Summary

Hold, $125 target: FY27 is secured, the FY28-FY30 path looks too high

InitiationBy Penmark Team
ORCLHold$124.99+8%
Page 1 of 26
Equity research | Initiation of coverage | United States
Software and AI infrastructure | Coverage: Initiated | Published 9 November 2026, 07:30 ET
Penmark Team | research@penmarkresearch.com
Oracle Corporation (NYSE: ORCL)

Hold, $125 target: FY27 is secured, the FY28-FY30 path looks too high

Rating
Initiation
Hold
12-month target
$124.99
Price, 6 Nov 2026
$116.10
Upside to target
+7.7%
Total return
+9.4%
Bull case
$229.80
Bear case
$26.50

Hold, $125 target: 7.7% upside to target and 9.4% total return. Oracle sells database software and now AI cloud capacity; the shares are down 52% in a year to 9.5x FY27E EBITDA. Our $114 weighted value rolls to the $125 target.

  • Backlog converts below plan: FY30E revenue $175bn vs the $225bn target. OpenAI concentration is worth −$32.6 a share (Chapter 10).
  • Leases hold infrastructure margin near 60%, not 66%. $288bn of leases start by FY29; one point is $4.04 a share (Chapter 11).
  • WACC decides the call: one point is $27.76 a share. Three of four alternative discount-rate tests move the rating; conviction is Low (Chapter 12).

What the market is missing. Our FY28E non-GAAP EPS of $8.29 is 25% below the Street’s $11.05, but the price already implies a 58.9% FY30E infrastructure margin against our 60.0%.

Catalysts. FQ2 FY27 results (Dec 2026); FQ3 RPO update (Mar 2027) decides +$24.3 or −$32.6; FY28 capex guide (Jun 2027).

Valuation and target. Ten-year DCF at 9.8% WACC: base $120.46; EV/EBITDA cross-check $152.52, which ignores $182bn of capex. Bear $26.50 (30%), base $120.46 (50%) and bull $229.80 (20%) weight to $114.14, which rolls to the $125 target.

Risks. OpenAI funding (−$32.6), lease costs (−$20.2), political backlash (−$16.9), funding costs (−$13.2).

Exhibit 1: Revenue grows 34% in FY27E, but free cash flow stays negative until FY29E
Year to 31 May, US$FY25AFY26AFY27EFY28EFY29E
Revenue ($bn)57.467.490.2119.3149.1
Growth y/y8.4%17.4%33.9%32.3%24.9%
EBITDA ($bn)23.929.949.864.882.4
Margin41.6%44.4%55.3%54.3%55.3%
Net income ($bn)12.417.120.520.728.1
EPS, GAAP ($)4.345.836.526.438.77
Growth y/y17.0%34.3%11.9%−1.4%36.4%
Free cash flow ($bn)(0.4)(23.7)(25.1)(24.3)6.7
At $116.10FY25AFY26AFY27EFY28EFY29E
P/E26.8x19.9x17.8x18.1x13.2x
EV/EBITDA19.8x15.8x9.5x7.3x5.7x
P/B16.3x9.0x4.1x3.4x2.6x
FCF yield−0.1%−6.4%−6.8%−6.6%1.8%
Dividend yield1.5%1.7%1.7%1.7%1.7%
ROE85%59%32%21%22%
Net debt / EBITDA3.4x3.3x2.0x2.0x1.6x
P/E at our $125 target28.8x21.4x19.2x19.4x14.2x

Source: Company filings; NYSE closes (Signals, Yahoo Finance; weekly, SPY as S&P 500); Penmark Research model. Shaded = estimates. Multiples at the 6 Nov 2026 close on GAAP EPS (non-GAAP FY27E P/E 14.3x at price, 15.4x at target). FCF includes customer prepayments; net debt excludes finance leases.

Base case is worth $120; the bear case is −77% and the bull case +98% (upside $113.7, downside $89.6)

Exhibit 3: Rolled 12 months, the weighted value reaches the $125 target; the bear case sits 75% below the price
$0$50$100$150$200$250$300Jan-26Jul-26Jan-27Jul-27ActualEstimatesPrice $116Bull $252 (+117%)Base $132 (+14%)Bear $29 (−75%)Target $125 (+8%)

Source: NYSE closes (Signals, Yahoo Finance); Penmark Research model. Scenario values at 6 Nov 2026 rolled at 9.5% a year (cost of equity less dividend yield).

Scenarios and price target basis
Bull$229.80+97.9%20% prob.

Management plan on time: FY30E revenue $226bn, infrastructure margin 66%, capex $100bn in FY28E.

Base$120.46+3.8%50% prob.

FY27E at guidance; FY30E revenue $175bn as OpenAI concentration slows growth; margin 60% as leases start.

Bear$26.50−77.2%30% prob.

OpenAI commitments renegotiated, older capacity churns; margin falls to 52%, FY30E revenue $118bn.

Probability-weighted value$114.14rolls to target
12-month target, upside$124.99+7.7%
Exhibit 4: Revenue rises 34% in FY27E and EBITDA 67%, but EBIT only 45% as D&A doubles
Income statement, $bnFY25AFY26AFY27EFY28EFY29E
Revenue57.467.490.2119.3149.1
EBITDA23.929.949.864.882.4
D&A(6.2)(9.3)(20.0)(33.7)(40.2)
EBIT17.720.629.831.142.2
Net interest(3.0)(3.8)(5.1)(5.8)(7.5)
Income taxes(1.7)(2.5)(4.0)(4.3)(6.2)
Net income, Oracle12.417.120.520.728.1
Net income, common12.417.020.220.427.9
Exhibit 5: FY27E free cash flow is −$47.6bn before $22.5bn of prepayments and −$25.1bn after
Cash flow, $bnFY25AFY26AFY27EFY28EFY29E
Operating cash flow20.832.067.465.771.7
Capex(21.2)(55.7)(92.5)(90.0)(65.0)
FCF, company basis(0.4)(23.7)(25.1)(24.3)6.7
Prepaid, received0.04.622.510.05.0
Prepaid, recognised0.00.00.0(4.1)(8.3)
FCF ex prepayments(0.4)(28.3)(47.6)(30.2)9.9
Dividends(4.7)(5.8)(6.1)(6.2)(6.2)
Cash, year end11.231.935.115.015.2
Exhibit 6: Net debt peaks at $130bn in FY28E; prepayments of $33bn fund part of the build
Balance sheet, $bnFY25AFY26AFY27EFY28EFY29E
Cash and securities11.231.935.115.015.2
Net PP&E43.5100.0173.5230.7256.2
Total assets168.4247.1322.8359.0384.0
Borrowings92.6129.5134.5145.2145.2
Customer prepayments0.04.627.133.029.8
Net debt ex leases81.497.699.4130.2130.1
Total equity21.043.192.4113.5143.2
Net debt / EBITDA3.4x3.3x2.0x2.0x1.6x
Exhibit 7: Infrastructure becomes 64% of revenue by FY29E; EPS growth waits for FY29E
Per share and KPIsFY25AFY26AFY27EFY28EFY29E
Diluted shares (m)2,8662,9143,0933,1753,175
EPS, GAAP ($)4.345.836.526.438.77
DPS ($)1.702.002.002.002.00
FCF per share ($)(0.14)(8.13)(8.13)(7.64)2.10
BVPS ($)7.112.928.033.844.6
Infra revenue ($bn)10.218.139.266.794.6
Infra % of revenue17.8%26.9%43.5%55.8%63.5%
Capex / revenue37%83%103%75%44%

Source: Company filings; Penmark Research model. Year to 31 May; shaded = estimates. GAAP; FCF = operating cash flow less capex; company-basis FCF includes net customer prepayments (received less recognised as revenue). Infra = cloud infrastructure.

Returns fall as capex is depreciated, and the one big gap to the Street is FY28E EPS: $8.29 against $11.05

Exhibit 8: EBITDA margin steps up 11pp to 55% in FY27E, then eases to 54% in FY28E as leases commence
Growth, marginsFY25AFY26AFY27EFY28EFY29E
Growth y/y
Sales8.4%17.4%33.9%32.3%24.9%
EBITDA11.0%25.4%66.7%30.0%27.2%
EBIT15.1%16.6%44.7%4.2%35.7%
EPS, GAAP17.0%34.3%11.9%−1.4%36.4%
DPS6.2%17.6%0.0%0.0%0.0%
Margins and cover
EBITDA margin41.6%44.4%55.3%54.3%55.3%
EBIT margin30.8%30.6%33.1%26.1%28.3%
Net margin21.7%25.2%22.4%17.1%18.7%
FCF margin−0.7%−35.2%−27.9%−20.3%4.5%
EBITDA / net interest8.0x7.8x9.8x11.2x11.0x
Exhibit 9: ROE falls from 59% to 21% by FY28E as equity is raised and capex is depreciated
Ratios and valuationFY25AFY26AFY27EFY28EFY29E
Returns and leverage
ROE85.4%58.6%32.5%21.0%22.4%
ROA8.0%8.2%7.2%6.1%7.6%
ROCE16.9%14.4%14.9%12.8%15.4%
Net debt / equity, ex leases3.9x2.3x1.1x1.1x0.9x
Net debt / EBITDA, ex leases3.4x3.3x2.0x2.0x1.6x
Valuation at $116.10
P/E, GAAP26.8x19.9x17.8x18.1x13.2x
EV/EBITDA19.8x15.8x9.5x7.3x5.7x
FCF yield−0.1%−6.4%−6.8%−6.6%1.8%
Dividend yield1.5%1.7%1.7%1.7%1.7%
P/E at $125 target28.8x21.4x19.2x19.4x14.2x

Source: Company filings; Penmark Research model. ROE on average common equity; ROA on average assets; ROCE = EBIT / average (equity + borrowings); net interest = interest expense less interest income. Multiples at the 6 Nov 2026 close; last row at our target. The model does not forecast gross profit, so there is no gross margin row. FY25A DPS from dividends paid. Net debt in these statement tables is borrowings less cash and excludes finance leases ($9.2bn at 31 Aug 2026).

