# Final S&P 500 AI P&L Evidence Report — 30 Days Ending August 17, 2026

## Status

This is the final auditable report for the fixed observation window. The
issuer universe is frozen at the August 17, 2026 close. The observation window
is July 19 through August 17, 2026, inclusive, in America/New_York.

- Universe: 500 issuers.
- Issuer reporting-window dispositions: 500/500.
- Minimum qualifying reporting cohort: 433 issuers.
- Qualifying periodic filings: 429 events from 428 issuers.
- Confirmed call-only reporters: 5.
- Resolved outside-window or no-call dispositions: 67.
- Unresolved call-only inventory: 0.
- Completed integrated issuer adjudications: 433.
- Qualifying reporters still awaiting structured adjudication: 0.
- Additional source packets triaged: 0.
- Companies with complete structured adjudications: 433.

The reporting-window disposition and structured economic adjudication of all
433 qualifying reporters are complete. An absent AI reference is not evidence
of zero company activity, spending, or benefit. The
[500-issuer disposition ledger](sp500-issuer-dispositions-2026-08-17.csv)
retains all reporting states and current structured-evidence coverage.

The remaining 67 issuers are not missing research: 65 reported outside the
fixed window and two had no earnings call found. They retain explicit
dispositions in the 500-row ledger but do not enter the 433-reporter evidence
cohort. The final reporter cohort separates 411 adopter-headline issuers from
22 AI providers. Five rows retain an `ACCESS_LIMITED` source flag because an
official replay or transcript was gated; each has a completed source-backed
adjudication using retained official corroboration and a complete alternative
transcript.

Final audit artifacts:

- [433-company evidence matrix](company-evidence-final.csv)
- [source and denominator coverage](source-coverage-final.csv)
- [statistics ledger](statistics-final.csv)
- [500-issuer disposition ledger](sp500-issuer-dispositions-2026-08-17.csv)
- [final audit manifest](final-audit-public.json)

## Population filing scan

All 429 qualifying periodic filings were acquired and scanned successfully.
Among the 428 unique filing reporters, 253 (59.1%) explicitly referenced AI or
machine learning and 175 (40.9%) did not. The all-index disclosure lower bound
is 253/500 (50.6%). Dexcom was added after human review found a machine-
learning sentence that the initial explicit-AI scan missed.

The automated scan produced discovery candidates, not final classifications:

| Discovery signal | n/N | Rate |
|---|---:|---:|
| Possible AI-spending reference | 77/428 | 18.0% |
| Possible AI-benefit reference | 120/428 | 28.0% |
| Possible realized operational/product result | 60/428 | 14.0% |
| Possible P&L/revenue-linked result | 8/428 | 1.9% |

Manual review is required because these candidates mix internal adoption,
provider supply, customer demand, risk disclosures, and broad transformation
programs containing AI.

## Structured adopter grade profile

The complete structured adopter cohort contains 411 companies; 22 additional
structured issuers are provider-side and reported separately. The counts below
are derived directly from the final D/S/B/G/J grades.

| Evidence gate | n/N | Rate |
|---|---:|---:|
| Substantive discussion (`D2`) | 47/411 | 11.4% |
| Explicit investment mechanism (`S2+`) | 18/411 | 4.4% |
| Quantified spending claim (`S3+`) | 3/411 | 0.7% |
| Total or ring-fenced AI spending in dollars | 1/411 | 0.2% |
| Quantified operational or financial result (`B2+`) | 13/411 | 3.2% |
| Current issuer direct-expense effect | 2/411 | 0.5% |
| Strict independently confirmed AI-only 2027 guidance | 1/411 | 0.2% |
| Complete spend-to-return reconciliation (`J1/J2`) | 0/411 | 0.0% |

