Q1 of fiscal 2027 is the first quarter in which the build shows up in revenue at scale: cloud infrastructure revenue was , up , on megawatts of AI capacity delivered. Management still gives no AI infrastructure revenue figure; the ledger rolls its estimate forward to , of total revenues of . Remaining performance obligations rose to , with expected within twelve months.
The money funding the build changed shape. Capital expenditures were , against which the company collected of customer prepayments and issued of common stock, and raised no debt. Depreciation rose to from , operating lease expense to from and interest expense to from ; lease commitments not yet commenced reached . The filing still explains all of it by infrastructure expenses and the expansion of data centers.
On who pays, the sources move a little. One site took up to of the quarter’s GPUs and is said to have trained a model released by OpenAI, the first time the stored sources connect a named lab to delivered capacity. Asked directly whether the customer-funded contracts come from AI labs, management answers that demand is broad based, from start-ups to the most valuable investment-grade companies, and describes payers that have raised money. That is the first direct support for the mixed funding the ledger has read on those contracts since the first covered quarter.
Elsewhere: embedded AI moves to stated usage, more than uses and tokens, which sizes the inference behind it as small; priced agentic capacity loses its customer count; a forward-deployed engineering channel opens; adjacent services and the company’s own coding-tool bill go unmentioned. The part of the rise in infrastructure expenses that depreciation and leases do not explain, read as power and other running cost, is . Research and development fell and sales and marketing on the year under a restructuring plan the filing ties in part to AI adoption and has now enlarged by . Every size in this exhibit is the ledger’s inference. Database demand, the restructuring cost and both staff savings credit AI beside other causes and are left unsized. Also unsized: the embedded agents (no price, counts of use), priced agentic capacity (no count, price or revenue), the halo (AI named beside sovereignty and the applications) and the application-revenue toll (denied by management).
Sized channels against the income statement, Q3 2026
9 of 20 channels sized
Each blue mark is one channel's dollars for the quarter; a bar is the range of an estimate. Grey marks are the company's reported lines. The distance between them is the point: how large the AI channel is next to the line it sits in.
Total revenuesreported line
$19.34bn
Total operating expensesreported line
$12.62bn
Depreciationreported line
$3.16bn
Research and developmentreported line
$2.40bn
Sales and marketingreported line
$1.81bn
Interest expensereported line
$1.43bn
General and administrativereported line
$376mn
Amortization of intangible assetsreported line
$202mn
Restructuringreported line
$165mn
Capital expenditures on data centers and AI computespend · our inference
AI infrastructure: GPU and accelerator capacity sold for training and inferencerevenue in · our inference
Depreciation of the data center and AI compute buildspend · our inference
Data center lease cost of the AI buildspend · our inference
Interest on the debt raised to fund the buildspend · our inference
Power, network and other running cost of the AI fleetspend · our inference
Oracle AI Data Platform, agent tooling and access to third-party modelsrevenue in · our inference
Forward-deployed engineers for the AI Data Platform and agent studiospend · our inference
Model inference behind the AI embedded in the applicationsspend · our inference
$100k$1mn$10mn$100mn$1bn$10bn$100bn
AI channel, dollars for the quarter low to high of an estimate reported lineLog scale: each gridline is ten times the one before.
New money and old money, Q3 2026
4 new13 expanded3 relabelled
Each channel is tagged once for whether its money existed before language models, from the company's annual report and call at the start of the period. A bar splits one flow's sized dollars by that tag. The incremental total is the part that would not be there without the models: a new channel counts in full, an expanded one only for what AI added, a relabelled one at zero.
Flow, sized total
Split by novelty
Incremental total
Paid for AI$2.37bn to $4.27bn sized
new $450k to $18mnexpanded $2.37bn to $4.25bn
Incremental total $2.37bn to $4.27bnpoint $3.53bn
Revenue arriving through AI$3.28bn to $5.40bn sized
new not sizedexpanded $3.28bn to $5.40bnrelabelled not sized
Incremental total $5.0mn to $5.40bnpoint $40mn$4.48bn in 1 channel has no traced baseline
The sized total counts every channel the company credits to AI, including relabelled money that existed before language models and the ledger's own estimates for it. The incremental total counts relabelled channels at zero. A flow is split by layer where its dollars sit at more than one: end use, compute sold to builders, and hardware. The same dollar can be a buyer's spend, a cloud's revenue and a chipmaker's revenue, so the layers are never added together.
Paid for AI9 channels · $2.37bn to $4.27bn sized · $2.37bn to $4.27bn incremental · 2 not sized
other
Capital expenditures on data centers and AI compute
103.1% to 140% of the quarter’s revenue; capital spending on the build, not added into totals (its depreciation is)
Matched line moved : the whole line, not this channel.
our inferencedisclosure: quantified· motive: offensive· before LLMs: expanded· layer: compute
Capital expenditures were in the quarter against a year earlier, which the filing attributes to the expansion of data centers (claim c75). Net of customer prepayments and supplier financing flows the cash outlay was , and management guides to as much as for the fiscal year with not more than net (claims c8, c9). The size, , is the ledger’s own: the quarter’s cash capital expenditures times an assumed AI share, sized as a level; management says the general-purpose cloud needs some capital but far less than the AI clusters (claim c62). Cash capital expenditures leave out equipment bought on supplier financing, which shows up as financing flows, net this quarter, and spending accrued but not yet paid, at the quarter end. Funding stays mixed and its composition shifts: of customer prepayments, of new equity and supplier financing, beside operating cash flow of (claims c57, c10). This release is the first to tabulate capital expenditures by quarter, putting the quarter ended 2026-02-28 at .
Evidence: 19 quotes, 16 figures, 3 confounds, 3 from before coverage
Cash paid for servers, accelerators, networking and data center fit-out to deliver contracted AI capacity. It is capitalized, not expensed: it reaches the income statement later as depreciation, which counts in the spend total in its place, so this channel is traced and shown and left out of totals (the methodology rule on capital spending; the buildout itself is a separate design). The filing attributes the increase to the expansion of data centers without naming AI; management ties the capital program to AI cloud infrastructure. The payees are chip and equipment suppliers and contract manufacturers. The measure is cash capital expenditures as the cash flow statement reports them: equipment bought on supplier financing is in it only as the financing is repaid, and accrued capital spending not yet paid is not in it. A budget redirected to an AI build is sized as a level: the size is the quarter’s capital expenditures times an assumed AI share, and the traced pre-build quarter is the baseline.
Why this motive
Carried from the prior quarters: capital committed against contracted AI capacity, with each project described as generating strong free cash flow once ramped (claims c11, c38).
Before LLMs: expanded
Capital spending on data center capacity predates the build: in fiscal 2024, about a quarter, and in Q1 of that year, when management already named building data centers as its biggest challenge and AI development companies had contracted for more than of training capacity. The baseline for the incremental total is the one quarter of fiscal 2024 the anchor traces, . The size is the whole of an activity that existed before.
“As of today, AI development companies have signed contracts to purchase more than $4 billion of AI training capacity in Oracle's Gen 2 Cloud. That's twice as much AI training as we had booked at the end of the last Q4.”
“We have invested in the rapid expansion of the Oracle Cloud by increasing existing data center capacity and adding data centers in new geographic locations to meet current and expected customer demand. We expect this trend will continue.”
Capital expenditures (cash paid; excludes equipment bought on supplier financing until repaid and accrued spending not yet paid) · 2026-CQ3
Capital expenditures (cash paid), prior-year quarter · 2025-CQ3
Net cash outlay for capital expenditures (capital expenditures less supplier financing flows and customer prepayments) · 2026-CQ3
Customer prepayments with a significant financing component received in the quarter · 2026-CQ3
Short-term financing related to capital expenditures, net (negative: a net repayment in the quarter); supplier-financed equipment reaches the cash flow statement here as it is repaid, not in capital expenditures · 2026-CQ3
Unpaid capital expenditures, a non-cash investing item at the quarter end: capital spending accrued and not yet paid, outside cash capital expenditures · as-of 2026-08-31
Capital expenditures expected for fiscal 2027, high end of the stated range · FY2027
Net cash capital expenditures expected for fiscal 2027 (not more than) · FY2027
Operating cash flow · 2026-CQ3
Free cash flow (negative: an outflow) · 2026-CQ3
Property, plant and equipment, net · as-of 2026-08-31
Construction in progress, mostly servers, networking equipment and leasehold improvements not yet deployed · as-of 2026-08-31
Unconditional purchase and other obligations, primarily component supply for cloud infrastructure assets and data center power · as-of 2026-08-31
Net proceeds of common stock issued through the at-the-market program in the quarter · 2026-CQ3
Capital expenditures (cash paid) in the quarter ended 2026-02-28, as first tabulated by quarter in this release · 2026-CQ1
AI capacity delivered to customers since the end of the prior quarter, megawatts · 2026-CQ3
Reported line it is matched to
Capital expenditures were against , an increase of . The filing attributes the increase to the expansion of data centers and does not name AI.
2026-CQ3: 2025-CQ3:
What else could explain it
line composition: Capital expenditures also fund multicloud database regions, general-purpose capacity and facilities; the AI share is assumed.
other: Customer prepayments counted in operating cash flow fund part of the spending, so free cash flow and net cash outlay give different pictures of the same quarter.
line composition: Cash capital expenditures leave out equipment bought on supplier financing until it is repaid and capital spending accrued but not yet paid; neither cash nor incurred capital spending is fully measured.
Quotes
“The vast majority of those new contracts were via prepay or bring your own hardware or similar mechanic, so will not require incremental capital from Oracle. Also, that new RPO will not impact our CapEx or revenues until fiscal 2028 or beyond.”