Exhibit 10: We are 25% below consensus on FY28E EPS and in line on FY27E
Year to 31 MayFY27EFY28E
Revenue ($bn)
Penmark estimate90.2119.3
Consensus90.5131.8
Prior-year period67.490.2
Diff. vs consensus−0.3%−9.5%
Growth y/y (Penmark)+33.9%+32.3%
Non-GAAP EPS ($)
Penmark estimate8.118.29
Consensus8.1411.05
Prior-year period7.638.11
Diff. vs consensus−0.4%−25.0%
Growth y/y (Penmark)+6.3%+2.2%

Source: S&P Global via StockAnalysis (6 Nov 2026); company guidance (FY27 revenue at least $90.0bn, non-GAAP EPS $8.10); Penmark Research model.

Exhibit 11: Oracle trades at 13.9x trailing EV/EBITDA, 18.6% below the peer median despite a 52% margin
Latest four quartersORCLPeersDiff.
Valuation
EV/EBITDA, trailing 4Q13.9x17.1x−18.6%
Growth and margin
Revenue growth, latest Q29.6%19.7%+9.9pp
EBITDA margin, latest Q52.1%33.9%+18.2pp
Leverage and risk
Net debt / market value0.24x0.01x+0.22x
Beta (ORCL relevered)1.28x1.08x+0.20x

Source: Signals and Yahoo Finance prices (6 Nov 2026), company filings; Penmark Research model. Median of six peers (Microsoft, Amazon, Alphabet, Salesforce, CoreWeave, ServiceNow); difference in % for multiples, pp for ratios. Peer beta = median unlevered beta of the four peers with a published beta (Microsoft, Alphabet, Salesforce, CoreWeave).

The shares are cheap on trailing EBITDA and history, but the cheapness is the market pricing capex and funding risk

Exhibit 12: Four dated catalysts: the March RPO update alone can add $24.3 or remove $32.6 a share
  • Dec 2026FQ2 FY27 results: infrastructure +5pt vs base is worth +$4.0; Microsoft and Tencent churn is −$4.0. We sit on the guide (+30-34% revenue).
  • Jan-Feb 2027Hyperscaler CQ4 results: capex staying high keeps compute scarce (+$14.5); not in our base.
  • Mar 2027FQ3 FY27 RPO update: conversion of the $664bn backlog (+$24.3) against OpenAI funding (−$32.6).
  • Jun 2027FY28 capex guide: $90bn in our base; $15bn lower in FY29 is worth +$4.5.

Source: Company calendar; Penmark Research model catalyst re-runs.

Exhibit 13: Four downside risks cost $13-33 a share; one upside risk is worth $24
  • OpenAI concentration. Largest customer cannot fund its ramp. Likelihood 35%; −$32.6/share.
  • Leases ahead of revenue. $288bn of leases commence by FY29. Likelihood 40%; −$20.2/share.
  • Political backlash. Post-midterm limits slow data-center capacity. Likelihood 30%; −$16.9/share.
  • Funding cost. New notes at 7.00-8.65%; Baa2 outlook negative. Likelihood 30%; −$13.2/share.
  • RPO converts on schedule (against our Hold). New contracts on prepaid terms. Likelihood 35%; +$24.3/share.

Source: Penmark Research model; likelihoods are analyst assumptions.

Exhibit 14: Trailing P/E of 18.2x is 18 turns below its four-year average of 36.2x
0x20x40x60x80x2023202420252026Mean 36.2x; bands ±1σ, ±2σ18.2x

Source: Signals (GAAP diluted EPS), Yahoo Finance prices; Penmark Research. Weekly, Sep 2022 to Nov 2026; trailing four quarters, includes the FQ2 FY26 Ampere gain.

Exhibit 15: EV/EBITDA does not rise with margin (R² 0.27); Oracle sits at 13.9x with a 52% margin
10x20x30x40x10%20%30%40%50%60%70%EV/EBITDA, trailing 4QEBITDA margin, latest quarterORCLMSFTAMZNGOOGLCRMCRWVNOWR² 0.27, n=7

Source: Signals and Yahoo Finance (6 Nov 2026), company filings; Penmark Research. Peers include CoreWeave and ServiceNow, which we do not cover.

Exhibit 16: Six indicators confirm or break the Hold; the RPO update in March 2027 is the decisive one
IndicatorThesis confirmed ifThesis broken ifDate
Infrastructure revenue growthFQ2 near +116% y/y run-rate, margin about 60%Growth below 90% or margin below 55%Dec 2026
RPO and OpenAI shareRPO above $664bn with prepaid termsRPO falls or OpenAI commitments are cutMar 2027
Funding versus $20bn capPrepayments of $22.5bn in FY27E; no new equitySecond equity raise or Moody's to Baa3 or S&P to BB+1H CY27
Capex guideFY28 at or below $90bnAbove $100bn without new prepaid contractsJun 2027
Lease commencement costInfrastructure margin holds at 60%Margin at 52% or lowerFY27-FY28
Cost of capitalWACC near 9.8%Each +1pt takes value to $90.49Ongoing
What would change our view. A confirmed RPO conversion path (+$24.3 a share) takes base-case value to about $145; a cut to OpenAI commitments (−$32.6) takes it to about $88. The March update decides which path the weighted value follows. Source: Penmark Research model.
PART 1 · SUMMARY · 3 THE THESIS IN CHARTS

Six exhibits carry the thesis: the stock trades on backlog news, and we differ from the Street after FY27

Exhibit 17: Results-day moves track the change in RPO, not its level (correlation 0.75, six results)
−20%0%20%40%$0bn$100bn$200bn$300bn$400bnNext-day share moveChange in RPO in the quarterQ4 FY25Q1 FY26Q2 FY26Q3 FY26 Q4 FY26Q1 FY27R² 0.56, n=6

Source: Signals (RPO, prices), Yahoo Finance; Penmark Research. Jun 2025 to Sep 2026; one outlier (FQ1 FY26) drives the fit.

Exhibit 18: RPO is up 4.8x in six quarters; the share price is 10% lower than at FQ4 FY25
RPO, $bnShare price at quarter end, $ (right)$0bn$200bn$400bn$600bn$800bn$140$160$180$200$220$240FQ4FY25FQ1FY26FQ2FY26FQ3FY26FQ4FY26FQ1FY27

Source: Company filings; Signals; Yahoo Finance (closes at quarter end). RPO $bn: 138, 455, 523, 553, 638, 664.

Exhibit 19: We match FY27E guidance but sit 9% below the Street on FY28E revenue and 25% below on EPS
Our estimateConsensusFY27E revenue, $bnFY27E non-GAAP EPSFY28E revenue, $bnFY28E non-GAAP EPSFY30E revenue, $bn$90.2bn$8.11$119.3bn$8.29$175.3bn

Source: S&P Global via StockAnalysis (6 Nov 2026); LSEG (FY30); company guidance 10 Sep and 28 Oct 2026; Penmark Research model.

Exhibit 20: Across our DCF cases and Street targets the range is $27 to $400, with our target of $125 near the price
DCF, bear to bullWeighted value, WACC±1ptStreet targets, low to highEV/EBITDA cross-check$0$100$200$300$400$27$230$90$146$110$400$153Price $116Target $125

Source: Penmark Research model (6 Nov 2026 prices); S&P Global via StockAnalysis (42 analysts; marker = mean $235.77).

Exhibit 21: Earnings growth added $116 a share, but a lower P/E took away $244
$0$100$200$300$4006 Nov 2025EarningsgrowthP/E change6 Nov 2026$244+$116−$244$116

Source: Signals, Yahoo Finance; Penmark Research. Trailing GAAP EPS $4.32 to $6.38; P/E 56.4x to 18.2x; dividends excluded.

Exhibit 22: Infrastructure reaches $119bn by FY30E; FCF turns positive in FY29E
Infra revenue, $bnFY30E: $119FY26AFY30ECapex, $bnFY30E: $53FY26AFY30EFree cash flow, $bnFY30E: $29FY26AFY30EEBITDA marginFY30E: 56%FY26AFY30E

Source: Company filings; Penmark Research model. Year to 31 May; FCF includes customer prepayments.

PART 1 · SUMMARY · 4 KEY DEBATES

Three debates decide the stock: we are more cautious than the market on backlog and margin, slightly less on the cost of capital

Backlog conversion, infrastructure margin and the cost of capital drive our value for Oracle’s AI build. Backlog and margin resolve by March 2027; the cost of capital swings value by about $58 across a ±1pt range.

Exhibit 23: We differ from the market on three debates, each worth $13-33 a share if it breaks our way
DebateMarket viewPenmark viewEvidenceWhat would change our viewResolves
1. Backlog conversion$664bn RPO converts on the management path: FY30E revenue $225bn, FY28E $131.8bnOpenAI concentration slows it: FY28E infrastructure +70%, FY30E revenue $175bnOpenAI revenue about $20bn below estimates (8 Oct); Microsoft wind-down; Tencent lease cancelledRPO above $664bn on prepaid terms (+$24.3 a share)Mar 2027
2. Infrastructure margin66% FQ1 run-rate holds; group non-GAAP margin 40% by FY3060% as $288bn of leases commence ahead of revenueFQ1 EBITDA margin 52.1%; gross margin step-down guided; CoreWeave at about 52%Margin above 63% in two quarters; +$12.1 at +3ptDec 2026
3. Cost of capitalFunding is solved: $20bn cap, about half equity, net debt/EBITDA below 3.5xOur 9.82% WACC is below the 9.98% the price implies, so we are slightly less cautious; FY28E needs a $10.7bn credit-line draw after the capNotes priced 2 Nov at 7.00-8.65%; Moody's Baa2 negative; $29.8bn equity issuedPrepayments above $22.5bn in FY27E; no new equity1H 2027

Source: Company filings and calls (10 Sep, 28 Oct 2026); S&P Global via StockAnalysis; Penmark Research model.

Exhibit 24: We sit below the market on infrastructure revenue (16%) and margin (6pp), but our WACC is 0.16pp below the price-implied rate
PenmarkConsensusFY28E infra revenue, $bnFY28E infra margin, %WACC, % (price-implied)$66.7bn60.09.82

Source: Penmark Research model; Street-implied infrastructure revenue and the FQ1 FY27 margin run-rate; WACC the price implies (Sensitivities tab).