## Strongest evidence

### Completed structured batches

Airbnb, Equifax, Sysco, Booking, CVS, Apple, FedEx, WTW, Axon, Tesla, Cigna,
Intel, Cognizant, ADP, ExxonMobil, Microsoft, Meta, IBM, News Corp, Honeywell,
Arista, AMD, GoDaddy, Block, Akamai, Seagate, Amazon, Western Digital, Fortinet,
Lumentum, Broadridge, Qualcomm, Datadog, KLA, Palantir, American Express,
CoStar, Expedia, Fiserv, McDonald's, Nasdaq, PayPal, Pfizer, Clorox, Coherent,
Fox, Lam Research, ResMed, Roper, Sandisk, S&P Global, Universal Health
Services, Equinix, Gen Digital, The Trade Desk, Jacobs Solutions, Goldman
Sachs, Omnicom, Generac, CDW, Alphabet, PNC, Iron Mountain, BNY Mellon,
Microchip, Tapestry, FIS, Align Technology, PTC, Vertiv, Amgen, Vistra,
Moderna, Citigroup, Cintas, AppLovin, Motorola Solutions, Procter & Gamble,
eBay, Robinhood Markets, DoorDash, Uber, and Teradyne now have structured
adjudications. FedEx Freight, Pinnacle West, Baxter, Cardinal Health, Johnson
Controls, Moody's, ServiceNow, and Blackstone are also complete. The first
fifteen are packet-backed; four hundred eighteen additional
issuers were completed through source-backed integrated transcript or filing
passes. Cadence, Incyte, Zebra, PPL, State Street, Verizon, T-Mobile, and
Entergy are also complete. American Electric Power, Corning, GE HealthCare,
F5, Ares, Keurig Dr Pepper, Marsh McLennan, and C.H. Robinson are also
complete. Amcor, Veralto, WEC Energy, Wells Fargo, Aptiv, Brown & Brown,
Domino's, and Globe Life are also complete. High-impact claims receive
selective skeptical review. CenterPoint, Danaher, PSEG, Bank of America,
Cincinnati Financial, Henry Schein, Colgate-Palmolive, and Philip Morris are
also complete. Amphenol, TE Connectivity, Qnity, Trimble, Biogen, Regeneron,
Gilead, and Disney are also complete. HCA, Morgan Stanley, Labcorp, Revvity,
Quest Diagnostics, Carvana, IFF, and Becton Dickinson are also complete.
Huntington, Hasbro, Truist, ON Semiconductor, Emerson, Gallagher, Expeditors,
and BXP are also complete. MSCI, CBRE, Principal, Eversource, AES, ADM, Avery
Dennison, and Apollo are also complete. AT&T, AbbVie, Chevron, Tyson Foods,
Mastercard, Caterpillar, NextEra Energy, and Intuitive Surgical are also
complete. Interactive Brokers, Darden, CMS Energy, IDEX, Tyler Technologies,
Intercontinental Exchange, Invitation Homes, and Ventas are also complete.
AIG, Paramount Skydance, Leidos, T. Rowe Price, Waters, UnitedHealth,
Ferguson, Charles Schwab, Ralph Lauren, Regions Financial, MetLife, Kenvue,
JPMorgan, Freeport-McMoRan, Evergy, and Consolidated Edison are also complete.
3M, PulteGroup, United Rentals, Hartford, Verisign, West Pharmaceutical,
J.B. Hunt, Teledyne, Baker Hughes, Capital One, Centene, Expand Energy,
FirstEnergy, IQVIA, NXP, UDR, Northern Trust, and Franklin Resources are also
complete.
Welltower, Xylem, Aon, Ford, Fortive, PACCAR, Starbucks, Waste Management,
Altria, Builders FirstSource, Coinbase, Deckers, First Solar, Huntington
Ingalls, Martin Marietta, Cboe, Eaton, Flex, Ingersoll Rand, Stryker, W. R.
Berkley, Alexandria Real Estate, Ameren, and Arch Capital are also complete.
Ball, Cummins, Duke Energy, DaVita, Edwards Lifesciences, Fifth Third,
W. W. Grainger, Gartner, Kimco, NRG, Rockwell Automation, Crown Castle,
Cencora, Healthpeak, Devon, Global Payments, Honeywell Aerospace, Eli Lilly,
Monolithic Power, Insulet, Solventum, Zimmer Biomet, Assurant, and BlackRock
are also complete. Ball's sole scan hit is the legal name Ayna.AI and is a
false positive; Cummins and Lilly reach only generic S1.
Ecolab, A. O. Smith, APA, Abbott, Aflac, Air Products, Albemarle, Allegion,
Alliant Energy, Allstate, American Tower, American Water Works, Ameriprise,
Ametek, Atmos Energy, AvalonBay, Berkshire Hathaway, Boeing, Boston
Scientific, Bristol Myers Squibb, Bunge Global, CF Industries, CME Group, and
CRH are also complete. Ecolab reaches filing-level D1 from forward-looking
reliance language; the other twenty-three filings contain no explicit AI
reference and therefore remain filing-level D0 rather than companywide zeroes.
CSX, Camden Property Trust, Carrier Global, Charles River Laboratories,
Charter Communications, Chipotle, Chubb, Church & Dwight, Citizens Financial,
Coca-Cola, Comcast, Comfort Systems USA, ConocoPhillips, Constellation Energy,
Corpay, Corteva, D. R. Horton, DTE Energy, Dexcom, Diamondback Energy, Digital
Realty, Dominion Energy, Dover, and Dow are also complete. Dexcom reaches
filing-level D1 from planned machine-learning product development; the other
twenty-three filings remain filing-level D0. Data-center load, facilities,
cooling, contracting, payments automation, and generic technology are not
internal AI economics.
DuPont, EOG Resources, EQT, EchoStar, Edison International, Emcor, Equity
Residential, Erie Indemnity, Essex Property Trust, Everest Group, Exelon,
Extra Space Storage, Fair Isaac, Federal Realty, GE Vernova, Garmin, General
Dynamics, General Motors, Genuine Parts, Halliburton, Hershey, Hilton, Host
Hotels, and Howmet Aerospace are also complete. All twenty-four fixed-window
filings contain no explicit AI or machine-learning reference and remain
filing-level D0. Analytics, scoring, automation, data-center construction,
grid software, autonomous driving, defense intelligence, and generic
technology are not promoted to AI evidence.
Hubbell, Humana, IDEXX, Illinois Tool Works, International Paper, Invesco,
Johnson & Johnson, KKR, KeyCorp, Kimberly-Clark, Kinder Morgan, Kraft Heinz,
L3Harris, Las Vegas Sands, Lennox, Linde, Live Nation, Lockheed Martin, Loews,
LyondellBasell, M&T Bank, MGM Resorts, Marathon Petroleum, and Marriott are
also complete. All twenty-four fixed-window filings contain no explicit AI or
machine-learning reference and remain filing-level D0. Analytics, automation,
grid equipment, diagnostics, portfolios, data centers, defense intelligence,
digital gaming, R&D, and generic technology are not promoted to AI evidence.
Applied Materials, Cisco, Masco, McKesson, Merck, Mettler Toledo, Mid-America
Apartment Communities, Molson Coors, Mondelez, Monster Beverage, Mosaic, NVR,
Newmont, NiSource, Norfolk Southern, Northrop Grumman, Norwegian Cruise Line,
Nucor, O'Reilly Automotive, Occidental, Old Dominion, ONEOK, Otis, and PG&E
are also complete. Applied Materials reaches internal-adoption
`D2/S2/B1/G1/J0`, but its AIx and semiconductor-equipment economics are
incidental provider supply and its internal claims remain unquantified. Cisco
is excluded as an AI technology vendor; its approximately 6% FY2026
AI-infrastructure revenue share and $7.5B FY2027 AI-infrastructure revenue
guidance are provider-side. Cisco's 145,000 AI-resolved support cases support
an internal operational B2 but not a financial B4. The other twenty-two
sources grade D0 after NVR's incomplete viewer capture was replaced with a
complete direct SEC filing snapshot.
PPG Industries, Packaging Corporation of America, Parker Hannifin, Pentair,
Phillips 66, Progressive, Prologis, Prudential Financial, Public Storage,
Quanta Services, RTX, Raymond James, Realty Income, Regency Centers, Republic
Services, Rollins, Royal Caribbean, SBA Communications, SLB, Sempra,
Sherwin-Williams, Simon Property Group, Skyworks Solutions, and Smurfit
Westrock are also complete. All twenty-four retained sources grade
`D0/S0/B0/G0/J0`. The twenty-three SEC filings contain no explicit AI or
machine-learning reference. Parker Hannifin's full call and official supporting
materials also contain no issuer AI disclosure; publisher AI controls and its
LLM disclaimer were rejected as boilerplate. Data centers, industrial
automation, semiconductor demand, defense systems, and generic technology were
not promoted to AI evidence.
Snap-on, Southern Company, Southwest Airlines, Stanley Black & Decker, Steel
Dynamics, Steris, Synchrony, TKO, Take-Two, Targa, Texas Instruments, Texas
Pacific Land, Textron, Thermo Fisher, Tractor Supply, Trane Technologies,
TransDigm, U.S. Bancorp, Union Pacific, UPS, Valero, Verisk, and Vertex are
also complete at source-level `D0/S0/B0/G0/J0`. Their twenty-three retained
SEC filings contain no explicit AI or machine-learning reference. Supermicro
is excluded as an AI technology vendor and grades provider-side
`D2/S0/B0/G1/J0`: its reported AI revenue mix is not multiplied into dollars,
its $65B-$72B FY2027 range is total-company revenue, and its margin, orders,
and CapEx remain blended. No adopter high-impact claim qualifies.
Viatris, Vici Properties, Visa, Vulcan Materials, Wabtec, Warner Bros.
Discovery, Weyerhaeuser, Williams, Wynn Resorts, Xcel Energy, Yum! Brands,
and Zoetis complete the qualifying reporter cohort. All twelve retained SEC
filings grade source-level `D0/S0/B0/G0/J0`. Generic intelligent tools,
agentic commerce, analytics, automation, franchise software, customer
data-center load, diagnostics, R&D, and technology totals were not promoted
to explicit AI evidence. No provider exclusion or high-impact claim qualifies
in the final batch.
Axon, Tesla, Intel, Cognizant,
Microsoft, Meta, IBM, Arista, AMD, Akamai, Amazon, Fortinet, Lumentum, Qualcomm,
Datadog, Palantir, Coherent, and Supermicro are provider-side and are excluded
from adopter headlines. Alphabet and ServiceNow are also core providers and
are excluded. Cadence and Cisco are also AI technology suppliers and are
excluded.
Leidos
is an incidental vendor but fails provider materiality and remains in adopter
headlines. IQVIA and NXP are also incidental vendors that fail provider
materiality and remain in adopter headlines.
Fortive, Huntington Ingalls, Ingersoll Rand, and Stryker are also incidental
vendors that fail provider materiality and remain in adopter headlines.
Applied Materials is also an incidental AI vendor that fails provider
materiality and remains in adopter headlines.