“Our CapEx for the quarter was $28 billion, leading to negative free cash flow of $5 billion. Our net cash CapEx, so net of prepayments, was $18 billion for the quarter.”
“Lastly, we are quite pleased to announce that we completed our previously disclosed $20 billion at-the-market equity issuance in entirety during the Q1.”
“We delivered 850 megawatts of AI capacity containing more than 300,000 GPUs to customers since the end of Q4. Delivery in Q1 is almost three times what we delivered in all of Q4 and 73% of the total capacity we delivered last fiscal year.”
“We delivered 131,000 GPUs there in Q1, 1.9 times the volume delivered in Q4. Six of the eight campus buildings, representing 618 megawatts and 75% of total capacity, have now been delivered to the customer.”
“I think that we have to separate out in our minds what Oracle spends as CapEx, uncouple that directly from how we think about how the business can grow.”
“We just delivered 850 megawatts in Q1. What that means is that, as you can see, neither Shackelford, nor New Mexico, or Wisconsin, or Michigan were delivered in Q1.”
“And the technology that we'll be deploying there is Bloom fuel cells, which is by far the most environmentally friendly way that we can do on-site power generation.”
“But the intention of what we're trying to say is that, while there clearly are capital expenditures, it does not require Oracle to go out and find additional cash to do it.”
“Sometimes it's working with our suppliers, through different financing arrangements that allows us to pay for the capacity as the customers pay us. That's one mechanism.”
“There is an understanding of this model right now that the access to capital and different ways of funding it are a constraint, and the industry adapts to allocate that in the most efficient way possible.”
“Obviously, we also have a very large and very rapidly growing general purpose cloud business that we do not talk about quite as much. But that is growing rapidly and has great growth rates, great margins, and does require some capital, but not as much as these giant AI clusters.”
“During Q1 FY 2027, Oracle successfully completed the sale of $20 billion of common stock (before commissions) through an At-the-Market (ATM) equity program, as part of its previously disclosed capital investment program.”
“Free cash flow was negative $5 billion for Q1 as Oracle continued to execute on investments to support the growth of its Cloud Infrastructure business.”
“Cash used for capital expenditures increased from $8.5 billion in the first quarter of fiscal 2026 to $28.5 billion in the first quarter of fiscal 2027 primarily due to the expansion of our data centers.”
“As of August 31, 2026, our unconditional purchase and certain other obligations with terms of one year or greater, which were primarily related to long-term supply arrangements for purchasing components for cloud infrastructure assets and power supply arrangements for data centers, were as follows (in millions):”
Depreciation of the data center and AI compute build
7.3% to 12.1% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: direction only· motive: offensive· before LLMs: expanded· layer: compute
Depreciation was against a year earlier, and the filing attributes a rise of in infrastructure expenses to support of the cloud infrastructure offering (claims c76, c73). The finance chief says gross margin declined as expected and may flatten over the next couple of years; the infrastructure chief says GPUs four years old or more were renewed or resold at a premium of , which bears on useful life (claims c2, c13, c41). The size, , is the ledger’s own: depreciation above its level before the build, times an assumed AI share. It counts in the spend total; the capital expenditure behind it is shown and left out of totals.
Evidence: 8 quotes, 5 figures, 1 confound, 3 from before coverage
Depreciation of the servers, accelerators and data center assets bought for the AI build, the largest part of the infrastructure expenses the filing gives as the reason cloud and software expenses rose. It is earlier capital spending recognized over the life of the equipment: the cost of the build as the income statement carries it, so it counts in the spend total while the capital expenditure channel does not. The filing does not attribute it to AI.
Why this motive
Carried from the prior quarters: the cost of delivering separately priced AI capacity, for which management says its margin guidance still holds (claim c55).
Before LLMs: expanded
Depreciation was in fiscal 2024 and in its last quarter, before the build accelerated; the anchor call already described GPU capacity as profitable for an automated cloud. The size is the change AI made, not the whole line.
“Now, as far as GPUs, are GPUs a low-margin business? Not for a 100% automated cloud with very, very low costs.”
“We have invested in the rapid expansion of the Oracle Cloud by increasing existing data center capacity and adding data centers in new geographic locations to meet current and expected customer demand. We expect this trend will continue.”
Infrastructure expenses within the cloud and software business, year-over-year increase in constant currency · 2026-CQ3
Cloud and software expenses · 2026-CQ3
Cloud and software expenses, prior-year quarter · 2025-CQ3
Cloud and software expenses, increase on the prior-year quarter · 2026-CQ3
Price premium at which GPUs coming up for renewal were renewed or resold · 2026-CQ3
Reported line it is matched to
Depreciation was against , an increase of . The filing attributes higher cloud and software expenses to infrastructure expenses that support the cloud infrastructure offering.
2026-CQ3: 2025-CQ3:
What else could explain it
line composition: Depreciation covers all property and equipment, including general-purpose cloud and office assets.
Quotes
“Our gross margin did decline as expected, driven by impacts from ramping up our data centers and the acceleration of infrastructure revenue. However, this was offset in the quarter by lower operating costs and strong operating leverage tied to simplification and efficiency actions.”
“That business, by nature, has much lower R&D and sales associated with it, at least in a company like Oracle, where we can effectively gain from all of the R&D that is already been done and being done across the rest of the company.”
“Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are four years or older. We see a long, useful life with increasing value for the AI capacity we're deploying.”
“Excluding the unfavorable effects of currency rate fluctuations of 1% in the first quarter of fiscal 2027, the constant currency increase in expenses was primarily due to a $2.8 billion increase in infrastructure expenses, partially offset by a $240 million decrease in sales and marketing expenses in the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“Total margin as a percentage of revenues in constant currency decreased in the first quarter of fiscal 2027, relative to the first quarter of fiscal 2026, due to an increase in cloud and software business' total expenses driven by higher infrastructure expenses to support growth in our cloud infrastructure offering.”
“Depreciation expenses on property, plant and equipment during the three months ended August 31, 2026 and 2025 were $3.2 billion and $1.4 billion, respectively.”
Q1 2026direction only · our inference · offensive · $805mn to $1.34bn
Q2 2026direction only · our inference · offensive · $962mn to $1.60bn
Q3 2026direction only · our inference · offensive · $1.41bn to $2.35bn
cost of-revenue
Data center lease cost of the AI build
2.5% to 4.1% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: direction only· motive: offensive· before LLMs: expanded· layer: compute
Operating lease expense was against a year earlier. Lease commitments not yet commenced rose to , and unconditional obligations, now mainly component supply and data center power, to (claims c77, c78); these are future cost. The size, , is the ledger’s own: lease expense above the fiscal 2024 quarterly rate, times an assumed AI share. Power and the other running costs of the same sites are read in their own channel; management describes on-site fuel cells at one site and grid supply at another, both still in permitting (claim c53).
Evidence: 7 quotes, 2 figures, 2 confounds, 2 from before coverage
Operating lease cost of data centers leased from third-party developers for the build, inside cloud and software expenses. The much larger commitments for leases that have not yet commenced are recorded as a metric: they are future cost, not the quarter’s. Power and the other running costs of the same sites are read as their own channel. The payees are data center developers and landlords.
Why this motive
Carried from the prior quarters: a cost of delivering separately priced AI capacity.
Before LLMs: expanded
Operating lease expenses were in fiscal 2024, about a quarter, with of data center lease commitments not yet commenced at the end of that year. The size is the change AI made, not the whole line.
“We have invested in the rapid expansion of the Oracle Cloud by increasing existing data center capacity and adding data centers in new geographic locations to meet current and expected customer demand. We expect this trend will continue.”
“As of May 31, 2024, we have $22.9 billion of additional operating lease commitments, primarily for data centers, that are generally expected to commence between fiscal 2025 and fiscal 2027 and for terms of nine to fifteen years that were not reflected on our consolidated balance sheet as of May 31, 2024 or in the maturities table below.”
Additional lease commitments not yet commenced, substantially all data centers · as-of 2026-08-31
Unconditional purchase and other obligations, primarily component supply for cloud infrastructure assets and data center power · as-of 2026-08-31
Reported line it is matched to
Operating lease expense was against , an increase of . The filing says leases primarily relate to data centers and real estate facilities and does not name AI.
2026-CQ3: 2025-CQ3:
What else could explain it
line composition: Operating lease expense includes offices and data centers for general-purpose regions.
other: Finance lease expense is left out, so the figure understates the occupancy cost of the build; power is read in its own channel.
Quotes
“Our gross margin did decline as expected, driven by impacts from ramping up our data centers and the acceleration of infrastructure revenue. However, this was offset in the quarter by lower operating costs and strong operating leverage tied to simplification and efficiency actions.”
“And the technology that we'll be deploying there is Bloom fuel cells, which is by far the most environmentally friendly way that we can do on-site power generation.”
“Excluding the unfavorable effects of currency rate fluctuations of 1% in the first quarter of fiscal 2027, the constant currency increase in expenses was primarily due to a $2.8 billion increase in infrastructure expenses, partially offset by a $240 million decrease in sales and marketing expenses in the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“Total margin as a percentage of revenues in constant currency decreased in the first quarter of fiscal 2027, relative to the first quarter of fiscal 2026, due to an increase in cloud and software business' total expenses driven by higher infrastructure expenses to support growth in our cloud infrastructure offering.”
“As of August 31, 2026, we had $288 billion of additional lease commitments, substantially all related to data center arrangements, that are generally expected to commence between the second quarter of fiscal 2027 and fiscal 2029 and for terms of fifteen to nineteen years that were not reflected on our condensed consolidated balance sheets as of August 31, 2026 or in the maturities table above.”