Exhibit 25: All three debates resolve within eight months, the largest in March 2027
Dec 2026FQ2 FY27resultsDebate 2Mar 2027FQ3 RPOupdateDebate 11H 2027Rating reviewsDebate 3Jun 2027FY28 capexguideDebates 1 and 3

Source: Company calendar; Penmark Research. Dates for results are estimates until confirmed.

Exhibit 26: Each debate maps to one thesis chapter and one value driver
DebateChapterStock driver (section 9)Value at stake, $/share
1. Backlog conversion10 Backlog converts below planRPO change on results day+24.3 / −32.6
2. Infrastructure margin11 Leases hold margin near 60%Lease cost and GPU pricing+12.1 / −20.2
3. Cost of capital12 The discount rate decidesRates and funding cost−13.2 per +0.5pt

Source: Penmark Research model catalyst re-runs, 6 Nov 2026.

Where we could be wrong. Backlog (about $57 between $144.80 and $87.85) and WACC (about $58 across ±1pt) move value most. Three of four alternative discount-rate tests move the rating, so conviction is Low.
PART 2 · THE BUSINESS · 5 HOW THE COMPANY MAKES MONEY

Cloud is now half of Oracle’s $67.4bn revenue, and 56 cents of each dollar goes to cash operating costs

Exhibit 27: Five numbers describe the business: $664bn of backlog, infrastructure up 121%, EBITDA margin 52%
$664bnRPO at 31 Aug 2026(+$26bn q/q)+121%FQ1 FY27 infrastructurerevenue growth52.1%FQ1 FY27 EBITDA margin$28.5bnFQ1 FY27 capex97.9%GPU utilisation (company,Sep 2026)

Source: Company filings and FQ1 FY27 call (10 Sep 2026); Signals; Penmark Research model.

Exhibit 28: Business model map: half of $67.4bn revenue is cloud, and $29.9bn (44%) is left as EBITDA
Cloud infrastructure $18.1bn$18.1bnCloud applications $15.9bn$15.9bnLicense support $19.8bn$19.8bnLicenses $4.7bn$4.7bnHardware $3.1bn$3.1bnServices $5.7bn$5.7bnRevenue $67.4bn$67.4bnEBITDA $29.9bn (44%)$29.9bnCash operating costs $37.5bn (56%)$37.5bnOfferings, FY26ATotal revenueWhat is left

Source: Company filings (10-K FY26); Penmark Research model. Cloud (infrastructure plus applications) is $34.0bn, 50% of revenue. Oracle does not disclose profit or capex by offering.

Exhibit 29: Infrastructure rises from 18% of revenue in FY25A to 68% by FY30E
Cloud infrastructureCloud applicationsSupport, license, hardware, services$0bn$50bn$100bn$150bn$200bnFY25AFY26AFY27EFY28EFY29EFY30EActualEstimates$57bn$67bn$90bn$119bn$149bn$175bn

Source: Company filings; Penmark Research model. Year to 31 May.

Exhibit 30: Capex has exceeded EBITDA since FY26A; the gap closes only in FY29E
EBITDA, $bnCapex, $bnFCF, $bn (right)$0bn$20bn$40bn$60bn$80bn$100bn−$40bn−$20bn$0bn$20bn$40bnFY25AFY26AFY27EFY28EFY29EFY30EActualEstimates

Source: Company filings; Penmark Research model. FCF includes customer prepayments.

Oracle sells the database and applications that run company records, and now rents AI computing capacity. Support and licenses ($24.5bn) are sticky; infrastructure customers buy scarce GPU capacity (utilisation 97.9%, renewals about 20% higher). New backlog is mostly on prepaid or customer-hardware terms.

EBITDA of $29.9bn became free cash flow of −$23.7bn in FY26A because capex was $55.7bn. Profit by offering is not disclosed, so we value the group on a DCF.

PART 2 · THE BUSINESS · 5 HOW THE COMPANY MAKES MONEY (CONTINUED)

Infrastructure supplies 92% of FY27E revenue growth, and its margin is the lever that matters most

Exhibit 31: Of each $1 of revenue, 31 cents became operating profit in FY26A
Cloud infrastructure$18.1bnCloud applications$15.9bnLicense support$19.8bnLicenses, hardware, services$13.6bnRevenue $67.4bn$67.4bnCost of revenue andother $26.5bn$26.5bnAfter cost of revenue$40.8bn$40.8bnResearch and development$10.3bnSales, general and administrative$9.9bnOperating profit $20.6bn (31%)$20.6bnRevenue, FY26ACostsOperating profit

Source: Company filings (10-K FY26); Penmark Research model. Cost of revenue and other is derived as revenue less R&D, SG&A and operating income; it includes depreciation of $9.3bn.

Operating leverage is strong in software but the cash cost of cloud is capex, not opex. R&D and SG&A together took 30 cents of each dollar. Depreciation rises from $9.3bn in FY26A to $20.0bn in FY27E and $33.7bn in FY28E in our model, so EBIT grows 4% in FY28E while EBITDA grows 30%.

Why it mattersDepreciation on $182bn of FY27-FY28 capex is why we do not value Oracle on P/E.
Exhibit 32: Revenue is infrastructure plus three slow-growing groups; infrastructure adds $21.1bn of the $22.8bn FY27E increase
Cloud infrastructure$39.2bn, +116.6%Cloud applications$17.5bn, +10.0%License support$19.8bn, 0.0%Licenses, hardware, services$13.7bn, +0.9%Revenue FY27E$90.2bn, +33.9%

Source: Penmark Research model; FY26A base. Infrastructure growth is the section 9 driver; its FY28E-FY31E path is +70%, +42%, +26%, +14%.

Exhibit 33: D&A grows from $9.3bn to $33.7bn by FY28E, so EBIT margin falls 7pp in FY28E even as EBITDA margin is 10pp above FY26A
D&A, $bnEBITDA margin, % (right)EBIT margin, % (right)$0bn$10bn$20bn$30bn$40bn$50bn20%30%40%50%60%FY26AFY27EFY28EFY29EFY30EActualEstimates

Source: Company filings; Penmark Research model. Year to 31 May.

Exhibit 34: Infrastructure growth slows from 117% in FY27E to 26% in FY30E as the base gets larger
0%25%50%75%100%125%FY26AFY27EFY28EFY29EFY30EActualEstimates77%117%70%42%26%

Source: Company filings; Penmark Research model. Year to 31 May.

PART 2 · THE BUSINESS · 6 THE MARKET: SUPPLY, DEMAND AND COST OF CAPACITY

Demand for AI capacity exceeds supply, but Oracle shares the market with peers earning 19-63% margins

Exhibit 35: Oracle books revenue in two of five links of the AI compute chain and leases the sites it needs
Chips and systemsNVIDIA, AMDOracle buys; capex$55.7bn FY26AData-center shells andpowerDevelopers, utilities$288bn of leases not yetcommencedCloud infrastructureOracle, Microsoft, Amazon,Alphabet, CoreWeaveOracle $18.1bn FY26A,+77%Platforms and databasesOracle, Microsoft,Salesforce, ServiceNowOracle applications$15.9bn; support $19.8bnAI model developersOpenAI and othersLargest customer;backlog $664bnOracle revenue booked: infrastructure, applications, support

Source: Company filings (10-K FY26, 10-Q Note 6); FQ1 FY27 call (10 Sep 2026); Penmark Research. Customer names from company statements and press.

Exhibit 36: Oracle grew 29.6% in the latest quarter, behind only CoreWeave among seven AI infrastructure peers
CoreWeaveORCLAlphabetServiceNowMicrosoftAmazonSalesforce112.3%29.6%24.6%20.5%18.8%12.1%10.8%

Source: Signals and company filings; latest reported quarter against the year-ago quarter (Peers tab, 6 Nov 2026).

Exhibit 37: Oracle's 52.1% EBITDA margin is third, behind Microsoft's 62.9% and CoreWeave's 52.5%
MicrosoftCoreWeaveORCLAlphabetSalesforceAmazonServiceNow62.9%52.5%52.1%38.7%29.1%24.3%18.9%

Source: Signals and company filings; latest-quarter EBITDA / revenue (Peers tab, 6 Nov 2026).

Demand drivers and cycle position

Demand exceeds supply today, and the signs are in hyperscaler spending. Microsoft guides FY27 capex of about $170bn and Google is leasing outside compute at $920m a month. Management reports 97.9% GPU utilisation.

Threats

The threats are capacity churn, funding and politics. Microsoft is winding down a 2023 rental and Tencent cancelled a $7bn lease. After the 3 Nov midterms, data-center permits are a new risk (Bear 5, −$16.9 a share).

Exhibit 38: Infrastructure revenue has tripled in seven quarters, from $2.4bn to $7.4bn
$0bn$2bn$4bn$6bn$8bnFQ2FY25FQ3FY25FQ4FY25FQ1FY26FQ2FY26FQ3FY26FQ4FY26FQ1FY27$2.4bn$2.7bn$3.0bn$3.3bn$4.1bn$4.9bn$5.8bn$7.4bn

Source: Company quarterly releases; Signals KPI (cloud infrastructure revenue).

Exhibit 39: Our revenue to Aug 2029 is $383bn, 1.15x the $332bn of backlog management expects to convert
$0bn$200bn$400bn$600bn$800bnRPO, Aug 2026Converts in 36monthsOur revenue, 36months$664bn$332bn$383bn

Source: FQ1 FY27 call (10 Sep 2026); Penmark Research model (Revenue build tab); our figure is total revenue, not only backlog.

PART 2 · THE BUSINESS · 7 COMPETITIVE POSITION

Oracle pairs 30% growth with a 52% margin, and trades on the lowest EV/EBITDA of its peers (13.9x)

Exhibit 40: Oracle's 30% growth is second only to CoreWeave's, and its 52% margin is third
0%20%40%60%80%5%10%20%50%100%200%EBITDA margin, latest quarterRevenue growth, latest quarter y/yORCLMSFTAMZNGOOGLCRMCRWVNOW

Source: Signals and company filings (Peers tab, 6 Nov 2026). Log scale on growth. Competitors we do not cover: CoreWeave, ServiceNow.