| Company | Spend | Current benefit | 2027 guidance | Strict result |
|---|---|---|---|---|
| Airbnb | Material 2026 increase, unquantified | $17M Q2 and $15M six-month vendor-cost reductions | None specific | Realized direct-expense effect; no net ROI |
| Equifax | Internal deployment; no dedicated spend disclosed | No isolated realized financial effect | $150M over 2026–28, not 2027-specific | Forward mixed target; not realized AI-only savings |
| Sysco | No dedicated AI spend disclosed | No isolated realized FY2026 result | ~$100M FY2027 bottom-line cost-out | Forward blended AI/prior-cost-out guidance |
| Booking | Low-single-digit share of technology spend | Double-digit service-cost-per-booking decline | None specific | Realized mixed-attribution unit-cost effect; no aggregate dollars |
| CVS | Unquantified AI investment within broad technology envelopes | Operational metrics; >$1B savings blends technology and AI | None AI-specific | B2 only; the AI share of savings is undisclosed |
| Apple | Rising AI-related R&D, OpEx, and COGS; no isolated amount | No internal-adopter financial result | None AI-specific | Spend direction only; provider-product evidence is excluded |
| FedEx | No dedicated amount | Qualitative lower-cost-to-serve mechanism | None AI-specific | B1 only; no measured result or isolated AI share |
| WTW | $625M cash plus $25M noncash Propel costs, blended | $400M gross/$350M net run-rate savings by end-2028, blended | Directional contribution in 2027; no 2027 amount | Forward B3/G3; no realized AI P&L or ROI |
| Axon | $4.4M SG&A and $5.1M quarterly/$6.5M six-month R&D expense increases | Nearly 700% AI-product revenue growth rate | Directional provider-product growth | Provider appendix only; no internal-adopter financial result |
| Tesla | $10.823B gross AI PP&E balance; $1.95B equity-funded AI-hardware acquisition | Higher FSD subscription revenue, unquantified | No quantified AI-specific 2027 amount | Provider appendix; balance is not a spend flow and total CapEx is blended |
| Cigna | Dedicated mechanism, no in-window AI amount | ~$2,000 lower annual customer medical cost; 42% fewer avoidable stays | $200M customer savings over three years, not allocable to 2027 | Customer-side mixed-attribution B4; not Cigna P&L |
| Intel | $637M Mentee Robotics acquisition | AI-driven businesses grew >70% and represented ~70% of revenue | Directionally higher 2027 total CapEx, unquantified | Provider appendix; Gaudi charge nonrecurrence is not an AI-created benefit |
| Cognizant | Explicit investment, unquantified; Leap totals are blended | Internal activity only; Leap savings are blended and forward | No AI-specific 2027 amount | Provider appendix; D2/S2/B1/G1/J0 |
| ADP | Significant AI-tool investment, unquantified | Zone access 10%→48%; contacts/client down 4% | Directional FY2027 efficiency only | Incidental vendor; internal D2/S2/B2/G2/J0 |
| ExxonMobil | Deployment only; no dedicated amount | ~90% validation and four unconfirmed Guyana prospects | Larger 2027 rollout, no financial mechanism | Non-provider D2/S1/B2/G1/J0 |
| Microsoft | AI infrastructure investment, no isolated amount | Foundry revenue more than doubled; GitHub Copilot revenue +60% QoQ | Companywide FY2027 CapEx growth only | Provider appendix; top-line B4 without bottom-line bridge |
| Meta | AI talent and token-cost mechanisms, no isolated amount | Advantage+ exceeded $75B annual revenue run-rate | 2027 CapEx amount explicitly withheld | Provider appendix; top-line B4 without bottom-line bridge |
| IBM | Hybrid-cloud-and-AI investment, no isolated amount | GenAI was ~50% of Consulting signings and >30% of backlog | No AI-specific 2027 amount | Provider appendix; signings/backlog are not revenue |
| News Corp | Internal investment unquantified | Qualitative productivity only | None AI-specific | Adopter; licensing and customer products excluded |
| Honeywell | Internal investment unquantified | No isolated internal financial result | 2027 Forge rubric is provider-side | Incidental vendor; Forge economics excluded |
| Arista | No dedicated AI spend | No recognized AI-fabrics revenue or margin bridge | FY2026 AI-fabrics goal ≥$3.5B; no 2027 amount | Provider appendix only |
| AMD | Total R&D/OpEx is blended; no dedicated AI amount | Instinct sales more than doubled, without profit bridge | 2027 Data Center revenue expected to more than double | Core-provider appendix only |
| GoDaddy | AI compute costs acknowledged, unquantified | Voicebot repeat-contact rate fell >16 percentage points | None AI-specific | Incidental vendor; internal B2 only |
| Block | AI infrastructure investment unquantified | Code changes per engineer reportedly rose 150% | None AI-specific | Incidental vendor; internal B2 only |
| Akamai | Up to $500M of provider GPU CapEx | Multi-year CIS commitments, not recognized revenue | Robotics contract ramps in 2027; company growth to low teens | AI-technology-provider appendix only |
| Seagate | Internal manufacturing AI investment unquantified | No measured internal financial result | Companywide FY2027 growth only | Adopter; AI-driven storage demand is supplier context |
| Amazon | $1B Forward Deployed Engineering; total CapEx blended | AI and chip run rates >$25B, not recognized revenue | Capacity direction only; no AI financial amount | Core-provider appendix only |
| Western Digital | Employee training only; no dedicated amount | No measured internal financial result | Companywide Q1 FY2027 guide only | Adopter; AI-driven storage demand is supplier context |
| Fortinet | Provider product and infrastructure investment unquantified | Internal support efficiency unquantified | No 2027 amount | AI-technology-provider appendix only |
| Lumentum | $167M Q4 CapEx for cloud-and-AI customer capacity | No internal-adopter financial result | Provider product timing into 2027 | AI-technology-provider appendix only |
| Broadridge | No dedicated AI spend | $25M expected FY2027 AI-driven productivity | $25M independently confirmed by CFO Ashima Ghei | Incidental vendor; forward B3/G3, not realized B4 or ROI |
| Qualcomm | Data-center and Modular investment, unquantified | No internal-adopter financial result | $5B fiscal-2027 Data Center revenue target | AI-technology-provider appendix only |
| Datadog | AI-product and infrastructure investment unquantified | No internal-adopter financial result | None AI-specific | AI-technology-provider appendix only |
| KLA | No dedicated amount | No measured internal financial result | Supplier growth only | Adopter; AI-infrastructure equipment demand is supplier context |
| Palantir | $741M Q2 adjusted expense is blended | Companywide provider margins and cash flow | FY2026 company guidance only | Core-provider appendix only |
| American Express | Named AI expense category; no amount | Intended operating efficiencies only | None AI-specific | Adopter D2/S2/B1/G1/J0 |
| CoStar | +$3M quarterly / +$5M six-month software-development expense, primarily AI software | Research efficiencies, unquantified | None AI-specific | Incidental vendor; quantified S3 expense increase, not total spend |
| Expedia | No dedicated amount | Intended conversion and operational efficiency | None AI-specific | Adopter D2/S1/B1/G1/J0 |