“As of August 31, 2026, our unconditional purchase and certain other obligations with terms of one year or greater, which were primarily related to long-term supply arrangements for purchasing components for cloud infrastructure assets and power supply arrangements for data centers, were as follows (in millions):”
Q1 2026direction only · our inference · offensive · $255mn to $425mn
Q2 2026direction only · our inference · offensive · $337mn to $562mn
Q3 2026direction only · our inference · offensive · $481mn to $801mn
operations
Power, network and other running cost of the AI fleet
1.1% to 2.7% of the quarter’s revenue
Incremental total: counts in full.
our inferencedisclosure: described· motive: offensive· before LLMs: expanded· layer: compute
The filing gives a rise of in infrastructure expenses as the main reason cloud and software expenses rose and does not break it down or name AI (claim c73). Depreciation rose and operating lease expense on the year; the rest of the increase is power, network and other running cost. Unconditional obligations, now mainly component supply and data center power, stand at , and management describes on-site fuel cells for one large site (claims c78, c53). The size, , is the ledger’s own: that residual times an assumed AI share, taken against the prior-year quarter.
Evidence: 4 quotes, 4 figures, 3 confounds, 2 from before coverage
The cost of running the data centers of the build other than depreciation and lease cost: power, network and the other infrastructure expenses inside cloud and software expenses. The filing gives the year-over-year increase in infrastructure expenses as a single figure and does not break it down or name AI; power is disclosed only as purchase obligations. The size is the part of that increase not explained by the rise in depreciation and in operating lease cost, times an assumed AI share. The payees are utilities, network carriers and other suppliers.
Why this motive
Carried from the prior quarters: a cost of delivering separately priced AI capacity, for sites that need new power supply (claim c53).
Before LLMs: expanded
Running the data centers predates the build: the fiscal 2024 annual report gives higher technology infrastructure expenses as the main reason cloud costs grew and names the energy mix of the data center operations as a source of pressure, with no amount for power. Total operating expenses were in the last quarter of fiscal 2024. The size is the change AI made, not the whole line.
“In constant currency, our total cloud and license business’ expenses increased in fiscal 2024 relative to fiscal 2023 primarily due to higher technology infrastructure expenses to support the increase in our cloud and license business’ revenues.”
“Regulatory, market, carbon tax and competitive pressures regarding the greenhouse gas emissions and energy mix for our data center operations may also grow.”
Infrastructure expenses within the cloud and software business, year-over-year increase in constant currency · 2026-CQ3
Depreciation, increase on the prior-year quarter · 2026-CQ3
Operating lease expense, increase on the prior-year quarter · 2026-CQ3
Unconditional purchase and other obligations, primarily component supply for cloud infrastructure assets and data center power · as-of 2026-08-31
What else could explain it
line composition: Infrastructure expenses are not defined in the filing; the residual also holds third-party capacity, maintenance and data center staff, and company-wide depreciation is subtracted though part of it sits outside the line.
fx: The increase is stated in constant currency while the depreciation and lease figures subtracted are reported.
other: The residual is a small difference between large rounded figures; the range carries the rounding.
Quotes
“And the technology that we'll be deploying there is Bloom fuel cells, which is by far the most environmentally friendly way that we can do on-site power generation.”
“Excluding the unfavorable effects of currency rate fluctuations of 1% in the first quarter of fiscal 2027, the constant currency increase in expenses was primarily due to a $2.8 billion increase in infrastructure expenses, partially offset by a $240 million decrease in sales and marketing expenses in the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“Total margin as a percentage of revenues in constant currency decreased in the first quarter of fiscal 2027, relative to the first quarter of fiscal 2026, due to an increase in cloud and software business' total expenses driven by higher infrastructure expenses to support growth in our cloud infrastructure offering.”
“As of August 31, 2026, our unconditional purchase and certain other obligations with terms of one year or greater, which were primarily related to long-term supply arrangements for purchasing components for cloud infrastructure assets and power supply arrangements for data centers, were as follows (in millions):”
Matched line moved : the whole line, not this channel.
our inferencedisclosure: direction only· motive: offensive· before LLMs: expanded· layer: compute
Interest expense was against a year earlier, attributed by the filing to the senior notes issued in the prior fiscal year (claim c79). No debt was raised in the quarter; the company issued of common stock, which the release ties to its capital investment program, and says the new contracts do not change its plans to raise capital (claims c80, c65, c64). The size, , is the ledger’s own: interest above its level before the build, times an assumed share. The cost of the equity raised, dilution, is not an income statement item and is not sized.
Evidence: 6 quotes, 1 figure, 2 confounds, 1 from before coverage
Interest expense on the senior notes and other borrowings raised while capital spending exceeded operating cash flow. The filing attributes the rise in interest expense to higher average borrowings issued for general corporate purposes, which may include capital expenditures; management ties its capital raising to the AI data center build. The payees are bondholders and lenders.
Why this motive
Carried from the prior quarters: the cost of capital raised for contracted capacity; this quarter the raising was equity (claim c10).
Before LLMs: expanded
Debt and interest predate the build: of indebtedness at the end of fiscal 2024 and interest expense of in that year’s last quarter. The size is the change AI made, not the whole line.
“As of May 31, 2024, we had an aggregate of $86.9 billion of outstanding indebtedness that will mature between calendar year 2024 and calendar year 2061.”
Net proceeds of common stock issued through the at-the-market program in the quarter · 2026-CQ3
Reported line it is matched to
Interest expense was against , an increase of . The filing attributes it to higher average borrowings.
2026-CQ3: 2025-CQ3:
What else could explain it
line composition: Interest expense covers all debt, most of it issued before the build.
other: Customer prepayments with a financing component carry imputed interest, which the filing calls immaterial so far.
Quotes
“Lastly, we are quite pleased to announce that we completed our previously disclosed $20 billion at-the-market equity issuance in entirety during the Q1.”
“Sometimes it's working with our suppliers, through different financing arrangements that allows us to pay for the capacity as the customers pay us. That's one mechanism.”
“Based on the structuring of those new contracts, the Company confirms there is no incremental impact on its plans to raise capital. Since the end of Q4, Oracle also delivered more than 300,000 GPUs to its AI Cloud customers and almost triple the capacity delivered in Q4 FY26.”
“During Q1 FY 2027, Oracle successfully completed the sale of $20 billion of common stock (before commissions) through an At-the-Market (ATM) equity program, as part of its previously disclosed capital investment program.”
“Interest expense increased in the first quarter of fiscal 2027, relative to the first quarter of fiscal 2026, primarily due to higher average borrowings from the issuances of $43.0 billion of senior notes in fiscal 2026, partially offset by lower interest expense due to scheduled repayments of $8.1 billion of debt made during the first quarter of fiscal 2027 and full year of fiscal 2026.”
“During the first quarter ended August 31, 2026, we fully utilized the ATM Program and issued approximately 141 million shares of common stock under the ATM Program for net proceeds of $19.9 billion.”
Q1 2026direction only · our inference · offensive · $151mn to $302mn
Q2 2026direction only · our inference · offensive · $281mn to $561mn
Q3 2026direction only · our inference · offensive · $275mn to $550mn
vendor bill
AI coding tools and model usage in engineering
Not sized
The quarter is silent on the company’s own coding-tool use. The rise in computer equipment expense within research and development is a reported movement the filing does not attribute, and is not taken as the size.
inscrutabledisclosure: not mentioned· motive: efficiency· before LLMs: new
The call and the release say nothing this quarter about the company’s own use of AI coding tools; coding agents are discussed only as products and integrations. The filing reports computer equipment expenses in research and development up on the year without saying what they are for.
Evidence: 0 quotes, 1 figure
What the company pays for the AI coding tools and model usage of its own engineers, which management says it is adopting rapidly and which let it restructure product development into smaller teams. It sits inside research and development. Management praises two labs’ coding tools and says the company is adopting AI coding tools; it does not say whom the company pays or how much, and as a seller of the same capacity part of the use may run on its own cloud.
Why this motive
Carried from the prior quarter: tools adopted so that fewer engineers deliver more.
Before LLMs: new
No outside AI coding-tool bill is in the anchor: on the anchor call the founder describes generating application code with the company’s own low-code tool. Research and development was in the last quarter of fiscal 2024, with employees.
Figures
Research and development computer equipment expenses, year-over-year increase in constant currency · 2026-CQ3
By quarter
Q1 2026direction only · our inference · efficiency · $846k to $141mn
Q2 2026direction only · our inference · efficiency · $1.3mn to $129mn
Q3 2026not mentioned · inscrutable · efficiency
cost of-revenue
Model inference behind the AI embedded in the applications
0% to 0.09% of the quarter’s revenue
Incremental total: counts in full.
our inferencedisclosure: direction only· motive: product-defensive· before LLMs: new
Management states a volume for the first time: tokens consumed by AI production usage in Fusion alone during the quarter, with embedded AI usage up on the prior quarter (claims c22, c21). No cost is given. The size, , applies a public price class to the stated volume and scales it for the other suites; it is small beside every other spend channel. The counterparty is mixed: outside model providers and the company’s own cloud capacity, which no source splits. A token count is a count of AI work, so the state is directional.
Evidence: 3 quotes, 3 figures, 2 confounds
The cost of running third-party and open models for the agents and generative features that ship in the applications without a separate charge. It sits inside cloud and software expenses and is served largely from the company’s own infrastructure; management gives usage counts and no cost. The payees are model providers and the company’s own cloud, so the counterparty is mixed.