Exhibit 41: Advantage scorecard: switching costs score 4 of 4, while cost of funds and customer breadth score 2
Source of advantageRatingEvidence
Switching costs●●●●License support of $19.8bn is flat but recurring; databases are embedded in customer systems
Scale●●●○$664bn backlog; 850 MW delivered in FQ1 FY27
Cost position●●○○New debt at 7.00-8.65%; net debt/market value 0.24x vs 0.01x for peers
Customer breadth●●○○OpenAI: S&P estimates about half of the $638bn RPO (about $320bn); Oracle does not disclose it
Pricing power●●●○Renewals about +20% on older GPUs; utilisation 97.9%

Source: Company filings and FQ1 FY27 call; S&P Global Ratings (OpenAI share of RPO, an external estimate); Penmark Research ratings (analyst judgement, four dots = strongest).

Exhibit 42: Oracle trades at 13.9x trailing EV/EBITDA, below all six peers
ServiceNowMicrosoftAlphabetSalesforceAmazonCoreWeaveORCL38.7x18.9x18.2x16.1x15.8x14.7x13.9x

Source: Signals and Yahoo Finance prices (6 Nov 2026); company filings; Peers tab. Trailing four quarters.

Exhibit 43: Oracle carries net debt of 24% of market value; only CoreWeave (116% of equity) is more leveraged
CoreWeaveORCLSalesforceAmazonServiceNowAlphabetMicrosoft116%24%16%2%1%0%0%

Source: Signals and company filings; net debt / equity market value (Peers tab, 6 Nov 2026). Microsoft and Alphabet are net cash.

Exhibit 44: Peer data: Oracle is 18.6% below the median multiple with 9.9pp faster growth and an 18.2pp higher margin
CompanyEV, $bnEV/EBITDARevenue growthEBITDA marginNet debt / equity
Oracle473.413.9x29.6%52.1%0.24x
Microsoft3,884.118.9x18.8%62.9%0.00x
Amazon2,900.015.8x12.1%24.3%0.02x
Alphabet3,383.218.2x24.7%38.7%0.00x
Salesforce203.316.1x10.8%29.1%0.16x
CoreWeave55.314.7x112.3%52.5%1.16x
ServiceNow106.938.7x20.5%18.9%0.01x
Peer median17.1x19.7%33.9%0.02x

Source: Signals and Yahoo Finance prices (6 Nov 2026), company filings; Penmark Research. Trailing four quarters for EV/EBITDA; growth and margin latest quarter. Oracle EV $473.4bn: 31 Aug 2026, pro forma for the 4 Nov issue, incl. finance leases and prepayments.

What this shows. Oracle is the cheapest on EV/EBITDA and the second most leveraged: the market prices funding risk, not growth.
PART 2 · THE BUSINESS · 8 MANAGEMENT, GOVERNANCE AND CAPITAL ALLOCATION

Management has put $97bn into capex in five years, and the FY30 plan is 28% above our base case

Exhibit 45: The one open-market purchase was a $3.5m insider buy in September; disposals were $3.8m
ExecutiveRoleRecent Form 4 activity
Mike SiciliaCo-CEO since 22 Sep 2025Disposed 22,562 shares at $151.59 ($3.4m), 22 Sep 2026
Safra CatzExecutive vice chairTax settlement, 2,550 shares at $125.54, 23 Oct 2026
Larry EllisonChairman and CTOOwns 1,158m shares, 38.2% of the class (proxy, 21 Sep 2026); 413m shares are pledged as collateral, which is not a sale
Stephen RusckowskiInsiderBought 25,000 shares at $139.35 ($3.5m), 29 Sep 2026

Source: Oracle investor relations; SEC Form 4 filings via Signals (Sep-Oct 2026). Tax settlements are not discretionary sales.

Exhibit 46: Insider activity nets to $0.3m of sales: a $3.5m buy against $3.8m of disposals
Rusckowski purchaseSicilia disposalSmith disposal$3.5m−$3.4m−$0.4m

Source: SEC Form 4 filings via Signals, 22-29 Sep 2026; $m (shares x price). Smith: 2,631 shares at $151.70.

Exhibit 47: Over five years Oracle spent $166bn: $97bn on capex, $28bn on the Cerner deal, $19bn on buybacks and $22bn on dividends, with operating cash flow covering 59%
Operating cash flow $98.2bn$98.2bnDebt, cash and other $68.2bn$68.2bnCapex $97.0bn$97.0bnAcquisitions $27.9bn (Cerner $27.7bn, FY23)$27.9bnBuybacks $19.4bn (FY22 $16.2bn)$19.4bnDividends $22.0bn$22.0bnSourceUse

Source: Company filings (10-K FY22-FY26); Penmark Research. Sums of FY22A-FY26A; capex $4.5bn, $8.7bn, $6.9bn, $21.2bn, $55.7bn ($97.0bn). Cerner $27.7bn (FY23); buybacks $16.2bn in FY22. Debt, cash and other is the residual.

Exhibit 48: Non-GAAP EPS beat the Street by $0.18 in FQ1 FY27 ($1.92 against $1.74), and FY27 guidance was raised
Consensus$0.0$0.5$1.0$1.5$2.0$2.5FQ1 FY26FQ2 FY26FQ3 FY26FQ4 FY26FQ1 FY27$1.47$2.26$1.79$2.11$1.92

Source: Signals KPI (non-GAAP EPS); company releases; consensus from MarketBeat for FQ1 FY27. FQ2 FY26 includes the Ampere gain.

Exhibit 49: Management has set an FY30 target of $225bn revenue; our base case is $175bn, 22% below
$0bn$50bn$100bn$150bn$200bn$250bnFY26AFY30E PenmarkFY30 target$67bn$175bn$225bn

Source: Company filings; Investor Day 28 Oct 2026 (target at least $225bn); Penmark Research model.

PART 3 · THE THESIS · 9 WHAT MOVES THE STOCK

The stock trades on backlog, margin and funding news rather than on EPS, and the market has priced about 70% of the weighted backlog risk

The shares fell 52% in a year while earnings grew: the P/E fell from 56.4x to 18.2x. Three drivers each pass at least two of four tests. Against our probability-weighted values, the market has priced about 70% of the OpenAI backlog risk ($7.87 of $11.41), and has moved by more than our weighted values on margin and funding.

Exhibit 50: The stock trades on three things: backlog news, infrastructure margin and funding; each reaction is set against our probability-weighted model value
DriverWhat moves the stockTests confirmedMarket reaction vs our model valueDebateNext number to watch
Backlog and OpenAIChange in RPO; news on the largest customer●●●○
Event-day, call Q&A, decomposition
8 Oct OpenAI shortfall: −$7.87 (−5.5%). Model: Bear 1 −$32.61 × 35% = −$11.411. Backlog conversionRPO, Mar 2027: we expect conversion on the $332bn path; moves value +$24.3 or −$32.6
Infrastructure marginLease cost and GPU pricing against the 66% run-rate●●●○
Event-day, call Q&A, decomposition
Investor Day 28 Oct: +$14.01 (+10.7%). Model: Bull 3 +$12.11 × 35% = +$4.242. Infrastructure marginMargin, Dec 2026: we model 66% FY27E and 60% from FY28E
Funding and equityCapital raises, debt cost and the $20bn cap●●●○
Event-day, call Q&A, decomposition
2 Nov $20bn raise: −$11.45 (−8.8%). Model: Bear 4 −$13.24 × 30% = −$3.973. Cost of capitalCapex guide, Jun 2027: we model $90bn for FY28E

Source: Signals, Yahoo Finance, earnings-call transcripts, Penmark Research model. A driver is kept only if two of four tests confirm it (filled dots = tests passed).

Exhibit 51: Results-day moves ranged from +36% to −11%; the change in RPO explains 56% of the variation
−20%−10%0%10%20%30%40%FQ4 FY25FQ1 FY26FQ2 FY26FQ3 FY26FQ4 FY26FQ1 FY2713.3%36.0%−10.8%−2.4%−8.5%−1.7%

Source: Signals (prices, RPO), Yahoo Finance; Penmark Research. Last six results, Jun 2025 to Sep 2026.

Exhibit 52: Over three years earnings growth added $97 a share and the P/E took away $90, leaving the price up $7
Earnings growthP/E changePrice change$97−$90$7

Source: Signals (GAAP EPS, trailing four quarters), Yahoo Finance; Penmark Research. Three years to 6 Nov 2026; dividends excluded. Over one year earnings growth added $116 and the P/E took away $244 (Exhibit 21).

Exhibit 53: The market has priced $7.87 of the $11.41 probability-weighted backlog risk (69%); margin and funding moves were larger than our weighted values
Market reaction, $ a shareProbability-weighted model value, $ a shareBacklog: OpenAI shortfallMargin: Investor DayFunding: $20bn raise−$11.41−$7.87$4.24$14.01−$11.45−$3.97

Source: Yahoo Finance closes (7-8 Oct, 27-28 Oct, 30 Oct-2 Nov 2026); Penmark Research model catalyst re-runs (Bear 1 at 35%, Bull 3 at 35%, Bear 4 at 30%; value per share times probability). Events are broader in scope than the model re-runs.

PART 3 · THE THESIS · 9 WHAT MOVES THE STOCK (CONTINUED)

Rates and the EPS print do not move Oracle; the next numbers are RPO in March and infrastructure margin in December

Exhibit 54: Market and sector explain 51% of weekly moves; rates add 3pp and are not a driver
S&P 500 (SPY)Technology (XLK)Software (IGV)10-year yieldUnexplained25.0%11.8%14.6%3.2%45.4%

Source: Yahoo Finance weekly returns, Nov 2024 to Nov 2026 (n=106); Penmark Research. Incremental R² of nested OLS: total 54.6%. The 10-year yield alone explains 0.8%.

Exhibit 55: Applications drew most questions (26%), but backlog, margin and funding together drew 58%
Applications and databaseCapex and fundingMargin and cost to serveCapacity and GPU supplyRPO, OpenAI, customers26%21%21%16%16%

Source: Signals earnings-call transcripts, six calls from Mar 2025 to Sep 2026 (19 questions coded to a topic); Penmark Research coding.

