASSHow much more it earns on capital than that capital costs: +0.25ppXBRL · Oracle earns 0.3 points more on the capital it employs than that capital costs. That clears the bar, but only just: a slightly worse year or a slightly higher cost of money and this business is running to stand still. The gap is no wider than it was at the start of the window.
• PASSStock sold by executives and directors on the open market, last 6 months: $0.4MEDGAR · Insiders at Oracle sold $0.4M of stock on the open market in the last 6 months, which is inside the $5 million line I draw. Selling that size is ordinary diversification rather than a signal about the business. The sellers: Smith Maria $0.4M. A further $68.6M went out under pre-planned 10b5-1 schedules, which I don't count: those sales are set up months ahead and say nothing about what management thinks today. Options being exercised, and shares held back to cover the tax on them, are not sales and I don't count them either. filing ↗
• FAILHow long the CEO has run this company: 1.1 yearsTEXT · Oracle's chief executive took the job on September 22, 2025, 1.1 years ago. I want a full cycle at the company, which I take as 5 years, before a management record tells me anything: a short run has not yet been tested by a bad year. filing ↗
• N/AProfit margin held up during the last sales slump: n/aXBRL · There's no revenue downturn in Oracle's filed record, or gross profit isn't tagged, so this test has nothing to measure. It needs a bad patch to look at: without one, whether customers would stay through a price rise is untested rather than proven.
• FAILNormal profit margin across a full business cycle: FCF margin -33.8% vs 13.1% medianXBRL · Oracle is earning a -33.8% cash margin against a long-run average of 13.1%, 357% worse than normal. Either this is a trough the business climbs out of, or the average no longer describes it. My valuation grows the average, so it assumes the first: if that's wrong, the value on this page is too high. Across the window the margin ran between -33.8% in August 2026 and 29.3% in February 2024.
• N/AHow much of the stock trades on an average day: $4,070MMARKET · Oracle trades so much volume that this basically isn't a concern. There's no pass or fail here because the answer depends on how big a position I'd actually want to take.
• PASSShare of revenue from its single biggest customer: no customer 10%+ of revenue disclosedTEXT · Oracle's June 22, 2026 10-K doesn't disclose any customer at 10% or more of revenue. The accounting rules require that disclosure only when it applies, so its absence means no single customer clears that bar, which is well inside the quarter of revenue where I stop work. This comes from the filing's prose rather than a tagged number: the percentages sit in an XBRL field the SEC's bulk data drops. filing ↗
• PASSWhether the company switched auditors in the last 2 years: noneEDGAR · Companies sometimes switch accounting auditors right before something bad is uncovered which can be a warning sign. Oracle hasn't changed auditors in the last 2 years, so there's no such red flag here.