Why this motive
Carried from the prior quarters: inference served for features that ship with the regular updates at no separate charge (claim c21).
Before LLMs: new
The anchor reports no model inference cost: generative AI in the applications was newly announced and no usage or cost was given.
Figures
Tokens consumed by AI production usage across Fusion in the quarter · 2026-CQ3
Uses of embedded AI capabilities by customers in the quarter (a floor) · 2026-CQ3
Embedded AI usage, growth on the prior quarter · 2026-CQ3
What else could explain it
other: Much of the inference runs on the company’s own capacity, where the cost is depreciation and power already read in other channels.
line composition: Tokens are given for Fusion only; the health, NetSuite and industry suites are an assumed multiple.
Quotes
“Customers used our embedded AI capabilities more than 150 million times during the quarter, with usage growing 42% sequentially. Our AI agents executed more than 3.5 million times in production during the quarter, nearly doubling quarter- over- quarter.”
“Customers have over 2,300 AI agents in production, and that is up 90% quarter- over- quarter. Overall, AI production usage across Fusion alone consumed 900 billion tokens during the quarter.”
Q1 2026described · described, no size · product-defensive
Q2 2026described · described, no size · product-defensive
Q3 2026direction only · our inference · product-defensive · $450k to $18mn
back office
Severance from restructuring attributed in part to AI adoption
Not sized
The restructuring expense is the cost of the whole plan: the filing names the adoption of AI technologies among the purposes of the plan, beside strategic measures and other operational activities (claim c81), and no AI share is given, so the channel is not sized (the methodology rule for AI named beside another cause).
described, no sizedisclosure: bounded· motive: exploratory· before LLMs: expanded
The filing repeats that the restructuring plan improves efficiency in part through the adoption of artificial intelligence technologies across certain functions (claim c81). Restructuring expense fell to from a year earlier, with recorded to date under a plan of up to , and after the quarter end management added approximately to the plan for further actions (claims c83, c82). The plan’s cost bounds the AI part from above; no split is given. Because AI is named beside other purposes of the plan and no split is given, the channel is left unsized.
Evidence: 5 quotes, 5 figures, 1 confound, 1 from before coverage
The part of restructuring expense, mostly employee severance, that follows from AI: management says AI code generation let it restructure product development into smaller teams, and from the annual report the filing names the adoption of AI technologies across certain functions among the purposes of the restructuring plan. The filing gives the plan’s cost and no AI share.
Why this motive
Carried from the prior quarter: severance for workforce reductions the filing ties in part to the adoption of AI (claim c81). The result is credited to several causes together, so it does not meet the efficiency tell for any one of them (the methodology's motive table): exploratory.
Before LLMs: expanded
Restructuring plans predate AI adoption: of restructuring expense in fiscal 2024 under a plan of up to , attributed to acquisitions and other operational activities. The size is the change AI made, not the whole line.
“During fiscal 2024, our management approved, committed to and initiated plans to restructure and further improve efficiencies in our operations due to our acquisitions and certain other operational activities (2024 Restructuring Plan).”
Fiscal 2026 restructuring plan costs to date · as-of 2026-08-31
Total estimated cost of the fiscal 2026 restructuring plan before the supplement (up to) · as-of 2026-08-31
Supplement to the fiscal 2026 restructuring plan approved after the quarter end (approximately) · as-of 2026-09-11
What else could explain it
transformation program: The plan also implements strategic measures and other operational activities; the filing gives no AI share.
Quotes
“During fiscal 2026, our management approved, committed to, initiated and further supplemented plans to restructure to implement certain strategic measures and further improve operational efficiencies, including through the adoption and integration of artificial intelligence technologies across certain functions and other operational activities (2026 Restructuring Plan).”
“activities in the first quarter of each of fiscal 2027 and 2026 primarily related to the 2026 Restructuring Plan that our management approved, committed to and initiated during fiscal 2026 to implement certain strategic measures and further improve operational efficiencies, including through the adoption and integration of artificial intelligence technologies across certain functions and other operational activities.”
“Subsequent to August 31, 2026, our management supplemented the 2026 Restructuring Plan by approximately $700 million to reflect additional actions that we expect to take.”
“We recorded $167 million and $415 million of restructuring expenses in connection with the 2026 Restructuring Plan for the three months ended August 31, 2026 and 2025, respectively.”
“Certain of the cost savings realized pursuant to the 2026 Restructuring Plan initiatives were offset by investments in resources and geographies that we believe better address the development, marketing, sale and delivery of our cloud-based offerings, including investments in the development and delivery of our second-generation cloud infrastructure.”
Forward-deployed engineers for the AI Data Platform and agent studio
0.01% to 0.2% of the quarter’s revenue
Incremental total: counts in full.
our inferencedisclosure: described· motive: exploratory· before LLMs: expanded
The channel opens this quarter. Asked whether it would place engineers with customers for the AI Data Platform, management says it is already doing so, for the platform and for the agent studio (claim c36). No number is given. Services business expenses were and fell on the year, with employee-related expense down (claim c85), so the investment is not visible in the line. The size, , is the ledger’s own, an assumed share of services expenses.
Evidence: 3 quotes, 2 figures, 2 confounds, 1 from before coverage
Engineers the company places with customers to get the AI Data Platform and the agent studio into production. Management says it is already investing in them and gives no number; the cost would sit in services or sales expenses.
Why this motive
Management says it is already investing in forward-deployed engineers and calls the offering necessary, for products that have no disclosed revenue (claims c36, c37): the investing-ahead tell.
Before LLMs: expanded
Deployment help was an established services line at the anchor, with services employees at the end of fiscal 2024 and consulting described as helping customers deploy the cloud offerings. Engineers placed with customers for AI products are not in the anchor. The size is the change AI made, not the whole line.
“consulting services, which are designed to help our customers and global system integrator partners more successfully architect and deploy our cloud and license offerings, including IT strategy alignment, enterprise architecture planning and design, implementation, integration, application development, security assessments and ongoing software enhancements and upgrades.”
Services business employee-related expenses, year-over-year decrease in constant currency · 2026-CQ3
What else could explain it
line composition: The cost may sit in services, sales or research and development; the filing does not say.
transformation program: Services headcount is falling under the restructuring plan at the same time.
Quotes
“We are already investing in deploying forward deployed engineers at our customers. That's true for the AI Data Platform. It's also true for our Fusion Agentic Studio, as we see them really as a combination platform running on a single control plane in OCI.”
“the constant currency decrease in services expenses was primarily due to a $107 million decrease in employee-related expenses, partially offset by a $28 million increase in bad debt expenses and a $27 million increase in external contractor expenses, in each case during the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
Engineering cost avoided through AI code generation
Not sized
The fall in research and development comes under a restructuring plan that names AI adoption beside strategic measures and other operational activities (claim c81), and no AI share of it is given, so the channel is not sized (the methodology rule for AI named beside another cause).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: exploratory· before LLMs: expanded
Research and development was against , a change of while revenue grew ; the filing attributes it to less employee-related expense, partly offset by more computer equipment expense (claim c84). The call does not attribute the saving to AI this quarter: the finance chief speaks of simplification and efficiency actions (claim c2). The link to AI rests on the filing’s description of the restructuring plan and on the claims of the two prior quarters. The line fell by on the year, but it falls under a plan with several purposes and nothing separates AI’s part, so the channel is left unsized.
Evidence: 4 quotes, 5 figures, 2 confounds, 2 from before coverage
Research and development cost the company says it avoids because AI code generation lets smaller teams build more software with fewer people. The displaced cost is engineering staff, in research and development, where the filing explains movements by employee-related and computer equipment expenses without naming AI.
Why this motive
The displaced line shrinks: research and development fell on the year with employee-related expense down , under a plan the filing ties in part to AI adoption (claims c84, c81). The headcount tell of the efficiency motive. The result is credited to several causes together, so it does not meet the efficiency tell for any one of them (the methodology's motive table): exploratory.
Before LLMs: expanded
Generating code to save programmer labor was already described at the anchor, with the company’s own low-code tool. Research and development was in the last quarter of fiscal 2024, with employees. The size is the change AI made, not the whole line.
“During fiscal 2024, our management approved, committed to and initiated plans to restructure and further improve efficiencies in our operations due to our acquisitions and certain other operational activities (2024 Restructuring Plan).”
“With Cerner, the rewrite of Cerner, it's not armies of programmers that are gonna be rewriting this. We are generating the new Millennium software using APEX, and that's also going to save us a lot of human labor and generate higher quality code and higher quality user interfaces and better security, all at once.”
Research and development, prior-year quarter · 2025-CQ3
Research and development, year-over-year change · 2026-CQ3
Research and development employee-related expenses, year-over-year decrease in constant currency · 2026-CQ3
Research and development computer equipment expenses, year-over-year increase in constant currency · 2026-CQ3
Reported line it is matched to
Research and development was against , a change of , or . The filing attributes the decrease to lower employee-related expenses, partly offset by higher computer equipment expenses.
2026-CQ3: 2025-CQ3:
What else could explain it
transformation program: The restructuring plan has several stated purposes, and the filing says some of its savings are reinvested.
operating leverage: Revenue growth comes mostly from infrastructure, which management says needs little added research and development.
Quotes
“Our gross margin did decline as expected, driven by impacts from ramping up our data centers and the acceleration of infrastructure revenue. However, this was offset in the quarter by lower operating costs and strong operating leverage tied to simplification and efficiency actions.”
“During fiscal 2026, our management approved, committed to, initiated and further supplemented plans to restructure to implement certain strategic measures and further improve operational efficiencies, including through the adoption and integration of artificial intelligence technologies across certain functions and other operational activities (2026 Restructuring Plan).”