1. Backlog news moves the stock more than the backlog itself.

What the market does: RPO change vs next-day move has a correlation of 0.75 over six results; the 8 Oct OpenAI shortfall cost $7.87 a share. Our model: Bear 1 is worth −$32.61 and Bull 2 +$24.34 a share; at 35% each, −$11.41 and +$8.52 weighted. Gap: the market has priced 69% of the weighted Bear 1 value ($7.87 of $11.41), a gap of $3.54. Debate 1. Next: RPO in March 2027; Bull 2 +$24.3, Bear 1 −$32.6.

2. Margin sets what the AI build earns.

What the market does: Investor Day targets (40% group non-GAAP margin) lifted the shares $14.01 (+10.7%) on 28 Oct; the gain has since gone. Our model: +3pt of infrastructure margin is worth $12.11; −5pt is −$20.19. Gap: market paid $9.77 more than the weighted +$4.24; we sit at 60%, not 66%. Debate 2. Next: FQ2 FY27 margin in December 2026; we model 66% in FY27E and 60% from FY28E.

3. Funding decides whether the build is self-financed.

What the market does: The $20bn raise priced on 2 Nov cost $11.45 (−8.8%) in one session. Our model: +0.5pt on WACC is −$13.24; our base adds a $10.7bn funding-line draw in FY28E after the $20bn cap is used. Gap: market paid $7.48 more than the weighted −$3.97 (Bear 4 at 30%). Debate 3. Next: FY28 capex guide in June 2027; we model $90bn.

Exhibit 56: Four candidates were tested and dropped: only one passed a single test
CandidateTest resultVerdict
10-year Treasury yieldWeekly changes explain 0.8% of returns; no event-day linkDropped
EPS surpriseFQ1 FY27 beat by $0.18 yet the stock fell 1.7% next dayDropped
Applications and databaseMost questions (5 of 19) but no price reaction on any eventSingle test, dropped
Capacity and GPU supply3 of 19 questions; no event-day linkSingle test, dropped

Source: Yahoo Finance, Signals, Penmark Research. Limits: six results and 106 weeks are small samples; correlations are descriptive, and the FQ1 FY26 move (+36%) drives the fit.

Exhibit 57: Backlog conversion moves base-case value most: $87.85 to $144.80 against $120.46
Funding: WACC +1pt / −1ptBacklog: Bear 1 / Bull 2Margin: −5pt / +3ptBase $120−$25+$33−$33+$24−$20+$12

Source: Penmark Research model re-runs (Sensitivities tab, 6 Nov 2026); base-case value per share at the price date.

PART 3 · THE THESIS · 10 BACKLOG CONVERTS BELOW THE MANAGEMENT PATH

Backlog converts below the management path: FY30E revenue $175bn against the $225bn target

Claim: the $664bn backlog converts, but too slowly and too concentrated to reach the $225bn FY30 revenue target: we model $175bn.

What the market believes

Street FY28E revenue is $131.8bn and FY30E $224.2bn (LSEG); the target is reaffirmed. At $116.10 the price implies a 9.98% WACC, so it already discounts some risk.

What it is worth

FY28E infrastructure revenue $66.7bn against about $79bn implied by the Street: FY28E non-GAAP EPS $8.29 against $11.05. Our 36-month revenue of $383bn is 1.15x the backlog expected to convert. Value: +$24.34 if it converts on time, −$32.61 if OpenAI cannot fund its ramp.

The evidence, from three sources

  1. 1. Reported data. RPO rose from $138bn to $664bn in six quarters; the quarterly increase has slowed from $317bn to $26bn. Management expects about half ($332bn) to convert in 36 months.
  2. 2. Call statements. Backlog, OpenAI and customers were 3 of 19 analyst questions; new RPO is mostly on prepaid or customer-hardware terms, with no capex or revenue effect before FY28 (call, 10 Sep 2026).
  3. 3. Industry and news. S&P Global Ratings estimates roughly half of the $638bn RPO reported at 31 May (about $320bn) is linked to OpenAI, an external estimate as Oracle does not disclose RPO by customer. OpenAI annualised revenue about $20bn below investor estimates (8 Oct); Microsoft is winding down a 2023 rental and Tencent cancelled a $7bn lease (5-6 Nov).
Exhibit 58: Infrastructure growth has accelerated for four quarters, to +121% in FQ1 FY27
0%25%50%75%100%125%FQ2 FY26FQ3 FY26FQ4 FY26FQ1 FY2768%84%93%121%

Source: Company quarterly releases; Signals KPI (cloud infrastructure revenue); growth against the year-ago quarter.

Exhibit 59: Backlog risks are skewed down: Bear 1 costs $32.61, the largest single item, against Bull 2 at +$24.34
Bull 2: RPO convertsBull 5: compute scarceBull 1: FQ2 accelerationBear 2: older capacitychurnsBear 5: political backlashBear 1: OpenAI shortfall$24.34$14.51$4.00−$4.00−$16.86−$32.61

Source: Penmark Research model catalyst re-runs (Sensitivities tab); $ per share against the $120.46 base.

Signposts (rests on driver: RPO change on results day)

DateEventConfirms / disproves
Dec 2026FQ2 FY27: infrastructure growthAbove +116% for FY27E confirms; below +100% breaks
Mar 2027FQ3 FY27 RPO updateAbove $664bn on prepaid terms confirms Bull 2; a cut breaks
Jun 2027FY28 capex guideAbove $100bn without new prepaid contracts disproves
PART 3 · THE THESIS · 11 LEASES HOLD INFRASTRUCTURE MARGIN NEAR 60%

Leases hold infrastructure margin near 60%, and each point is worth $4.04 a share

Claim: $288bn of leases arrive ahead of revenue, so infrastructure EBITDA margin holds near 60%, not the 66% run-rate.

What the market believes

The Street models about a 66% infrastructure margin (the FQ1 FY27 run-rate) and a 40% group non-GAAP margin by FY30. The price implies 58.9% FY30E, close to our 60.0%.

What it is worth

One point of infrastructure margin is worth $4.04 a share. At 60% we are 6pt below the run-rate; EBIT margin falls from 33.1% to 26.1% in FY28E as D&A reaches $33.7bn. Value: +$12.11 at +3pt, −$20.19 at −5pt.

The evidence, from three sources

  1. 1. Filings. $288bn of data-center leases commence between FQ2 FY27 and FY29 (10-Q Note 6); FQ1 FY27 capex was $28.5bn and D&A $3.4bn.
  2. 2. Call statements. Management guided a gross margin step-down in FY27, flattening over the next couple of years; margin and cost to serve was 4 of 19 questions.
  3. 3. Peers. CoreWeave, the closest pure play, earns a 52.5% EBITDA margin; Oracle's quarter is 52.1% and group margin of 55.3% in FY27E is already above FY26A's 44.4%.
Exhibit 60: EBITDA margin rose from 40.5% to 52.1% in six quarters; the next step down is guided
0%20%40%60%FQ1 FY26FQ2 FY26FQ3 FY26FQ4 FY26FQ1 FY2741%43%47%47%52%

Source: Company quarterly filings; Signals; Penmark Research (operating income plus D&A, over revenue).

Exhibit 61: Each 5pt of infrastructure margin is worth about $20 a share; each 10pt a year of growth about $50
$65$80$95$100$120$141$144$170$197Infrastructure margin FY28-36Margin −5ptBaseMargin +5ptInfrastructure growth FY28-31Growth −10ptGrowth baseGrowth +10pt

Source: Penmark Research model re-runs (Sensitivities tab); value per share at the price date.

Signposts (rests on driver: Lease cost and GPU pricing)

DateEventConfirms / disproves
Dec 2026FQ2 FY27 EBITDA marginAbove 55% with D&A rising confirms +3pt; a fall below 50% disproves
FY27-FY28Lease commencement and gross marginMargin at 55% equals Bear 3 (52% is the full bear case)
Jan-Feb 2027Hyperscaler resultsSustained capex keeps GPU pricing (renewals +20%)
PART 3 · THE THESIS · 12 THE DISCOUNT RATE DECIDES THE CALL

The discount rate decides the call: each point of WACC moves value $27.76 a share

Claim: the discount rate decides the call: each point of WACC moves value $27.76 a share, and three of four alternative discount-rate tests move the rating.

What the market believes

The Street is not debating funding: it assumes the $20bn capital cap (about half equity) is enough and net debt/EBITDA stays below 3.5x. The price implies a 9.98% WACC against our 9.82%.

What it is worth

WACC −1pt gives a $159.89 target (+37.7%, Buy); +1pt gives $99.09 (−14.7%, Sell), each rolled 12 months from the weighted value ($146.02 and $90.49). Net debt/EBITDA peaks at 3.3x in FY26A and falls to 1.6x by FY29E in our model. Value: −$13.24 for +0.5pt of WACC.

The evidence, from three sources

  1. 1. Filings and prospectus. Notes priced on 2 Nov at 7.00-8.65%, spreads +145-175bp wider than in February; $29.8bn of equity issued; net debt $97.6bn at May 2026.
  2. 2. Ratings and calls. S&P lowered Oracle to BBB- on 9 Jul; Moody's affirmed Baa2 with a negative outlook on the new notes; capex and funding was 4 of 19 analyst questions, the second most.
  3. 3. Our model. Cash held at $15bn needs a $10.7bn funding-line draw in FY28E; the $20bn capital-markets cap was used in full by the notes and shares priced on 2 Nov (closed 4 Nov), so we assume the draw comes from the credit line and commercial paper (about $19.7bn of headroom), not new issuance.
Exhibit 62: Each point of WACC moves base-case value by about $27.76: $153.86 at 8.82%, $120.46 at 9.82%, $95.69 at 10.82%
$0$50$100$150$200WACC 8.82%WACC 9.82%WACC 10.82%$154$120$96

Source: Penmark Research model (Scenarios tab); terminal growth 3.0%, value per share at the price date. The price implies a 9.98% WACC.

