After the US market closed on June 10th, Oracle (ORCL) released its Q4 FY2026 earnings report, surpassing expectations on nearly all key metrics. Revenue reached $19.18 billion, a 21% year-over-year increase, exceeding the Wall Street consensus of $19.1 billion. Non-GAAP EPS was $2.11, nearly 8% higher than the $1.96 consensus.
Despite the strong numbers, the stock price plunged over 10% in after-hours trading. While the Nasdaq itself fell nearly 2% that day due to a three-year high in CPI data putting pressure on the broader market, Oracle's steeper decline indicated investor concerns beyond just macroeconomic noise.
Investors Look Beyond the Impressive Figures
Examining the growth side, IaaS (Cloud Infrastructure) revenue was $5.79 billion, surging 93% year-over-year. The growth rate over the past eight quarters shows a sequence of 45%, 52%, 49%, 52%, 55%, 68%, 84%, and 93%—an acceleration that is exceptionally rare for a large tech company, especially on an expanding base.
Remaining Performance Obligations (RPO) increased from $553 billion last quarter to $638 billion, a single-quarter addition of $85 billion. Management explicitly stated on the earnings call that the RPO growth in Q3 and Q4 was "almost entirely from large-scale AI contracts."
These figures together paint a picture of extremely robust demand. However, the market that evening was less focused on the strength of demand and more concerned with the cost required to meet it.
FY2026 full-year capital expenditures reached $55.7 billion, significantly exceeding the prior guidance of $50 billion. Q4 alone saw $15.9 billion in CapEx. Concurrently, management announced a plan to raise $40 billion in FY2027, with $20 billion coming from an At-The-Market (ATM) equity offering and the remainder through debt. The $20 billion share offering implies direct dilution for existing shareholders, serving as one trigger for the after-hours sell-off.
A Client-Funded Hardware Model
One of the most notable pieces of information from the call was the progress of the Bring Your Own Hardware (BYOH) model.
This model involves clients purchasing GPUs themselves or making substantial prepayments, then having Oracle operate the hardware in its data centers. Cumulative client investment through this model has reached $75 billion this quarter.
This capital is not counted as CapEx for Oracle, but the resulting operational revenue is included in IaaS revenue.
In other words, Oracle is exploring a "light-asset" path—having clients bear the hardware acquisition cost while it earns operational fees. CEO Clay Magouyrk mentioned on the call that current GPU utilization is at 97.5%. More interestingly, among expired GPU leases, 49% of clients chose to renew, retaining 92% of the GPUs, while the un-renewed GPUs were immediately resold to customers on the waitlist.
Management cited a project-level return on investment "approaching 30%." If this figure is credible, the BYOH model could indeed drive IaaS growth without massively expanding the balance sheet.
However, it's important to note: the company-wide Non-GAAP operating margin is 44.8%, a figure buoyed by high-margin traditional Software business. The pure IaaS margin is likely below the company average.
Management's long-term target margin for IaaS is 30-40%—which itself implies the current margin has not yet reached that range.
Deciphering the $55.7 Billion in CapEx
Placing CapEx next to revenue: FY2026 capital expenditure was $55.7 billion against revenue of $67.4 billion, resulting in a CapEx/Revenue ratio of 83%. Operating cash flow was $32 billion (up 54% year-over-year), but was completely consumed by CapEx, leading to negative free cash flow of $23.7 billion for the year.
Total debt increased from $92.6 billion to $129.5 billion, a 40% rise in one year. In November 2025, Barclays downgraded Oracle's debt rating to underweight, warning of a potential slide to BBB-—the final threshold for investment grade. However, in April 2026, CreditSights offered a contrary view, upgrading Oracle's credit to Outperform, citing the RPO as providing a substantial anchor for future revenue.
Management made a commitment on the call: no new debt issuance for the 2026 calendar year. Of the $40 billion financing planned for FY2027, half is from the ATM offering and half from debt, but new debt issuance won't occur until 2027. This statement attempts to draw a line between growth and leverage, though whether investors accept it remains to be seen.
Breaking down the $40 billion reveals a structure more complex than the headline number. Oracle had already completed two major financings early in the 2026 calendar year: a $25 billion senior unsecured bond issuance in early February (led by Goldman Sachs, attracting $129 billion in oversubscription across eight tranches from 3 to 40 years) and a $5 billion mandatory convertible preferred stock offering (6.50% coupon, conversion price around $160, led by Citi, also heavily oversubscribed).
This means the $45-$50 billion financing plan for the 2026 calendar year is largely executed. If the "FY2027 $40 billion" mentioned on the call includes the already authorized but unused $20 billion ATM facility, the actual incremental new debt ceiling is only $20 billion, with management committed to no new debt in 2026. In other words, the new supply pressure on credit markets is significantly less than the "$40 billion" figure suggests.
Another easily overlooked detail: the dilution effect of the $5 billion mandatory convertible is delayed until mandatory conversion at maturity, while the ATM offering causes ongoing but flexible dilution, allowing management to sell opportunistically based on the stock price. Their impact mechanisms on the share price differ—one is a distant, certain dilution; the other is a near-term, uncertain pressure.
Changes in revenue structure are also noteworthy. Traditional Software business declined 2% year-over-year this quarter, marking a second consecutive quarter of negative growth. SaaS growth has remained around 10% for eight consecutive quarters. Oracle is betting almost all its resources on the IaaS track—if AI computing demand falls short of expectations, it lacks a Plan B.
How Long Will a $638 Billion Order Book Last?
An RPO of $638 billion is an astronomical figure. Dividing it by annualized IaaS revenue (approximately $23.1 billion) theoretically "fills" 27 years of capacity. However, large enterprise contracts don't work that way.
Management disclosed the RPO's term structure: 12% of revenue is expected to be recognized within 12 months, 34% within 13-36 months, and 54% beyond 36 months.
Most of the order book won't convert to revenue until over three years from now. Short-term revenue growth depends more on existing capacity and the pace of new signings than the absolute size of the RPO.
Another unanswered question concerns cancellation terms. Management did not disclose the cancellation or breach conditions for these large AI contracts. However, under the BYOH model, where clients have prepaid significant funds or purchased hardware, the sunk cost is high, making large-scale cancellations less likely.
The FY2027 revenue guidance remains unchanged at $90 billion, implying 34% growth. Q1 guidance projects revenue growth of 27-29%, with Cloud growth of 58-64%. Management also raised the Non-GAAP EPS guidance to $8.05.
Key Areas to Monitor Moving Forward
Three key focal points exist for FY2027.
First, whether Q1 IaaS growth can be maintained above 90%. If growth begins to slow, market confidence in the $638 billion RPO would rapidly diminish.
Second, whether the proportion of the BYOH model continues to expand. If client prepayments and self-purchased GPU volumes keep growing, Oracle's own CapEx pressure would ease, and free cash flow could turn positive sooner.
Third, the direction of credit ratings. The divergence between Barclays and CreditSights reflects two market interpretations of Oracle's balance sheet. If rating agencies downgrade further, Oracle's financing costs would rise, directly eroding margins.
Management has provided a long-term target for FY2030: 31% compound annual revenue growth and 28% compound annual EPS growth. If achieved, Oracle's revenue scale in five years would approach $180 billion, comparable to AWS's current size. This is an extremely aggressive target, and its realization depends on whether AI computing demand continues to explode as management expects.
Prior to the earnings report, five investment banks raised their valuation expectations for Oracle, with TD Cowen's $300 price target being the most aggressive. However, the stock's performance that evening indicated that, at least in the short term, the market is less concerned with whether Oracle can sustain high growth, but rather with how much capital that growth will consume.
Comments