| Fiserv | $187M quarterly / $329M six-month One Fiserv costs are blended | Intended operating-model efficiency | None AI-specific | Adopter; client products separated |
| McDonald's | No dedicated amount | Restaurant > NEXT efficiency is intended, not measured | None AI-specific | Adopter D2/S1/B1/G1/J0 |
| Nasdaq | $139M Adenza costs are blended with location strategy | $160M+ net synergies are blended, not AI-attributed | Program closed; none for 2027 | Incidental vendor; Verafin and Dasseti excluded |
| PayPal | $120M–$140M expected charge is blended | $1.5B run-rate savings plan blends reorganization, AI, and automation | Two-to-three-year horizon, not AI-only | Adopter D2/S1/B1/G1/J0 |
| Pfizer | $7.3B program total and nested components are blended | Program savings combine digital enablement, automation, AI, and simplification | 2027 amounts are not AI-specific | Adopter D2/S1/B1/G1/J0 |
| Clorox | No dedicated amount | AI-supported R&D and marketing are qualitative | None AI-specific | Adopter D1/S1/B1/G1/J0 |
| Coherent | NVIDIA $2B equity financing; segment/R&D/CapEx totals not AI-only | AI-datacenter optics demand, no isolated bottom-line amount | Supply agreement through 2030; no 2027 AI amount | AI-technology-provider appendix under M2 |
| Fox | Potential AI compliance costs, unquantified | Internal tools and content generation, no measured result | None AI-specific | Adopter D2/S1/B1/G1/J0 |
| Lam Research | Internal AI tools unquantified | AI drives wafer-fab-equipment demand | None internal and AI-specific | Adopter; supplier demand treated like KLA |
| ResMed | Product R&D unquantified | Patient/product features and intended internal efficiency | None AI-specific | Adopter; ordinary-care AI features are customer-side |
| Roper | No dedicated amount | Embedded vertical-software and operating uses, unmeasured | None AI-specific | Incidental vendor; materiality fails |
| Sandisk | No internal AI spend disclosed | AI-driven NAND demand, not internal benefit | 2027 demand direction only, no amount | Adopter; storage demand treated like WD/Seagate |
| S&P Global | Immaterial Enertel acquisition; no internal amount | AI-related issuance is customer demand | None AI-specific | Incidental vendor; internal redesign unquantified |
| Universal Health Services | ~$115M genAI-company stake fair value; purchase cost undisclosed | No internal result | None AI-specific | External investment exposure; fair value is not AI spend |
| Equinix | Internal use unquantified; data-center totals excluded | No measured internal result | None AI-specific | AI-workload demand and xScale are supplier/customer context |
| Gen Digital | AI included in R&D; no allocation from $118M total | No measured internal result | None AI-specific | Embedded consumer-security features fail materiality |
| Trade Desk | +$2M quarterly / +$3M six-month mixed development-cost increments | No measured internal result | None AI-specific | Primarily AI tools, but not a dedicated AI budget |
| Jacobs Solutions | Continued AI investment, unquantified | Client efficiency only | None AI-specific | Client-facing services are not Jacobs P&L |
| Goldman Sachs | No internal AI amount | Governance and product integration only | None AI-specific | OneGS savings and technology expense are unallocated |
| Omnicom | New-technology investment unquantified | Qualitative employee productivity | None AI-specific | Omni is a client platform; restructuring totals excluded |
| Generac | No internal AI spend disclosed | AI-driven data-center demand | None AI-specific | Ordinary power-product sales are supplier-demand context |
| CDW | No internal AI spend disclosed | Customer AI-infrastructure demand | None AI-specific | Incidental reseller/integrator; materiality fails |
| Alphabet | Shared AI R&D and infrastructure, unallocated | TPU/Cloud/Search provider economics | 2027 TPU revenue timing, no amount | Core-provider appendix under M4 and C2/C3 |
| PNC | No internal AI spend disclosed | Macro AI-CapEx context only | None AI-specific | Economy-wide forecast language is not PNC economics |
| Iron Mountain | No concrete internal AI spend | No measured internal result | Data-center CapEx guide is unallocated | Data-center activity is not AI-only |
| BNY Mellon | No live deployment or amount | Threat and non-adoption risk only | None AI-specific | Filing hits do not establish BNY AI economics |
| Microchip | Employee AI-tool governance only | AI/ML end-market exposure | None AI-specific | Broad semiconductor demand is not internal benefit |
| Tapestry | Fiscal-2026 AI mechanism, unquantified | Qualitative engagement improvement | None AI-specific | D2/S2/B1; no financial bridge |
| FIS | Blended platform/AI investment, unquantified | No measured internal result | None AI-specific | Incidental vendor; transformation totals excluded |
| Align Technology | Product, service, and IT-system AI use, unallocated | No measured internal result | None AI-specific | Product and internal activity are not separated |
| PTC | Embedded software AI only | Customer transformation context | None AI-specific | Feature AI fails provider materiality; no internal economics |
| Vertiv | No internal AI spend disclosed | AI-driven data-center demand | None AI-specific | Power and thermal infrastructure are supplier context |
| Amgen | AI-enabled tools and adoption unquantified | No measured internal result | None AI-specific | R&D and operating totals remain unallocated |
| Vistra | Up to $1B external fund commitment | AI-enabling infrastructure investment | None AI-specific | Equity-method fund commitment is not internal AI spend |
| Moderna | Future AI cash needs blended with other programs | No measured internal result | None AI-specific | Operating cash use and R&D remain unallocated |
| Citigroup | Development and use acknowledged, unquantified | Risk and governance context only | None AI-specific | Transformation and productivity totals lack AI attribution |
| Cintas | Multi-function internal deployment, unquantified | Operational use without a measured KPI | None AI-specific | D2/S1/B1; no financial bridge |
| AppLovin | Axon AI data-center investment, unquantified | Advertising outcomes only | None AI-specific | Adtech fails provider materiality; cloud commitment unallocated |
| Motorola Solutions | Product acquisitions only | Customer public-safety workflows | None AI-specific | Incidental vendor; acquisition prices are not AI spend |
| Procter & Gamble | Operating-technology adoption, unquantified | General capability only | None AI-specific | Companywide productivity and expense totals unallocated |
| eBay | No dedicated amount | Qualitative marketplace efficiency | None AI-specific | D1/S0/B1; no measured bridge |
| Robinhood Markets | Product and process use, unquantified | Qualitative support/fraud/coding efficiency | None AI-specific | D1/S1/B1; no financial bridge |
| DoorDash | Broad platform costs, unallocated | Expected developer productivity | None AI-specific | D1/S1/B1/G1; result is forward and unmeasured |
| Uber | No internal AI amount | Aurora investment mark only | None AI-specific | External investment gain is not operating AI benefit |