Exhibit 63: Cash after capex stays negative to FY28E; the model draws $10.7bn after management's $20bn cap is used
−$40bn−$20bn$0bn$20bn$40bnFY26AFY27EFY28EFY29EFY30EActualEstimates−$24bn−$25bn−$24bn$7bn$29bn

Source: Company filings; Penmark Research model. Includes customer prepayments of $22.5bn in FY27E, $10.0bn in FY28E and $5.0bn in FY29E.

Signposts (rests on driver: Rates and funding cost)

DateEventConfirms / disproves
1H 2027Moody's and S&P rating reviewsMoody's to Baa3 or S&P to BB+ disproves; a stable outlook confirms
Jun 2027FY28 capex guide and funding planCapex at $90bn with prepayments of $10bn confirms
Any timeSecond equity raiseAny raise beyond the 4 Nov issue breaks the base case
PART 4 · THE NUMBERS · 13 TRACK RECORD AND QUALITY OF REPORTED NUMBERS

Revenue and EPS have compounded for four years, but profit stopped converting to cash in FY25A

Exhibit 64: Revenue grew 12% a year over four years and EPS 25%, but free cash flow went from +$11.8bn to −$23.7bn in two years
Year to 31 MayFY22AFY23AFY24AFY25AFY26A
Revenue ($bn)42.450.053.057.467.4
Growth y/yn/a17.7%6.0%8.4%17.4%
EBITDA ($bn)14.019.221.523.929.9
EBITDA margin33.1%38.4%40.6%41.6%44.4%
Operating margin25.7%26.2%29.0%30.8%30.6%
Net income ($bn)6.78.510.512.417.1
EPS, GAAP ($)2.413.073.714.345.83
Operating cash flow ($bn)9.517.218.720.832.0
Capex ($bn)4.58.76.921.255.7
Free cash flow ($bn)5.08.511.8(0.4)(23.7)
Stock compensation ($bn)2.63.54.04.74.8
FCF / net income0.75x1.00x1.13xn/mn/m

Source: Company filings (10-K FY22-FY26); Penmark Research model. GAAP; FCF = operating cash flow less capex.

Exhibit 65: Non-GAAP EPS exceeds GAAP by 23-46% in four of five quarters; the FY26 gap is $1.80
GAAP EPSNon-GAAP EPSFQ1 FY26FQ2 FY26FQ3 FY26FQ4 FY26FQ1 FY27$1.01$1.47$2.10$2.26$1.27$1.79$1.45$2.11$1.56$1.92

Source: Company releases; Signals (diluted EPS and non-GAAP KPI). FQ2 FY26 GAAP EPS includes the Ampere gain.

Exhibit 66: Operating cash flow tripled to $32.0bn in FY26A, but capex of $55.7bn was 1.7x of it
Operating cash flow, $bnCapex, $bnFree cash flow, $bn (right)$0bn$20bn$40bn$60bn−$30bn−$20bn−$10bn$0bn$10bn$20bnFY22AFY23AFY24AFY25AFY26A

Source: Company filings (10-K FY22-FY26); Penmark Research.

Accounting choices that flatter results

Stock compensation and leases are the two to watch. Non-GAAP EPS excludes stock compensation ($4.8bn in FY26A, 7.1% of revenue) and amortization; we treat it as a cash cost in the DCF. Operating lease cost sits in EBITDA, but $288bn of leases have not yet commenced.

Profit-to-cash conversion

Reported profit no longer converts to cash. FCF was 113% of net income in FY24A and −139% in FY26A. Customer prepayments ($4.6bn in FY26A, $11.4bn in FQ1 FY27 alone) are financing, not operating cash, in our model.

Counter-evidence to our caution. Reported margins are rising (EBITDA margin 52.1% in FQ1 FY27 against 40.5% a year earlier) and FQ1 FY27 non-GAAP EPS beat the Street by $0.18. The weakness is cash, not earnings.
PART 4 · THE NUMBERS · 14 FORECASTS

Our forecasts match FY27 guidance, but FY30E revenue is 22% and EPS 32% below the Investor Day targets

Exhibit 67: Revenue is infrastructure plus three slow-growing groups; infrastructure adds $21.1bn of the $22.8bn FY27E increase
Cloud infrastructure$39.2bn, +116.6%Cloud applications$17.5bn, +10.0%License support$19.8bn, 0.0%Licenses, hardware, services$13.7bn, +0.9%Revenue FY27E$90.2bn, +33.9%

Source: Penmark Research model (driver tree as Exhibit 32). Infrastructure growth FY28E-FY31E: +70%, +42%, +26%, +14%.

Exhibit 68: We match FY27 guidance and sit 8-78% below the FY30 Investor Day targets on four lines
MetricGuidancePenmarkGapStatus
Revenue, FY27 ($bn)90.0 floor90.2+0.2%Inside range
Non-GAAP EPS, FY27 ($)8.108.11+0.1%Inside range
Capex, FY27 ($bn)90-9592.5midInside range
Revenue, FY30 ($bn)225+175.3−22.1%Outside
Non-GAAP op. margin, FY3040%+36.7%−3.3ppOutside
Non-GAAP EPS, FY30 ($)21.00+14.36−31.6%Outside
FCF, FY29 ($bn)30+6.7−77.8%Outside
Net debt / EBITDA, FY29<3.5x1.6xbelow capInside range
Funding-line draw FY28E ($bn)20 new-capital cap used10.7beyond capOutside

Source: Company FQ1 FY27 call (10 Sep 2026) and Investor Day (28 Oct 2026, 8-K Ex 99.1); Penmark Research model. Reasons: OpenAI concentration, lease costs ahead of revenue, D&A on $182bn of capex.

Exhibit 69: Against the Street we are in line in FY27E and 9% lower on FY28E revenue, 25% lower on EPS
ConsensusPenmarkFY27E revenue, $bnFY28E revenue, $bnFY27E EPS, $FY28E EPS, $90.290.5119.3131.88.1, no change8.311.1

Source: S&P Global via StockAnalysis (6 Nov 2026); Penmark Research model. EPS is non-GAAP.

Exhibit 70: FY30E value is most sensitive to infrastructure growth: ±10pt a year moves value by $50
Growth ±10pt a yearWACC +1pt / −1ptMargin −5pt / +5ptBase $120−$40+$50−$25+$33−$20+$20

Source: Penmark Research model re-runs (Sensitivities tab); base-case value per share.

Costs, cash and balance sheet

D&A, not opex, is the cost that scales. We hold the software margin at 47% and set infrastructure margin at 66% in FY27E and 60% after. Capex is $92.5bn, $90.0bn and $65.0bn in FY27E-FY29E, with prepayments of $22.5bn, $10.0bn and $5.0bn.

What the model needs

Cash is held at a $15bn minimum. That needs a $10.7bn draw in FY28E on the credit line and commercial paper, which is outside the $20bn new-capital cap; net debt peaks at $130.2bn and falls to 1.6x EBITDA by FY29E.

PART 4 · THE NUMBERS · 15 VALUATION

The target of $125 rests on a DCF worth $120.46, cross-checked by a peer multiple at $152.52

Method one: ten-year DCF

A DCF fits because FY27-FY28 capex of $182bn exceeds EBITDA. Free cash flow to the firm to FY36, mid-period, at a 9.82% WACC (risk-free 4.84%, equity premium 5.0%, beta 1.28, cost of debt 8.0%). Terminal growth is 3.0% on 12% return on new capital; terminal value is 68% of EV. Stock compensation is a cash cost; prepayments are financing.

Method two: peer EV/EBITDA

The cross-check gives $152.52, 26.6% above the DCF. Peer median 17.1x on Oracle's trailing EBITDA of $33.9bn. It sits higher because trailing EBITDA does not charge $182bn of FY27-FY28 capex or the funding it needs. We reject sum-of-the-parts (no segment capex disclosure) and forward P/E (D&A timing).

Exhibit 71: DCF enterprise value of $481.5bn becomes $120.46 a share after $114.6bn of claims
$0bn$100bn$200bn$300bn$400bn$500bnPVFY27-36PVterminalEVDebt,leasesCashOtherclaimsEquity+$154bn+$327bn$482bn−$135bn+$37bn−$17bn$367bn

Source: Penmark Research model (Valuation tab), 31 Aug 2026. Other claims: prepayments $16.0bn, minorities and preferred $1.2bn. Equity value $366.9bn over 3,089m shares is $118.78, rolled to $120.46 at 6 Nov 2026 on 3,175m shares after the 4 Nov issue.

Exhibit 72: Our two methods give $120.46 and $152.52, a 26.6% gap explained by $182bn of capex the multiple ignores
EV/EBITDA cross-checkTarget (rolled)DCF baseProbability-weighted$152.52$124.99$120.46$114.14

Source: Penmark Research model (Valuation tab); value per share at the 6 Nov 2026 price date, target rolled 12 months at 9.5%.

Exhibit 73: Value runs from $94 to $165 across WACC and terminal growth; the price sits at 9.98% WACC
$146$149$154$159$165$130$133$136$139$144$116$118$120$123$126$104$106$107$109$111$94$95$96$97$98Terminal growth2.0%2.5%3.0%3.5%4.0%WACC8.82%9.32%9.82%10.32%10.82%

Source: Penmark Research model (Scenarios tab); base case at the price date.

Exhibit 74: Reverse DCF: the price needs a 9.98% WACC, or a 58.9% FY30E infrastructure margin, or 1.97% terminal growth
WACC, % (model 9.82)Infra margin FY30E, %(model 60.0)Terminal growth, %(model 3.0)9.9858.921.97

Source: Penmark Research model (Sensitivities tab); each driver solved alone to $116.10, others at base.

Key assumptions and why. Infrastructure revenue grows 117%, 70%, 42%, 26% and 14% in FY27E-FY31E: FY27E sits at the $90.0bn guide floor, and FY28E onward is below consensus on OpenAI concentration. Infrastructure margin is 66% in FY27E and 60% after as leases commence. Capex is $92.5bn, $90.0bn and $65.0bn in FY27E-FY29E. We value FY27-FY36 explicitly and charge stock compensation as cash.
PART 4 · THE NUMBERS · 15 VALUATION (CONTINUED)

Bear, base and bull cases tie to the drivers, and the weighted value of $114 supports a Hold

Exhibit 75: Scenarios weight to $114.14: bear $26.50 (30%), base $120.46 (50%), bull $229.80 (20%)
$0$50$100$150$200$250$27$120$230Bear30%Base50%Bull20%Weighted $114Price $116

Source: Penmark Research model (Scenarios tab). Probabilities are analyst assumptions; bear: infrastructure margin 52%, FY30E revenue $118bn; bull: management plan, margin 66%, $226bn.