“due to a $145 million decrease in employee-related expenses, partially offset by a $92 million increase in computer equipment expenses, in each case in the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“Certain of the cost savings realized pursuant to the 2026 Restructuring Plan initiatives were offset by investments in resources and geographies that we believe better address the development, marketing, sale and delivery of our cloud-based offerings, including investments in the development and delivery of our second-generation cloud infrastructure.”
Q1 2026direction only · our inference · narrative-defensive · $0
Q2 2026direction only · described, no size · exploratory
Q3 2026direction only · described, no size · exploratory
back office · cheap to verify
Staff cost avoided through AI adoption outside engineering
Not sized
Management credits the lower costs to efficiency actions and a simpler go-to-market model, and the filing names AI adoption only among the purposes of the restructuring plan (claims c2, c81); no AI share is given, so the channel is not sized (the methodology rule for AI named beside another cause).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: exploratory· before LLMs: expanded
Sales and marketing was against , a change of , and the filing reports employee-related expense in services down (claims c73, c85). Management attributes lower operating costs to simplification and efficiency actions, and separately describes AI tooling that cuts implementation times from months to weeks (claims c2, c28, c29). No saving is attributed to AI in dollars. The falls are credited to efficiency actions beside a plan that names AI among several purposes, with nothing to separate AI’s part, so the channel is left unsized.
Evidence: 9 quotes, 5 figures, 3 confounds, 3 from before coverage
Cost avoided in sales, services, support and administrative functions through the adoption of AI, which the annual report says has resulted in reductions to the workforce. The lines that fall are sales and marketing and services expenses, which management attributes to efficiency actions and a simpler go-to-market model without giving an AI share.
Why this motive
Displaced lines shrink: sales and marketing fell on the year and services employee-related expense fell , under a plan the filing ties in part to AI adoption, while management describes AI tooling that shortens implementations (claims c73, c85, c81, c28). The result is credited to several causes together, so it does not meet the efficiency tell for any one of them (the methodology's motive table): exploratory.
Before LLMs: expanded
Labor-saving automation and falling headcount predate AI adoption: the anchor call described the data centers as fully automated to save labor, and sales and marketing expense, in the last quarter of fiscal 2024, was already falling on lower headcount. The company had employees at the end of that year. The size is the change AI made, not the whole line.
“During fiscal 2024, our management approved, committed to and initiated plans to restructure and further improve efficiencies in our operations due to our acquisitions and certain other operational activities (2024 Restructuring Plan).”
“These constant currency expense increases were partially offset by lower sales and marketing expenses, which decreased primarily due to lower employee related expenses due to lower headcount.”
Sales and marketing, prior-year quarter · 2025-CQ3
Sales and marketing, year-over-year change · 2026-CQ3
Sales and marketing expenses of the cloud and software business, year-over-year decrease in constant currency · 2026-CQ3
Services business employee-related expenses, year-over-year decrease in constant currency · 2026-CQ3
Reported line it is matched to
Sales and marketing was against , a decrease of . The filing reports the decrease without giving a cause in this quarter.
2026-CQ3: 2025-CQ3:
What else could explain it
transformation program: Management credits simplification and efficiency actions; none of the decrease is attributed to AI on the call.
line composition: Only sales and marketing is used as the base; the services saving management’s tooling points to is not counted.
other: A simpler go-to-market model and efficiency actions are named for the same cost falls, and the plan names AI among several purposes.
Quotes
“Our gross margin did decline as expected, driven by impacts from ramping up our data centers and the acceleration of infrastructure revenue. However, this was offset in the quarter by lower operating costs and strong operating leverage tied to simplification and efficiency actions.”
“At AI World in October, we will unveil a new agentic AI accelerator poised to redefine how customers deploy Oracle applications faster, simpler, and at a dramatically lower cost. Working alongside Oracle and customer teams, AI agents will automate and orchestrate implementation at an unprecedented scale, compressing SaaS deployments from years to months, and months to weeks.”
“AI has given us a tremendous ability to accelerate those go-lives, and we have some proof points for the tooling that we have rolled out to date, and we will roll out a bunch more of it at Oracle AI World.”
“In our Oracle NetSuite applications, we have seen early customers leveraging these AI tools coming down from double-digit months down to single-digit weeks in order to be able to go live in production.”
“Number two, in some of these very complex industries, there are ramps associated with these go-lives, and it allows us to unlock the ramp and recognize revenue more quickly than we have in the manual implementation piece.”
“Excluding the unfavorable effects of currency rate fluctuations of 1% in the first quarter of fiscal 2027, the constant currency increase in expenses was primarily due to a $2.8 billion increase in infrastructure expenses, partially offset by a $240 million decrease in sales and marketing expenses in the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“During fiscal 2026, our management approved, committed to, initiated and further supplemented plans to restructure to implement certain strategic measures and further improve operational efficiencies, including through the adoption and integration of artificial intelligence technologies across certain functions and other operational activities (2026 Restructuring Plan).”
“the constant currency decrease in services expenses was primarily due to a $107 million decrease in employee-related expenses, partially offset by a $28 million increase in bad debt expenses and a $27 million increase in external contractor expenses, in each case during the first quarter of fiscal 2027 relative to the first quarter of fiscal 2026.”
“Certain of the cost savings realized pursuant to the 2026 Restructuring Plan initiatives were offset by investments in resources and geographies that we believe better address the development, marketing, sale and delivery of our cloud-based offerings, including investments in the development and delivery of our second-generation cloud infrastructure.”
Q2 2026direction only · described, no size · exploratory
Q3 2026direction only · described, no size · exploratory
Revenue arriving through AI8 channels · $3.28bn to $5.40bn sized · $5.0mn to $5.40bn incremental · 6 not sized
product revenue
AI infrastructure: GPU and accelerator capacity sold for training and inference
16.9% to 27% of the quarter’s revenue
Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.
our inferencedisclosure: quantified· motive: offensive· before LLMs: expanded· layer: compute
Management gives capacity and usage: megawatts with more than GPUs delivered since the prior quarter, utilization of , and renewals or resales at a premium of (claims c38, c40, c41). It gives no AI infrastructure revenue; cloud infrastructure revenue as a whole was , up , which the finance chief ties to record megawatts brought online and to database services (claim c1). The size is the ledger’s own: , of cloud infrastructure revenue, the prior-quarter estimate rolled forward by an assumed share of the sequential increase of . Remaining performance obligations rose to , with expected as revenue within twelve months. For the first time the sources point at a buyer: one site took up to of the quarter’s GPUs, and a model released by OpenAI is said to have been trained there (claims c42, c43). Asked whether its customers are AI labs, management answers that demand is broad based, from start-ups to investment-grade companies (claim c59). The counterparty is mixed: AI labs that train and serve models, and other advanced AI customers, which management does not split.
Evidence: 23 quotes, 20 figures, 4 confounds, 4 from before coverage
Revenue from accelerator capacity that customers rent to train and run AI models, the part of cloud infrastructure revenue management calls AI infrastructure. Management describes the buyers as model vendors and advanced AI customers and names none; the annual report says no single customer reaches its disclosure threshold for a share of total revenue. The revenue level is not disclosed: management gives a growth rate, a gross margin, megawatts and GPUs delivered, and the contract backlog. Remaining performance obligations are a backlog to be earned over many years and are recorded as a metric, never as the size.
Why this motive
A separately sold AI offering with stated capacity, usage and price movement: megawatts delivered, utilization of and renewals at a premium of (claims c38, c40, c41).
Before LLMs: expanded
At the anchor this money was AI training capacity in the second-generation cloud: AI development companies had contracted for more than , and the annual report listed GPU compute in the infrastructure catalog. No revenue was given for it; all of infrastructure cloud revenue was in Q1 of fiscal 2024, up , and remaining performance obligations were at the end of that year. Capacity rented on accelerators predates LLMs and not every buyer is an AI lab, so the tie-break gives expanded; no figure for one quarter of the same activity before AI can be traced. The size is the whole of an activity that existed before.
“As of today, AI development companies have signed contracts to purchase more than $4 billion of AI training capacity in Oracle's Gen 2 Cloud. That's twice as much AI training as we had booked at the end of the last Q4.”
“Using Oracle technologies, our customers build, deploy, run, manage and support their internal and external products, services and business operations, including, for example, an artificial intelligence (AI) product company that uses Oracle Cloud Infrastructure (OCI) to build and serve generative AI models;”
“OCI compute services range from virtual machines to graphics processing unit-based offerings to bare metal servers and include options for high I/O workloads and high performance computing.”
AI capacity delivered to customers since the end of the prior quarter, megawatts · 2026-CQ3
GPUs delivered to customers since the end of the prior quarter (a floor) · 2026-CQ3
GPUs delivered at the Abilene site in the quarter · 2026-CQ3
Abilene deliveries as a share of GPUs delivered in the quarter (an upper bound, since the total is a floor) · 2026-CQ3
GPU utilization · 2026-CQ3
Price premium at which GPUs coming up for renewal were renewed or resold · 2026-CQ3
Contract value of additional AI contracts closed in the quarter without requiring capital from the company (a floor; contract value, not revenue) · 2026-CQ3
Remaining performance obligations a year earlier · as-of 2025-08-31
Remaining performance obligations, increase on the prior quarter · 2026-CQ3
Share of remaining performance obligations expected as revenue within twelve months · as-of 2026-08-31
Share of remaining performance obligations expected as revenue in months thirteen to thirty-six · as-of 2026-08-31
Remaining performance obligations expected as revenue within twelve months, per quarter on average · as-of 2026-08-31
What else could explain it
other: With no growth rate or level for two quarters, the estimate rests on the first-quarter estimate rolled forward twice.
line composition: Cloud infrastructure growth is attributed jointly to new megawatt capacity and database services; the split is not disclosed.
other: Deliveries are concentrated in one site and one customer; the next large sites are not yet delivering and two face permitting and power questions.
other: Remaining performance obligations depend on capacity being built and on customers meeting obligations over many years.