| Teradyne | No internal AI spend disclosed | AI-driven test-equipment demand | Companywide 2027 investment direction | Supplier demand is not internal adoption |
| FedEx Freight | AI operations use, unquantified | Forecasting and risk-management activity | None AI-specific | Separate issuer; parent FedEx claims are not transferred |
| Pinnacle West | AI strategy and cameras, unquantified | Reliability and experience intentions | None AI-specific | D1/S1/B1; no measured bridge |
| Baxter | AI exploration, unquantified | Intended efficiency only | None AI-specific | Exploration is not deployment or realized benefit |
| Cardinal Health | AI/ML initiatives, unquantified | No measured internal result | None AI-specific | ERP and supply-chain program totals remain unallocated |
| Johnson Controls | Nantum AI acquisition unallocated | Customer building/data-center context | None AI-specific | Incidental vendor; $291M bundle is not AI spend |
| Moody's | CAPE product only | AI-financing issuance demand | None AI-specific | Incidental vendor; issuance revenue is blended |
| ServiceNow | Platform/R&D/cloud totals unallocated | Provider subscription economics | None AI-specific | Core-provider appendix under M2/M3 and C1 |
| Blackstone | No internal AI spend disclosed | Portfolio and ecosystem context | None AI-specific | Fund commitments and investment gains are not adoption |
| Cadence | Product/R&D/acquisition totals unallocated | AI-driven EDA product economics | None AI-specific | AI-technology-provider appendix under M3 |
| Incyte | $80.0M Genesis AI R&D prepayment; $2.7M Q2 expense | No measured AI benefit | Future use/cost unquantified | Adopter S3; equity and contingent milestones excluded |
| Zebra | No internal AI spend disclosed | Customer-facing AI agents only | None AI-specific | Incidental vendor; materiality fails |
| PPL | Internal grid use unquantified | Decision and asset-planning activity only | None AI-specific | D1/S1/B1; no measured bridge |
| State Street | AI integration unquantified | No measured internal result | None AI-specific | Financial-service features and technology totals unallocated |
| Verizon | Network AI use unquantified | No separable network result | None AI-specific | CapEx and broad efficiency totals unallocated |
| T-Mobile | No affirmative AI spend disclosed | Regulation and cybersecurity risk only | None AI-specific | Threat-actor AI is not issuer adoption |
| Entergy | No internal AI spend disclosed | AI/data-center load is customer demand | None AI-specific | Generation and transmission plans are supplier context |
| American Electric Power | GWAI software license unquantified | Equity investment and data-center load only | None AI-specific | $150M GWAI stake is not AI OpEx/CapEx |
| Corning | No internal AI spend disclosed | AI-infrastructure optical demand | None AI-specific | Optical products fail the purchased-layer test |
| GE HealthCare | Blended AI/product investment unquantified | Medical-product features only | None AI-specific | R&D and acquisitions remain unallocated |
| F5 | Internal use unquantified | AI-security product economics only | None AI-specific | Incidental vendor; materiality fails |
| Ares | No internal AI spend disclosed | Fund and portfolio context | None AI-specific | Carried-interest and data-center economics are not adoption |
| Keurig Dr Pepper | Multi-function use unquantified | No measured internal result | None AI-specific | D2/S1; technology bundle remains unallocated |
| Marsh McLennan | BCS investment unquantified | Thrive savings are blended | None AI-specific | $500M cost and $400M savings are not AI-only |
| C.H. Robinson | AI-enabled restructuring unquantified | Intended operational efficiency | None AI-specific | $50M-$75M program has no AI allocation |
| Amcor | Internal use unquantified | Intended efficiency only | None AI-specific | R&D and integration totals remain unallocated |
| Veralto | No internal AI spend disclosed | AI-augmented inspection product only | None AI-specific | Product and acquisition fail materiality |
| WEC Energy | Evaluation only | No measured internal result | None AI-specific | Utility load and capital are customer context |
| Wells Fargo | No affirmative AI spend disclosed | Cyber and technology risk only | None AI-specific | Technology expense remains unallocated |
| Aptiv | No internal AI spend disclosed | AI-enhanced vehicle products only | None AI-specific | Product and segment economics fail materiality |
| Brown & Brown | No affirmative AI spend disclosed | Technology-response risk only | None AI-specific | No deployment or financial bridge |
| Domino's | No affirmative AI spend disclosed | Content and consumer-behavior risk only | None AI-specific | Digital and franchise economics remain unallocated |
| Globe Life | No internal AI spend disclosed | Paid-search market-channel effect | None AI-specific | 4% sales decline has no isolated AI share |
| CenterPoint | Adoption/deployment unquantified | No measured internal result | None AI-specific | Data-center load and capital are customer context |
| Danaher | Business/product use unquantified | Risk language only | None AI-specific | R&D and Masimo acquisition remain unallocated |
| PSEG | Issuer/vendor use unquantified | No measured internal result | None AI-specific | Data-center legislation and grid capital are context |
| Bank of America | Emerging-technology use unquantified | Expected productivity/savings only | None AI-specific | Benefits are risk-qualified and unmeasured |
| Cincinnati Financial | Blended technology projects unquantified | Expected underwriting benefit only | None AI-specific | Reserve and ratio movements have no AI bridge |
| Henry Schein | No internal AI spend disclosed | Customer demand for AI solutions | None AI-specific | Product and segment economics fail materiality |
| Colgate-Palmolive | AI capability unquantified | No measured internal result | None AI-specific | Productivity program and R&D remain unallocated |
| Philip Morris | Internal AI use unquantified | Benefits explicitly uncertain | None AI-specific | Product, manufacturing, R&D, and CapEx remain unallocated |
| Amphenol | No internal AI spend disclosed | AI-driven customer demand only | None AI-specific | Connectors and sensors fail the purchased-layer test |
| TE Connectivity | No internal AI spend disclosed | AI-driven customer demand only | None AI-specific | Organic growth is supplier context, not adopter P&L |
| Qnity | No internal AI spend disclosed | AI-driven customer demand only | None AI-specific | Materials, packaging, PCB, and thermal products fail the purchased-layer test |
| Trimble | No internal AI spend disclosed | Customer-solution AI only | None AI-specific | Incidental vendor; $246.4M acquisition consideration is not internal AI spend |
| Biogen | Internal use unquantified | No measured internal result | Undated increased use only | R&D and operating totals remain unallocated |
| Regeneron | Internal use unquantified | No measured internal result | Undated further-use exploration | R&D and operating totals remain unallocated |
| Gilead | No affirmative deployment disclosed | Threat, misuse, and regulatory language only | None AI-specific | Same-filing Item 1A snapshot coverage is access-limited |