Exhibit 76: Each case ties to a driver: bear 1 and 3 cost $52.80 together; bull 2 and 3 add $36.46
Bull 2: RPO convertsBull 3: GPU pricing holdsBear 3: lease costsBear 1: OpenAI shortfall$24.34$12.11−$20.19−$32.61

Source: Penmark Research model catalyst re-runs (Sensitivities tab); $ per share against base $120.46.

Exhibit 77: The DCF bridge: $481.5bn of enterprise value less $114.6bn of claims leaves $366.9bn for 3,089m shares
US$bn at 31 Aug 2026Value
PV of FY27-FY36 cash flows154.1
PV of terminal value (68% of EV)327.4
Enterprise value481.5
Borrowings(125.3)
Finance leases(9.2)
Cash and marketable securities37.1
Customer prepayments received(16.0)
Minorities and preferred dividends(1.2)
Equity value366.9
Value per share at 31 Aug ($)118.78
Rolled to 6 Nov 2026, 3,175m shares ($)120.46

Source: Penmark Research model (Valuation tab).

Exhibit 78: Street targets average $235.77, 103% above the price; our $124.99 target is 47% below the mean
Street low $110Penmark target $125Street mean $236Price $116Street median $225Street high $400

Source: S&P Global via StockAnalysis (42 analysts, 6 Nov 2026); Penmark Research model.

Reconciliation to page 1. The rating basis is upside to the 12-month target, excluding dividends. The probability-weighted value of $114.14 rolled 12 months at 9.5% (cost of equity less the 1.7% dividend yield) is the $124.99 target: +7.7% on the $116.10 price (+9.4% with dividends), inside the Hold band of ±10%. Rolled the same way, the beta-high case ($96.60) is $105.78, −8.9%, still Hold; WACC ±1pt and beta-low move the rating, so three of four alternative tests flip it and conviction is Low.
PART 4 · THE NUMBERS · 16 CATALYSTS

Five dated events decide the rating, and two of them can add $24 or remove $33 a share

Exhibit 79: Five dated events in 13 months decide the rating; the March 2027 RPO update is the tie-break
Dec 2026FQ2 FY27 resultsInfra growth, marginJan-Feb 2027Hyperscaler resultsCapex, GPU scarcityMar 2027FQ3 FY27 RPO updateBacklog conversion1H 2027Rating reviewsMoody's, S&PJun 2027FY28 capex guideCapex, funding

Source: Company calendar; Penmark Research. Results dates are estimates until confirmed.

Exhibit 80: Penmark expects FY27 in line with guidance; the gaps to consensus open in FY28 and widen with each event
DateEventWhat decides itConsensusPenmark
Dec 2026FQ2 FY27 resultsInfra revenue +116% FY27E; revenue growth +30-34% guide; EPS $1.85-1.93 guideFY27E revenue $90.45bn, EPS $8.14$90.2bn, $8.11
Jan-Feb 2027Hyperscaler CQ4 resultsCapex stays high; compute scarce (Bull 5, +$14.5)Partly pricedNot in base
Mar 2027FQ3 FY27 RPO updateRPO above $664bn on prepaid terms (Bull 2, +$24.3) or OpenAI cut (Bear 1, −$32.6)FY28E revenue $131.8bn$119.3bn
1H 2027Moody's and S&P reviewsMoody's Baa2, negative outlook (affirmed on the new notes); S&P cut to BBB- on 9 Jul; a further cut adds funding cost (Bear 4, −$13.2 per +0.5pt)Priced in bond spreadsWACC 9.82%
Jun 2027FY28 capex guide$90bn FY28E; $15bn lower in FY29 is +$4.5 (Bull 4)FY27E FCF −$34.7bn (company basis)−$25.1bn

Source: Company guidance (10 Sep, 28 Oct 2026); S&P Global via StockAnalysis (6 Nov 2026); Penmark Research model catalyst re-runs. Events that could change the rating: the March RPO update and a rating cut.

Exhibit 81: The last eight sessions show the catalysts at work: +10.7% on Investor Day, then −20% in six sessions to $115.16
$110$120$130$140$150Nov-26$116

Source: Signals and Yahoo Finance closes, 28 Oct to 6 Nov 2026.

Next number to watch, by driver. Backlog: RPO in March 2027 (we model $332bn converting in 36 months). Margin: FQ2 FY27 infrastructure margin in December 2026 (we model 66%, then 60%). Funding: FY28 capex guide, June 2027 (we model $90bn).
PART 4 · THE NUMBERS · 17 RISKS

Four risks cost $13-33 a share, and the thesis ends if the March RPO update shows a cut or a conversion

Exhibit 82: Four downside risks cost $13-33 a share each, and the one upside risk to our Hold is worth $24
RiskWhat happensLikelihoodValue per shareSignpost
OpenAI concentrationLargest customer cannot fund its ramp; commitments renegotiated35%−$32.61RPO and OpenAI terms, Mar 2027
Leases ahead of revenue$288bn of leases commence by FY29; infrastructure margin 55% not 60%40%−$20.19FQ2 FY27 margin, Dec 2026
Political backlashPost-midterm limits slow new data-center capacity30%−$16.86State permits, from Jan 2027
Funding costNotes at 7.00-8.65%; Moody's to Baa3 or S&P to BB+; WACC +0.5pt30%−$13.24Rating reviews, 1H 2027
Upside risk to our HoldRPO converts on schedule on prepaid terms (Bull 2)35%+$24.34RPO, Mar 2027

Source: Penmark Research model catalyst re-runs (Sensitivities tab); likelihoods are analyst assumptions (Thesis tab). Value per share against the $120.46 base.

Exhibit 83: OpenAI is the only risk above $30 a share
$0$10$20$30$400%10%20%30%40%Value lost, $ a shareLikelihoodOpenAILeasesPoliticsFunding

Source: Penmark Research model (Thesis, Sensitivities tabs). Likelihoods are analyst assumptions.

Exhibit 84: Bear 1 and Bull 2 net to −$2.90 a share at 35% each, hence Hold
−$5.0−$2.5$0.0$2.5$5.0$7.5$10.0Bull 2 weightedBear 1 weightedNet+$8.5−$11.4−$2.9

Source: Penmark Research model (Thesis tab): $24.34 x 35% and −$32.61 x 35%. Weighted contributions are arithmetic on those inputs.

Exhibit 85: Three of four alternative tests change the rating; targets run from $99.09 to $159.89
WACC −1.0pt: +37.7% (Buy)Beta low: +28.1% (Buy)In use: +7.7% (Hold)Beta high: −8.9% (Hold)WACC +1.0pt: −14.7% (Sell)$159.89$148.74$124.99$105.78$99.09

Source: Penmark Research model (Sensitivities tab, 6 Nov 2026); 12-month target per share (probability-weighted value rolled at 9.5%). Upside to target: Buy above +10%, Sell below −10%, against the $116.10 price.

What would end the thesis

The March 2027 RPO update is the one number that matters. Base-case value is near $145 if commitments hold (Bull 2) and near $88 on a cut (Bear 1).

Regulatory, legal and key-person

Control and politics matter most. Larry Ellison (chairman, CTO) owns 38.2% of the shares, with 413m pledged as collateral (proxy, 21 Sep 2026). Data-center permits after the 3 Nov midterms are the main regulatory risk.

PART 5 · APPENDIX · 18 OWNERSHIP AND TRADING

Institutions hold at least 42% of Oracle, short interest is 1.6%, and the Street is 83% Buy

Exhibit 86: Ten institutions hold 618m shares, 20.5% of the 3,024m outstanding, led by Vanguard at 187m
VanguardBlackRockState StreetCapital ResearchGeodeMorgan StanleyFMRNorges BankBank of AmericaJPMorgan187m145m79m40m39m31m26m25m24m23m

Source: SEC 13F filings via Signals, 30 Jun 2026; shares in millions. 13F filers report 1,280m shares, 42.3% of shares outstanding.

Exhibit 87: Short interest is 47.5m shares, 1.6% of shares outstanding, and 1.66 days of volume
0m20m40m60m29 May30 Jun31 Jul31 Aug30 Sep15 Oct35m42m50m45m48m48m

Source: FINRA short interest via Signals, settlement dates 29 May to 15 Oct 2026. Days to cover: 1.88, 1.38, 1.40, 2.32, 1.47, 1.66.

Exhibit 88: Street ratings are 35 Buy, 6 Hold, 1 Sell, and our Hold is in the 14% minority
BuyHoldSell3561

Source: S&P Global via StockAnalysis, 42 analysts, 6 Nov 2026. Penmark Research rating is Hold.

Exhibit 89: Upside to target: Penmark +7.7% against +103% for the Street mean; even the lowest Street target is 5% below the price
Street highStreet meanStreet medianPenmarkStreet low244.5103.193.87.7−5.3

Source: S&P Global via StockAnalysis, 6 Nov 2026 (42 analysts); Penmark Research model. Percent change from the $116.10 price to each target.

Exhibit 90: Trading is liquid at $4.6bn a day, so a $100m position trades in under a day
Liquidity and positioningValue
Share price, 6 Nov 2026 ($)116.10
52-week range, close ($)114.99 - 248.15
Average daily value traded, 3 months ($bn)4.6
Shares outstanding, 30 Jun 2026 (m)3,024
Short interest, 15 Oct 2026 (m shares)47.5
Days to cover1.66
13F coverage of shares outstanding42.3%

Source: Yahoo Finance (price, volume); Signals (13F, short interest). Value traded is the three-month average of close x volume.

Insiders and control

Insider activity is mixed, not bearish. Stephen Rusckowski bought 25,000 shares ($3.5m) at $139.35 on 29 Sep 2026; Mike Sicilia sold 22,562 ($3.4m) on 22 Sep. Larry Ellison, chairman and CTO, owned 1,158m shares (38.2% of the class) at 21 Sep 2026 per the proxy; 413m of them are pledged as collateral, which is not a sale.