Quotes
“Cloud infrastructure revenue for Q1 was $7.4 billion, up 121%, reflecting strong execution as we brought record levels of new megawatt capacity online, supported by a continued strong demand environment for compute and our database services.”
“That business, by nature, has much lower R&D and sales associated with it, at least in a company like Oracle, where we can effectively gain from all of the R&D that is already been done and being done across the rest of the company.”
“We delivered 850 megawatts of AI capacity containing more than 300,000 GPUs to customers since the end of Q4. Delivery in Q1 is almost three times what we delivered in all of Q4 and 73% of the total capacity we delivered last fiscal year.”
“Customer demand continues to support this investment. We closed more than $30 billion of additional AI contracts in Q1 without requiring additional capital from Oracle.”
“Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are four years or older. We see a long, useful life with increasing value for the AI capacity we're deploying.”
“We delivered 131,000 GPUs there in Q1, 1.9 times the volume delivered in Q4. Six of the eight campus buildings, representing 618 megawatts and 75% of total capacity, have now been delivered to the customer.”
“We are delivering data center and GPU capacity at a pace that would've seemed impossible only a year ago. Customers are signing new contracts, renewing capacity at higher prices, and keeping the fleet almost fully utilized.”
“We just delivered 850 megawatts in Q1. What that means is that, as you can see, neither Shackelford, nor New Mexico, or Wisconsin, or Michigan were delivered in Q1.”
“To the question about impact on FY 2027 revenue, neither of these sites will have any impact into our previously stated FY 2027 revenue or earnings guidance.”
“In terms of the types of customers and where we see that demand, it is really broad based. It doesn't matter if it is a startup or the most valuable investment-grade companies.”
“Customer demand for AI Cloud Training and Inferencing Services continues to grow faster than supply. Oracle booked more than $30 billion of additional AI cloud contracts in Q1 increasing its RPO to $664 billion.”
“Based on the structuring of those new contracts, the Company confirms there is no incremental impact on its plans to raise capital. Since the end of Q4, Oracle also delivered more than 300,000 GPUs to its AI Cloud customers and almost triple the capacity delivered in Q4 FY26.”
“Total quarterly revenues increased 30% to $19.3 billion, reflecting strong execution in our infrastructure business, with the delivery of 850MW additional datacenter capacity.”
“Remaining performance obligations were $664 billion as of August 31, 2026, of which we expect to recognize approximately 13% as revenues over the next twelve months, 37% over the subsequent month 13 to month 36, 34% over the subsequent month 37 to month 60 and the remainder thereafter.”
“The increase in remaining performance obligations as of August 31, 2026 in comparison to August 31, 2025 was primarily attributable to certain significant cloud contracts that were entered into during the period.”
AI contracts funded by the customer: prepayments and customer-supplied hardware
Not sized
No revenue from these contracts is disclosed. What they earn is recognized inside AI infrastructure revenue, which is sized in its own channel, and a contract value or backlog is never a quarter’s size.
described, no sizedisclosure: quantified· motive: offensive· before LLMs: expanded· layer: compute
More than of additional AI contract value was closed without requiring capital from the company, most of it prepaid or with customer-supplied hardware, and management says this new backlog does not affect revenue until fiscal 2028 or later (claims c39, c4). The cash is now large: of prepayments with a significant financing component in the quarter, and non-current deferred revenue of against a quarter earlier (claim c72). No revenue from these contracts is given; what they earn is counted inside AI infrastructure revenue, so this channel is not sized. Funding stays mixed, as read since the first covered quarter: management now describes the payers as start-ups or established companies that have raised money, and demand as broad based up to the most valuable investment-grade companies (claims c58, c59), which is the first direct support for that reading and not a change in it.
Evidence: 16 quotes, 6 figures, 2 confounds, 1 from before coverage
Large AI infrastructure contracts in which the customer pays for the accelerators up front or buys them and supplies them, so that the company builds and operates the cluster without raising the capital itself. Management reports the contract value signed under this model and, from fiscal year end, the cash prepayments received. The cash is deferred revenue, and the revenue arrives later inside AI infrastructure revenue, where it is counted; hardware a customer supplies is never the company’s revenue. The channel is therefore not sized: contract value, prepayments and deferred revenue are recorded as metrics, and no backlog or contract value is spread into a quarter. Management calls the buyers broad based, from start-ups to investment-grade companies, and names none.
Why this motive
Carried from the prior quarters, with new stated levels: more than of such contracts closed and of prepayments received in the quarter (claims c39, c72).
Before LLMs: expanded
A prepaid commercial model existed at the anchor: cloud infrastructure was typically sold for a prepaid fee drawn down as services were consumed, inside remaining performance obligations of , of which was due within twelve months. Prepayment of the hardware itself, and hardware supplied by the customer, are not in the anchor. The size is the whole of an activity that existed before.
“We typically charge a prepaid fee that is decremented as the OCI services are consumed by the customer over a stated time period.”
Contract value of additional AI contracts closed in the quarter without requiring capital from the company (a floor; contract value, not revenue) · 2026-CQ3
Customer prepayments with a significant financing component received in the quarter · 2026-CQ3
Net cash outlay for capital expenditures (capital expenditures less supplier financing flows and customer prepayments) · 2026-CQ3
Deferred revenues, non-current · as-of 2026-08-31
Deferred revenues, non-current, at the prior fiscal year end · as-of 2026-05-31
Remaining performance obligations, increase on the prior quarter · 2026-CQ3
What else could explain it
other: Contract value and cash prepayments are not revenue; start dates and terms are not disclosed.
other: Management declines to say what kinds of customer these are; the funding reading rests on its description of the model, not on named payers.
Quotes
“The last point I will make on our financial highlights is that our Remaining Performance Obligations, or RPO, increased $26 billion from Q4.”
“The vast majority of those new contracts were via prepay or bring your own hardware or similar mechanic, so will not require incremental capital from Oracle. Also, that new RPO will not impact our CapEx or revenues until fiscal 2028 or beyond.”
“We drove record cash flow from operations of $23 billion in Q1, again reflecting our strong execution against a backdrop of strong demand, as well as collection of customer prepayments.”
“Our CapEx for the quarter was $28 billion, leading to negative free cash flow of $5 billion. Our net cash CapEx, so net of prepayments, was $18 billion for the quarter.”
“Customer demand continues to support this investment. We closed more than $30 billion of additional AI contracts in Q1 without requiring additional capital from Oracle.”
“I think that we have to separate out in our minds what Oracle spends as CapEx, uncouple that directly from how we think about how the business can grow.”
“But the intention of what we're trying to say is that, while there clearly are capital expenditures, it does not require Oracle to go out and find additional cash to do it.”
“Sometimes it's working with our suppliers, through different financing arrangements that allows us to pay for the capacity as the customers pay us. That's one mechanism.”
“A third option is that the customer has been able to raise money. Maybe it is a startup, maybe it is an established company, and says, "Hi, I would like to pay you upfront as a prepayment, and in return, that doesn't require you to front the cash to go out and spend your dollars on that CapEx."”
“In terms of the types of customers and where we see that demand, it is really broad based. It doesn't matter if it is a startup or the most valuable investment-grade companies.”
“There is an understanding of this model right now that the access to capital and different ways of funding it are a constraint, and the industry adapts to allocate that in the most efficient way possible.”
“Customer demand for AI Cloud Training and Inferencing Services continues to grow faster than supply. Oracle booked more than $30 billion of additional AI cloud contracts in Q1 increasing its RPO to $664 billion.”
“Based on the structuring of those new contracts, the Company confirms there is no incremental impact on its plans to raise capital. Since the end of Q4, Oracle also delivered more than 300,000 GPUs to its AI Cloud customers and almost triple the capacity delivered in Q4 FY26.”
“Remaining performance obligations were $664 billion as of August 31, 2026, of which we expect to recognize approximately 13% as revenues over the next twelve months, 37% over the subsequent month 13 to month 36, 34% over the subsequent month 37 to month 60 and the remainder thereafter.”
“During the first quarter of fiscal 2027, we received $11.4 billion of prepayments from customers that included a significant financing component. No prepayments from customers that included a significant financing component were received during the first quarter of fiscal 2026.”
Q1 2026quantified · described, no size · offensive
Q2 2026quantified · described, no size · offensive
Q3 2026quantified · described, no size · offensive
product revenue
General-purpose cloud services bought alongside AI accelerators
Not sized
The quarter is silent on this channel; the share of spend management gave is now two quarters old and is not carried into a quarter in which nothing is said.
inscrutabledisclosure: not mentioned· motive: offensive· before LLMs: expanded· layer: compute
Nothing in the call, the release or the filing addresses the general-purpose services AI customers purchase around their clusters or the share of spend given in the first covered quarter. Management does speak of the general-purpose cloud business as a whole, which is read under the halo channel.
Evidence: 0 quotes, 1 from before coverage
General-purpose compute, storage, load balancing, identity and security services that AI infrastructure customers purchase around their accelerator clusters. Management gives the typical share of an AI customer’s total spend that goes to these services and says they carry higher margins. The sources do not say whether this spend is counted inside the AI infrastructure revenue growth figure.
Why this motive
Carried from the prior quarter; the quarter is silent on services bought alongside accelerators.