| Disney | No affirmative deployment disclosed | Infringement and legal risk only | None AI-specific | Content, technology, and CapEx totals remain unallocated |
| HCA | Use efforts unquantified | Forward-looking efficiency mechanism only | None AI-specific | No measured result or financial bridge |
| Morgan Stanley | No affirmative internal spend disclosed | AI-supported capital-market activity only | None AI-specific | Market condition, not issuer adoption |
| Labcorp | No affirmative deployment disclosed | Adverse-use and regulatory risk only | None AI-specific | No measured result or financial bridge |
| Revvity | Significant investment required, unquantified | Internal expansion is still initial | Undated expansion only | Product and internal mechanisms remain unallocated |
| Quest Diagnostics | Internal use unquantified | Intended productivity and quality only | None AI-specific | Invigorate savings and restructuring remain blended |
| Carvana | Internal use unquantified | Lower calls and higher conversion/NPS, unquantified | None AI-specific | Operational B1 without an expense or profit bridge |
| IFF | $9M Q2/$18M half-year blended strategic costs | Expanded GBS AI use, unmeasured | None AI-specific | Costs mix strategy, GBS, automation, consulting, salaries, and AI |
| Becton Dickinson | Development/deployment/use unquantified | No measured internal result | None AI-specific | Product and operating references fail provider materiality |
| Huntington | No affirmative internal spend disclosed | Macro, governance, and cyber context only | None AI-specific | AI Risk Committee is oversight, not deployment |
| Hasbro | Named AI studio, no dedicated amount | Studio launch and intended integration only | None AI-specific | Behavioral licensing fails provider materiality |
| Truist | No affirmative internal spend disclosed | Glossary and technology/cyber risk only | None AI-specific | No issuer operating mechanism |
| ON Semiconductor | No internal AI spend disclosed | AI data-center customer demand only | None AI-specific | Power semiconductors fail the purchased-layer test |
| Emerson | No internal AI spend disclosed | Sold industrial-AI products only | None AI-specific | Incidental vendor; segment economics remain unallocated |
| Gallagher | Internal use unquantified | Intended efficiencies only | None AI-specific | Third-party AI providers and regulation are risk context |
| Expeditors | No internal AI spend disclosed | Customer AI-infrastructure demand increased logistics volume | None AI-specific | No AI share of revenue or profit is disclosed |
| BXP | Internal use unquantified | Tenant and office-demand context only | None AI-specific | Leasing, redevelopment, and FFO remain unallocated |
| MSCI | No internal AI spend disclosed | Vantager is a sold AI-enabled platform | None AI-specific | Incidental vendor; the $95.5M aggregate acquisition price is not allocable to AI |
| CBRE | No internal AI spend disclosed | AI/data-center customer demand increased critical-infrastructure work | None AI-specific | Customer demand; no AI share of revenue or profit |
| Principal | No internal AI spend disclosed | Beam had ~$175M of 2025 premiums before the acquisition closed | None AI-specific | Target economics are not Principal AI revenue or benefit |
| Eversource | Strategic AI investment, unquantified | Intended outage, maintenance, and analytics value | None AI-specific | D2/S2/B1/G1; no amount or measured result |
| AES | No internal AI spend disclosed | AI Fund appears inside a mixed investment segment | None AI-specific | Segment losses, CapEx, and investment values remain unallocated |
| ADM | Internal mechanism, unquantified | Targets lower transaction costs and fewer manual touchpoints | None AI-specific | D2/S1/B1/G1; intended mechanism without a measured result |
| Avery Dennison | ~$75M external Wiliot minority investment | No internal result | None AI-specific | External investment, not internal AI OpEx/CapEx or benefit |
| Apollo | No internal AI spend disclosed | Macro and managed-fund compute-infrastructure exposure | None AI-specific | $35B facility and Athene guarantee are not Apollo internal AI economics |
| AT&T | Dedicated production mechanism, unquantified | Already “saving millions”; up to 90% scoped AI-cost reduction | None AI-specific | B3 only; vague amount, ceiling rate, and no statement-line bridge |
| AbbVie | AI-enabled campus, no AI allocation | No measured result | None AI-specific | Bundled manufacturing infrastructure; campus totals are not AI spend |
| Chevron | No dedicated amount | Directional shale and exploration use | None AI-specific | Data-center power and project returns are supplier context |
| Tyson Foods | Tested Agentic AI products, no amount | “Looking good,” unmeasured | Expansion planned, no dated financial amount | D1/S1/B1/G1; filing AI hits are avian influenza |
| Mastercard | No affirmative spend disclosed | Regulation and risk language only | None AI-specific | Filing-only evidence does not establish deployment or provider materiality |
| Caterpillar | No internal AI spend disclosed | AI-data-center customer demand only | None AI-specific | Engines and turbines fail the purchased-layer test |
| NextEra Energy | No amount disclosed | Planned productivity may not be realized | No dated financial amount | Risk-qualified intended benefit, not a current result |
| Intuitive Surgical | No internal AI spend disclosed | Ion AI navigation is a customer-facing device feature | None AI-specific | Ordinary-care feature fails provider materiality; no AI revenue allocation |
| Interactive Brokers | No internal AI spend disclosed | AI-company investor interest and IPO-market context | None AI-specific | SpaceX proceeds and market activity are not Interactive Brokers AI economics |
| Darden Restaurants | No internal AI spend disclosed | Cybersecurity and board-risk language only | None AI-specific | No restaurant deployment, labor saving, or financial result |
| CMS Energy | No internal AI spend disclosed | AI-sector and data-center risk context | None AI-specific | Implementation ability is not deployment; utility economics remain unallocated |
| IDEX | No internal AI spend disclosed | Mixed AI-driven customer demand contributed to HST volume | None AI-specific | AI is inseparable from space, defense, price, acquisitions, and rebates |
| Tyler Technologies | No dedicated AI spend disclosed | Product-timeline and market-opportunity risk only | None AI-specific | No purchased AI layer or materiality; ARR, R&D, and commitments remain unallocated |
| Intercontinental Exchange | No internal AI spend disclosed | AI feature and market-offering language | None AI-specific | No purchased AI layer or materiality; segment economics remain unallocated |
| Invitation Homes | No internal AI spend disclosed | Technology, cybersecurity, and compliance risk only | None AI-specific | No property-management deployment, staffing effect, or financial result |
| Ventas | No internal AI spend disclosed | Ventas/manager/tenant/borrower AI risk | None AI-specific | Tenant activity does not transfer to Ventas; no measured benefit |