Overhangs

New supply is the overhang. Oracle sold $19.9bn of stock through its at-the-market programme in FQ1 FY27 and $9.9bn net (86.2m shares at about $116; priced 2 Nov, closed 4 Nov 2026). The mandatory convertible preferred adds up to 31.2m shares. Our base case assumes no further equity.

PART 5 · APPENDIX · 19 FINANCIAL MODEL

The model builds to $175bn of revenue and $29bn of free cash flow in FY30E, after two years of negative cash flow

Exhibit 91: Annual model: revenue reaches $175.3bn in FY30E and free cash flow turns positive in FY29E
US$bn, year to 31 MayFY25AFY26AFY27EFY28EFY29EFY30E
Income statement
Revenue57.467.490.2119.3149.1175.3
Growth y/y8.4%17.4%33.9%32.3%24.9%17.6%
EBITDA23.929.949.864.882.497.9
EBITDA margin41.6%44.4%55.3%54.3%55.3%55.8%
EBIT17.720.629.831.142.255.1
Net income12.417.120.520.728.138.4
EPS, GAAP ($)4.345.836.526.438.7712.08
EPS, non-GAAP ($)n/a7.638.118.2910.8814.36
Cash flow
Operating cash flow20.832.067.465.771.781.6
Capex21.255.792.590.065.052.6
Free cash flow, company basis(0.4)(23.7)(25.1)(24.3)6.729.0
Stock compensation4.74.85.26.67.88.8
Balance sheet
Net debt, ex finance leases81.497.699.4130.2130.1107.6
Net debt / EBITDA, ex leases3.4x3.3x2.0x2.0x1.6x1.1x
Total equity21.043.192.4113.5143.2184.2
Exhibit 92: Capex and funding: $92.5bn, $90.0bn and $65.0bn of capex drive a $10.7bn funding draw in FY28E
US$bn, year to 31 MayFY25AFY26AFY27EFY28EFY29EFY30E
Capital and funding
Capex21.255.792.590.065.052.6
Customer prepayments (model)n/a4.622.510.05.0n/a
Dividends paid4.75.86.16.26.26.6
Common equity issuedn/an/a29.80.00.00.0
Funding line drawnn/an/a0.010.70.00.0
Exhibit 93: Segment and KPI detail: infrastructure grows from $18.1bn to $119.3bn and reaches 68% of revenue by FY30E
US$bn, year to 31 MayFY25AFY26AFY27EFY28EFY29EFY30E
Cloud infrastructure
Infrastructure revenue10.218.139.266.794.6119.3
Growth y/yn/a76.9%116.6%70.0%42.0%26.0%
Share of revenue17.8%26.9%43.5%55.8%63.5%68.0%
Rest of Oracle
Other revenue47.249.351.052.754.456.1
Growth y/yn/a4.4%3.5%3.4%3.3%3.1%
Backlog at 31 Aug 2026
Remaining performance obligations ($bn)n/an/a664n/an/an/a
Converting within 36 months ($bn)n/an/a332n/an/an/a

Source: Company filings (10-K, 10-Q); Penmark Research model. Revenue from Sep 2026 to Aug 2029 in the model is $383.1bn, 1.15x the $332bn converting. Free cash flow is company basis, including customer prepayments.

PART 5 · APPENDIX · 20 REFERENCE

Reference: adjusted EPS runs $0.36-1.80 above GAAP, and every figure comes from one model dated 6 November 2026

Exhibit 94: Non-GAAP EPS is $1.80 above GAAP in FY26A and $0.36 above in FQ1 FY27
Per share ($)FY26AFQ1 FY27
GAAP diluted EPS5.831.56
Non-GAAP diluted EPS7.631.92
Difference1.800.36

Source: Company releases; Signals (GAAP diluted EPS, non-GAAP KPI). FQ2 FY26 GAAP EPS of $2.10 includes the Ampere gain.

Exhibit 95: Ratio definitions: returns use averages, valuation uses the $116.10 price
ROENet income to common over average common equity.
ROANet income over average total assets.
ROCEEBIT over average equity plus borrowings.
Net debtBorrowings less cash and marketable securities in the statement tables; the stock-data figure ($87.5bn at 31 Aug 2026, pro forma for the 4 Nov issue) also includes $9.2bn of finance leases.
EBITDAOperating income plus depreciation and amortization.
P/E at target$124.99 target over the fiscal-year EPS shown.
FCF yieldCompany-basis FCF over market value at $116.10.

Source: Penmark Research model (Methodology tab).

Exhibit 96: Six terms carry the thesis: backlog, prepayments, leases and cost of capital
RPORemaining performance obligations: contracted revenue not yet recognised ($664bn at 31 Aug 2026).
InfrastructureOracle Cloud Infrastructure: compute, storage and network sold for AI training and general workloads.
PrepaymentsCash customers pay ahead of delivery for capacity. We treat them as financing, not operating cash.
FCF, company basisOperating cash flow less capex, including customer prepayments. Ex-prepayments basis strips them out.
WACCWeighted average cost of capital; 9.82% in our model.
Lease commencementDate a leased data center starts generating lease cost; $288bn of leases have not yet commenced.

Source: Company filings; Penmark Research definitions.

Methodology note

Archetype: capacity and utilisation, with a projects and backlog element. The primary method is a ten-year FCFF DCF valued at 31 Aug 2026 and rolled to 6 Nov 2026; the cross-check is peer median trailing EV/EBITDA. We reject sum-of-the-parts (no segment capex disclosure) and forward P/E as primary. Rating: Hold on upside to the 12-month target of +7.7% (probability-weighted value of $114.14 rolled 12 months at 9.5%), excluding dividends.

Peer set and data sources

Six peers: Microsoft, Amazon, Alphabet, Salesforce, CoreWeave and ServiceNow, chosen for cloud infrastructure and enterprise software exposure; median trailing EV/EBITDA 17.1x. Sources, as of 6 Nov 2026: company filings and calls (through FQ1 FY27, 10 Sep 2026; Investor Day 28 Oct 2026); Signals (filings, 13F, Form 4, FINRA short interest, transcripts); Yahoo Finance (prices); S&P Global via StockAnalysis (consensus, Street targets); Penmark Research model.

PART 5 · APPENDIX · 21 DISCLOSURES

Analyst certification and important disclosures

Ratings

Ratings are based on the upside from the share price to our 12-month price target, excluding dividends. The price target is weighted across bull, base and bear scenarios.

BuyUpside to price target above +10%
HoldUpside to price target between −10% and +10%
SellUpside to price target below −10%

We review ratings at each set of results and whenever the facts change materially. A share price can move outside its band before the rating changes, and targets may not be reached.

Analyst certification

The Penmark Team certifies that (1) the views in this report accurately reflect its views about the subject company and its securities, and (2) no part of its compensation was, is or will be directly or indirectly related to the specific ratings, price targets or views in this report.

Contact

Penmark Research

Questions about our research:
research@penmarkresearch.com

All our research and coverage:
penmarkresearch.com

Basis of preparation

This report uses information publicly available up to its publication date. Share prices are closing prices on the trading day before the publication date, unless stated otherwise. Forecasts, valuations and price targets are Penmark Research’s own, and forecast periods are marked E. Sources are shown beneath each exhibit.

Valuation and price targets

Price targets are based on the valuation methods set out in each report and are weighted across bull, base and bear scenarios. The main risks to each rating and price target are described in the report. Penmark Research updates its coverage after each set of results but is not obliged to update a report at any other time.

Independence and conflicts of interest

Penmark Research is independent. It does not provide investment banking, brokerage, market-making or asset management services, does not trade securities, and does not accept payment from any company for research coverage. Companies do not see or review Penmark Research reports before publication.

No member of the Penmark Team holds a long or short position, directly or through derivatives, in the securities of any company Penmark Research covers.

Regulatory status

Penmark Research publishes general, impersonal investment research on a regular basis. It is not registered as an investment adviser with the US Securities and Exchange Commission or any state securities authority, is not a registered broker-dealer and is not a member of FINRA. Penmark Research does not give advice tailored to any person’s circumstances.

Not investment advice or an offer

This report is for information only. It is not personalised investment advice and is not an offer or solicitation to buy or sell any security or other financial instrument. It does not take into account any reader’s objectives, financial situation or needs. Readers should make their own investment decisions and seek independent professional advice where appropriate.

Forward-looking statements and risk

Forecasts, valuations and price targets are estimates based on assumptions that may prove wrong, and they may not be achieved. Opinions are current as of the publication date and may change without notice. Investing in equities involves risk, including the loss of capital. Share prices can be volatile, and past performance is not a reliable indicator of future results.

Data and sources

Company financials, operating data and filings come from company filings, earnings releases and earnings calls. Market data and any consensus estimates come from third-party sources named beneath the relevant exhibit; consensus figures from aggregators are unofficial. Penmark Research does not use other brokers’ research. Third-party data belong to their providers and may not be extracted from this report.

Accuracy and liability

Information is taken from sources believed to be reliable, but Penmark Research makes no representation or warranty as to its accuracy, completeness or timeliness. To the fullest extent permitted by law, Penmark Research accepts no liability for any loss arising from use of this report. Descriptions of litigation and regulatory matters reflect allegations and public records, not findings, unless stated. If a published report is corrected, the corrected version carries an “Updated” date at penmarkresearch.com.

Distribution and access

Penmark Research reports are available to registered users of penmarkresearch.com. Downloads and programmatic access are governed by the Penmark Research subscription terms. This report is not directed at, or intended for, any person in Singapore or India, and access from those countries is blocked. It is not intended for distribution to, or use by, any person in any jurisdiction where that would be contrary to law or regulation or would subject Penmark Research to registration or licensing requirements there.

Copyright and permitted use

© Penmark Research. All rights reserved. Penmark Research is a trade name of Brightriver Labs. Readers may share a link to this report but may not copy, reproduce, redistribute, sell or use its content, including for text and data mining, without written consent. Contact: research@penmarkresearch.com.

Keep reading

Sign in to Penmark