Before LLMs: expanded
The services themselves are the ordinary cloud infrastructure catalog of the anchor, compute, storage and networking sold beside GPU offerings, inside infrastructure cloud revenue of in Q1 of fiscal 2024. The anchor gives no attach rate for them at AI customers. The size is the change AI made, not the whole line.
“OCI compute services range from virtual machines to graphics processing unit-based offerings to bare metal servers and include options for high I/O workloads and high performance computing.”
Cloud and multicloud database demand attributed to AI
Not sized
Management credits database growth to AI adoption and to the availability of regions in partner clouds (claims c44, c45) and gives no measure of AI’s part, so the channel is not sized (the methodology rule for AI named beside another cause).
described, no sizedisclosure: direction only· motive: narrative-defensive· before LLMs: relabelled
Multicloud database revenue grew , which management ties to the completed regional footprint in partner clouds, and it expects the AI Data Platform to help drive further growth (claims c44, c45, c35). No level, no cloud database growth rate and no AI-attributed measure is given. Growth is credited to AI beside multicloud region availability with nothing to separate AI’s part, so the channel is left unsized; as a relabelled channel it would add nothing to the incremental total in any case.
Evidence: 4 quotes, 2 figures, 2 confounds, 3 from before coverage
The part of cloud database and multicloud database revenue that management attributes to customers adopting AI: moving data to the cloud to reach AI features, vector search and agent access. The product is the existing database, now named Oracle AI Database. Management reports growth rates for cloud database and multicloud database revenue and no AI-attributed measure; the dollars sit inside cloud infrastructure revenue.
Why this motive
Carried from the prior quarters: growth rates for the product and no AI-attributed measure (claims c44, c35).
Before LLMs: relabelled
The same product was on offer at the anchor as Oracle Database and Autonomous Database, with a vector database announced and the multicloud partnership with Microsoft being expanded. Cloud database services grew in Q1 of fiscal 2024, faster than in the covered quarters, and management already called them a growth leg driven by on-premise databases migrating to the cloud.
“Very importantly, as on-premise databases migrate to the cloud, we expect these cloud database services will be the third leg of revenue growth alongside strategic SaaS and Gen 2 OCI Cloud Services.”
“So we think this is very good for our database business, and Oracle's new vector database will contain highly specialized training data like electronic health records, while keeping that data anonymized and private, yet still training the specialized models that can help doctors improve their diagnostic capability and their treatment prescriptions for cancer and heart disease and all sorts of other diseases.”
“We will be substantially expanding our existing multi-cloud partnership with Microsoft by making it easier for Microsoft Azure customers to buy and use the latest Oracle Cloud database technology in combination with Microsoft Azure Cloud Services.”
Cloud infrastructure revenue other than the AI infrastructure estimate · 2026-CQ3
What else could explain it
relabel: The product is the existing database under the name Oracle AI Database.
other: Multicloud growth follows from regions opening in partner clouds, which management gives as the driver.
Quotes
“Cloud infrastructure revenue for Q1 was $7.4 billion, up 121%, reflecting strong execution as we brought record levels of new megawatt capacity online, supported by a continued strong demand environment for compute and our database services.”
“Our Database Cloud business is also growing quickly. Multicloud database revenue grew 353% year over year, and multicloud customers grew 180% year- over- year.”
“As we continue to invest in making that available in every cloud, in every region, we think the AI Data Platform will help drive that growth as well.”
Q1 2026direction only · described, no size · narrative-defensive
Q2 2026direction only · described, no size · narrative-defensive
Q3 2026direction only · described, no size · narrative-defensive
customer cohort
Sovereign, dedicated and general-purpose cloud demand attributed to the AI build
Not sized
Management ties sovereign and partner-cloud demand to AI as well as to sovereignty (claim c61), describes a general-purpose cloud business growing in its own right (claim c62) and credits the applications as lead generation (claim c32). Nothing separates AI's part, so the channel gets no ballpark on a judgment share. Cloud infrastructure revenue outside AI capacity, , is the ceiling.
described, no sizedisclosure: described· motive: narrative-defensive· before LLMs: relabelled
Management says sovereign and partner-operated cloud demand is tied to AI because those customers deploy GPU capacity, describes a rapidly growing general-purpose cloud business it does not usually discuss, and calls the applications business lead generation for infrastructure (claims c61, c62, c32). No figure is given. The channel is unsized: AI is named beside sovereignty and the applications; cloud infrastructure revenue outside AI capacity, , is the ceiling. The former estimate, , is no longer the size.
Evidence: 3 quotes, 1 figure, 3 confounds, 2 from before coverage
Demand for sovereign regions, dedicated regions, partner-operated clouds and general-purpose cloud workloads that management attributes to its AI infrastructure position, which it calls a halo effect. No figure separates it from the rest of cloud infrastructure revenue.
Why this motive
Carried from the prior quarters: links asserted between AI and sovereign or general-purpose demand, with no figure (claims c61, c32).
Before LLMs: relabelled
Sovereign and dedicated regions were established offerings at the anchor: dedicated regions were live in September 2023 with demand said to be increasing, and the annual report already described the sovereign cloud as a way to offer sovereign AI. No revenue is given for them.
“This capability now also allows us to offer sovereign AI to customers who want the latest in AI innovations while operating within their regulatory environments;”
“We also have nine dedicated regions live and 11 more planned, nine security regions, and 12 EU sovereign regions live with increasing demand for more of each.”
Cloud infrastructure revenue other than the AI infrastructure estimate · 2026-CQ3
What else could explain it
relabel: Sovereign and dedicated regions were sold before the AI build.
other: GPU capacity sold to sovereign customers may already sit inside AI infrastructure revenue.
other: Management names the pull on other business together with the applications, sovereignty and partner-operated clouds beside AI; nothing separates AI's part.
Quotes
“One other thing I will mention about our SaaS business is that our SaaS business is also a wonderful lead generation business for our IaaS business.”
“Also, it is actually tied to AI as well, because many of those customers, we offer GPU capabilities for those customers, and they deploy that for those on sovereign workloads.”
“Obviously, we also have a very large and very rapidly growing general purpose cloud business that we do not talk about quite as much. But that is growing rapidly and has great growth rates, great margins, and does require some capital, but not as much as these giant AI clusters.”
Q1 2026described · described, no size · narrative-defensive
Q2 2026described · described, no size · narrative-defensive
Q3 2026described · described, no size · narrative-defensive
product revenue · expensive to verify
AI agents embedded in Fusion and industry applications at no additional cost
Not sized
The measures are counts of uses, agents in production and tokens (claims c21, c22), and the features arrive with the regular updates at no separate charge (claim c26). A count sizes nothing and a product with no price is read described and unsized (the methodology rules for counts and for products with no price). Cloud applications revenue, , is the ceiling.
described, no sizedisclosure: direction only· motive: product-defensive· before LLMs: relabelled
Management gives usage for the first time: more than uses of embedded AI in the quarter, up on the prior quarter, over agents in production and tokens consumed in Fusion alone (claims c21, c22). These measure use of features delivered with the regular updates, not revenue. Cloud applications revenue was , up , with Fusion up (claim c17). The release presents an all-new agentic health care management and records system (claim c67); it replaces the patient care system the company already sells and the sources do not say it carries a separate charge, so it is read here rather than as priced capacity. The channel is unsized: the features carry no price and the measures are counts. The former estimate, , is no longer the size. Agent, use and token counts are counts of AI work, so the state is directional.
Evidence: 14 quotes, 9 figures, 3 confounds, 2 from before coverage
AI agents and generative features built into the Fusion, NetSuite and industry application suites and delivered in the regular quarterly updates without a separate charge. Any revenue effect is in cloud applications revenue, through retention, competitive wins and faster go-lives; management reports agent counts and usage, not revenue. The work the agents touch includes financial close, banking compliance and clinical records, where an error is costly. The agentic health records system announced from fiscal Q4 2026 replaces the patient care system the company already sells, and the sources do not say it carries a separate charge, so it is read here rather than under priced agentic capacity.
Why this motive
Carried from the prior quarters: the AI arrives with the regular application updates (claim c26), and the usage management now reports is of features that carry no separate charge.
Before LLMs: relabelled
At the anchor the cloud applications were already described as natively incorporating AI, machine learning and digital assistants, and management was using generative AI to differentiate Fusion, NetSuite and the industry applications. Cloud applications revenue was in Q1 of fiscal 2024, growing , faster than in the covered quarters.
“Next week, we have Oracle CloudWorld, which will showcase the latest innovations, including AI on OCI, the progress of Oracle Autonomous Database, our multi-cloud strategy, the use of Oracle Analytics throughout our portfolio to drive better decision-making, and the use of generative AI to differentiate Fusion, NetSuite, and our industry applications.”
“Our SaaS offerings are also designed to natively incorporate advanced technologies such as AI, Internet-of-Things (IoT), machine learning, blockchain, digital assistants and advances in the “human interface” and how users interact with Oracle Cloud SaaS offerings within a business context or to augment human capabilities to enhance productivity.”
bundling: The AI is folded into the subscription price; no part of the price is allocated to it.
other: Usage of a free feature shows adoption, not willingness to pay.
other: Whether the new agentic health records system is priced separately from the system it replaces is not disclosed; if it is, part of this channel belongs under priced agentic capacity.
Quotes
“The introduction of AI is an accelerator, not a replacement, for packaged applications.”
“Rather than asking every employee to navigate and execute a process exactly as the system expects, AI agents can perform tasks using the organization's established workflows and business rules.”