Across the one hundred fifty-one broad-headline-eligible companies, 1/151
disclosed a total or ring-fenced AI dollar-spend level, 3/151 reached `S3+`,
2/151 disclosed a current direct-expense effect, 2/151 disclosed quantified
FY2027 bottom-line guidance that explicitly includes AI, and 1/151
independently confirmed an
AI-specific FY2027 amount. No company
has a comparable spend-to-return pair. CoStar's $3 million quarterly and
$5 million six-month figures are overlapping period-over-period
software-development expense increases, primarily related to AI software; they
are not total AI spend. Booking reports a low-single-digit share of technology
spend, not dollars. Incyte is the first adopter with a total ring-fenced dollar
amount: an $80.0 million prepayment for Genesis Molecular AI R&D services,
with $2.7 million recognized in Q2. Airbnb is the only case with a current
issuer-side dollar benefit amount; Booking supplies unit economics but no
aggregate dollars. Broadridge CFO Ashima Ghei independently confirmed
$25 million of FY2027 AI-driven productivity gains. The amount is forward
B3/G3 guidance, not realized B4 or ROI. Cigna separately supplies customer
medical-cost evidence, not issuer P&L. Provider-side issuers are reported
separately.

The second skeptical pass classified Booking `D2/S3/B4/G1/J0`, CVS
`D2/S2/B2/G1/J0`, and Apple `D2/S2/B1/G1/J0`. Booking's directional
positive-ROI statement does not qualify as `J1` because no quantified return
or payback was disclosed. CVS's savings are inseparable from broader
technology efficiencies. Apple's spending claims are provider-product or
mixed-unseparable and are excluded from adopter spending numerators.