“At AI World in October, we will unveil a new agentic AI accelerator poised to redefine how customers deploy Oracle applications faster, simpler, and at a dramatically lower cost. Working alongside Oracle and customer teams, AI agents will automate and orchestrate implementation at an unprecedented scale, compressing SaaS deployments from years to months, and months to weeks.”
“In total, our SaaS business grew 10%, with Fusion growing at 14%. Our Oracle Health business continued to accelerate, and although we don't specifically call it out, our industry applications grew at greater than 20% in Q1.”
“Customers used our embedded AI capabilities more than 150 million times during the quarter, with usage growing 42% sequentially. Our AI agents executed more than 3.5 million times in production during the quarter, nearly doubling quarter- over- quarter.”
“Customers have over 2,300 AI agents in production, and that is up 90% quarter- over- quarter. Overall, AI production usage across Fusion alone consumed 900 billion tokens during the quarter.”
“Additionally, the NetSuite AI Connector Service, which lets customers securely connect their NetSuite data to leading AI assistants of their choice, including ChatGPT and Claude, is already one of the fastest adopted capabilities in the whole entire history of NetSuite, with more than 10,000 customers already using it.”
“Number two, in some of these very complex industries, there are ramps associated with these go-lives, and it allows us to unlock the ramp and recognize revenue more quickly than we have in the manual implementation piece.”
“Oracle's AI Health Care Management and Electronic Health Records system is an all-new 100% Agentic system made up of a collection of AI agents for every medical specialty from General Medicine to Oncology to Radiology.”
“Our products and services include enterprise applications and infrastructure offerings that incorporate and are enhanced by artificial intelligence (AI) technologies, including embedded AI-driven automation and analytics and generative AI capabilities.”
Q1 2026direction only · described, no size · product-defensive
Q2 2026direction only · described, no size · product-defensive
Q3 2026direction only · described, no size · product-defensive
product revenue
Oracle AI Data Platform, agent tooling and access to third-party models
0.03% to 0.9% of the quarter’s revenue
Incremental total: counts in full.
our inferencedisclosure: described· motive: exploratory· before LLMs: expanded
The release and the call give the AI Data Platform more space: automated enterprise ontologies, hundreds of data sources, integration with third-party coding agents, and resale of OpenAI products through the marketplace (claims c66, c34, c48, c46). Asked how it earns revenue, management says it may be pure consumption or used with the applications or the infrastructure (claim c33). No customer count, price or revenue is given. The size, , is the ledger’s own, an outside view from the class of new AI platform products.
Evidence: 11 quotes, 2 confounds, 2 from before coverage
Revenue from the AI Data Platform, the AI Agent Studio and access to third-party models sold through the cloud, with which customers build their own agents on their private data. Management describes consumption, standalone and bundled business models and gives no customer count, price or revenue. The size is the platform’s own revenue, all of it taken as what AI added.
Why this motive
Carried from the prior quarters: asked for the business model, management answers that it may be any of several and gives no number (claim c33).
Before LLMs: expanded
The anchor already sold the same kind of work under other names: cloud AI offerings designed to be embedded into customer applications, generative AI among them, and generative AI services listed among the cloud infrastructure offerings, with no revenue given. The AI Data Platform, the agent studio and the resale of third-party models are not in the anchor; under the tie-break a separately priced product that continues work the anchor already shows is expanded, sized as an increment because the size is the new product’s revenue. The size is the change AI made, not the whole line.
“OCI AI offerings are designed to be embedded into customer applications for a variety of predictive use cases, including, among others, the servicing of machine parts that are at risk of failing, using generative AI for fault detection on an assembly line, the stocking of retailer store shelves, credit fraud detection and financial modeling to stay within a business’ forecasts.”
“Our OCI offerings also include cloud-based compute, storage and networking capabilities, application development and cloud native services, among others, and new and innovative services such as AI Infrastructure offerings and emerging technologies such as generative AI, IoT and blockchain.”
bundling: Platform consumption may be billed as ordinary cloud infrastructure usage, and the agent studio ships inside the applications.
other: Third-party models resold through the marketplace are revenue passed through to the model provider at an undisclosed margin.
Quotes
“The next layer is our Oracle Fusion Agentic Applications AI studio, which allows customers and/or partners to build their own AI agents right inside the same platform.”
“It is the same control plane and the same platform that our applications are running on, which means that Oracle AI Agent Studio, which allows customers to build their own agents or partners, gets all of the same quarterly updates, gets all of the same security patching, and is available as a complete service to our customers.”
“I think the other piece that is quite important is, as we mentioned in the press release, again, part of the same platform that we are running our applications on, the Oracle AI Data Platform, which automates the creation of enterprise ontology.”
“So, whether it is just pure consumption of AI data platform, whether it is used in concert with our applications, or whether it is used in concert with OCI, we are more focused on allowing the customer to make the best choice, or the partner to make the best choice that suits them.”
“As we continue to invest in making that available in every cloud, in every region, we think the AI Data Platform will help drive that growth as well.”
“We are already investing in deploying forward deployed engineers at our customers. That's true for the AI Data Platform. It's also true for our Fusion Agentic Studio, as we see them really as a combination platform running on a single control plane in OCI.”
“AI Data Platform is now integrated with Codex and Cloud Code, allowing developers to work with AI Data Platform data, knowledge, and capabilities from the coding environments they already prefer.”
“The new Oracle AI Data Platform fully automates the creation of Enterprise Ontologies—making it inexpensive, easy and fast and for any enterprise to use the most advanced AI models to reason on their private data and automate their business processes.”
Management names customers that added Fusion Agentic Applications (claims c19, c20) and gives no count, price or revenue; the former estimate multiplied the prior quarter's customer count, itself not a size, by an assumed multiple and purchase (the methodology rule for counts).
described, no sizedisclosure: described· motive: offensive· before LLMs: new
Management names several large customers that added Fusion Agentic Applications in the quarter (claims c19, c20). The customer count for token bundles given a quarter earlier is not repeated, and no price or revenue is given. The channel is unsized: the only measure behind the former estimate was the earlier customer count, , which sizes nothing. The former estimate, , is no longer the size.
Evidence: 3 quotes, 2 confounds
AI in the applications that carries its own price: bundles of tokens bought in advance for more advanced reasoning and models, agents priced on an outcome such as candidates screened, and the agentic applications customers add to a suite. Management reports customer counts and named customers, not revenue.
Why this motive
Carried from the prior quarter: a separately priced AI offering, now with named customers adding it (claim c19).
Before LLMs: new
The anchor has no priced unit of AI work in the applications: generative AI was a means of differentiating the suites, sold inside the subscription. Outcome-based pricing existed in the construction and hospitality products, without agents.
What else could explain it
other: The sources do not say whether Fusion Agentic Applications carry a separate charge in every case or how they relate to the token bundles.
bundling: Most AI in the applications remains free, so priced capacity is bought only beyond what the subscription includes.
Quotes
“Johnson Controls, the Saudi National Bank, GuideWell Mutual Holding Corporation, a health solutions company serving more than 45 million people, and PETRONAS, Malaysia's national energy company, each added Fusion Agentic Applications this quarter to drive better outcomes.”
“The next layer is our Oracle Fusion Agentic Applications AI studio, which allows customers and/or partners to build their own AI agents right inside the same platform.”
Q2 2026direction only · described, no size · offensive
Q3 2026described · described, no size · offensive
Cost imposed, or revenue lost, by others’ AI1 channel · 1 not sized
customer cohort
Application revenue lost or delayed because customers weigh AI-built alternatives
Not sized
Management denies the effect: AI accelerates packaged applications rather than replacing them (claim c14), and slower NetSuite growth is credited to slower decision cycles with no cause named (claim c18). A toll whose money management says has not started gets no ballpark (the methodology rule for money not started).
described, no sizedisclosure: described· motive: imposed· before LLMs: new
Management restates that AI accelerates packaged applications and does not replace them, and attributes slower NetSuite growth to slower decision cycles in the last fiscal year without naming a cause (claims c14, c18). Cloud applications revenue grew , with Fusion up (claim c17). No magnitude is given. The channel is unsized, since management denies the effect this quarter. The former estimate, , is no longer the size.
Evidence: 3 quotes, 3 figures, 1 confound
Cloud applications revenue the company would lose or see delayed because customers consider replacing packaged applications with software built quickly using AI, the thesis management calls the SaaS apocalypse. Management says its suites are not at risk and later acknowledges delayed decision cycles; cloud applications growth is the only reported line near it.
Why this motive
Carried from the prior quarters: a toll, not the company’s choice.
Before LLMs: new
The anchor does not describe customers replacing packaged applications with AI-built software. Cloud applications revenue was in Q1 of fiscal 2024, growing .
“In total, our SaaS business grew 10%, with Fusion growing at 14%. Our Oracle Health business continued to accelerate, and although we don't specifically call it out, our industry applications grew at greater than 20% in Q1.”
“As I mentioned last quarter, NetSuite saw some slower decision cycles last fiscal year, and therefore, the growth is a little lower than the rest, but we have an exciting new product generally available that I will speak about in just a bit.”
Q3 2026. Growing slower than revenue: total operating expenses (+18.5%), research and development (−3.6%), sales and marketing (−12.2%), general and administrative (0.0%), amortization of intangible assets (−51.9%), restructuring (−59.0%). A displaced cost shows up as a line that stays under the dashed revenue line. These are the audited lines, as first reported; nothing here is attributed to AI by the filing. Not drawn: restructuring, depreciation, which one-time items move by more than 60% in a quarter; the values are in the table below.
Total operating expensesResearch and developmentSales and marketingGeneral and administrativeAmortization of intangible assetsInterest expenseRevenue