The third skeptical pass classified FedEx `D2/S1/B1/G1/J0` and WTW
`D2/S2/B3/G3/J0`. Axon is `OUT_AI_TECH_VENDOR`; its provider-appendix grades
are `D2/S3/B4/G2/J0`, while its separable internal-use evidence reaches only
`D2/S2/B1/G0/J0`.

The fourth skeptical pass classified Cigna `D2/S2/B4/G3/J0`, with `B4`
limited to customer medical-cost outcomes and excluded from issuer-P&L
headlines. Tesla is `OUT_AI_TECH_VENDOR` with provider grades
`D2/S3/B4/G3/J0` and no separable internal-adoption result. Intel is
`OUT_CORE_PROVIDER` with provider grades `D2/S3/B4/G2/J0`; its separable
internal evidence reaches only `D1/S1/B0/G1/J0`.

The fifth integrated pass, reviewed read-only with independent review, classified
Cognizant `OUT_AI_TECH_VENDOR` at `D2/S2/B1/G1/J0`, ADP as an
`IN_ADOPTER` with an incidental AI product at `D2/S2/B2/G2/J0`, and
ExxonMobil `IN_ADOPTER` at `D2/S1/B2/G1/J0`. None produced an accepted B4,
J1, J2, quantified S3+, or quantified G3+ claim. An independent review was completed.

The sixth integrated pass, also reviewed read-only with independent review, classified
Microsoft and Meta `OUT_CORE_PROVIDER` at `D2/S2/B4/G2/J0`, and IBM
`OUT_CORE_PROVIDER` at `D2/S2/B2/G2/J0`. Microsoft's Foundry and GitHub
Copilot growth and Meta's greater-than-$75-billion Advantage+ run-rate are
realized provider top-line evidence only. IBM's GenAI share of Consulting
signings and backlog is demand evidence, not revenue. None disclosed dedicated
AI dollar spending, a strict realized bottom-line amount, AI-specific 2027
financial guidance, or a comparable spend-return pair.

The seventh integrated pass, retrieved through public source archive and reviewed
read-only with independent review, classified News Corp `IN_ADOPTER` at
`D2/S1/B1/G1/J0`, Honeywell as an incidental vendor that remains
`IN_ADOPTER` at `D2/S1/B1/G1/J0`, and Arista `OUT_AI_TECH_VENDOR` at
`D2/S2/B3/G3/J0`. News Corp licensing and customer-product evidence,
Honeywell Forge economics, and Arista AI-fabrics economics are not
internal-adopter benefits. None disclosed dedicated AI dollars, AI-only
OpEx/CapEx, a realized AI-attributed bottom-line benefit, AI-specific 2027
financial guidance, or a reproducible spend-return pair.

The eighth integrated pass used the same low-cost workflow for AMD, GoDaddy,
Block, Akamai, Seagate, and Amazon. AMD and Amazon are `OUT_CORE_PROVIDER`;
Akamai is `OUT_AI_TECH_VENDOR`. GoDaddy and Block remain incidental-vendor
adopters for their internal operational evidence, and Seagate remains an
adopter whose AI-driven storage demand is supplier context rather than an
internal benefit. Akamai disclosed up to $500 million of provider GPU CapEx,
and Amazon announced a $1 billion Forward Deployed Engineering investment, but
neither amount belongs in adopter spending. No company in the batch disclosed
a strict realized AI bottom-line result, AI-only 2027 financial guidance, or a
reproducible spend-return pair.

The ninth integrated pass used public source archive for source retrieval and one
read-only independent review for Western Digital, Fortinet, Lumentum, Broadridge,
Qualcomm, Datadog, KLA, and Palantir. Western Digital and KLA remain adopters,
but their AI-linked economics are supplier-demand context. Fortinet, Lumentum,
Qualcomm, and Datadog are `OUT_AI_TECH_VENDOR`; Palantir is
`OUT_CORE_PROVIDER`. Broadridge remains an incidental-vendor adopter. Its
complete secondary transcript reports $25 million of expected fiscal-2027
AI-driven productivity gains, and CFO Ashima Ghei independently confirmed the
same amount in a public management-authored post. The claim qualifies as
forward B3/G3 guidance but not realized B4, retained earnings, or ROI. No
company in the batch disclosed a current realized AI-attributed bottom-line
amount or a reproducible spend-return pair.

The tenth integrated pass used exact SEC 10-Q snapshots retrieved through
public source archive and one read-only independent review for American Express, CoStar,
Expedia, Fiserv, McDonald's, Nasdaq, PayPal, and Pfizer. All eight remain
adopter-headline eligible after customer-facing CoStar, Fiserv, Expedia, and
Nasdaq products are separated. CoStar disclosed the batch's only quantified
internal AI-spend evidence: software-development expense increased $3 million
for the quarter and $5 million for six months, primarily because of AI
software. The periods overlap, the amounts are expense increases rather than
total spend, and “primarily” is not an exclusive AI allocation. PayPal's
$1.5 billion run-rate-savings plan, Pfizer's $7.3 billion cost program,
Fiserv's One Fiserv costs, and Nasdaq's $139 million cost/$160 million-plus
synergy envelope all blend AI with other actions and are rejected as AI-only
economics. No company in the batch disclosed a realized AI-attributed
bottom-line amount, AI-specific 2027 guidance, or a reproducible
spend-to-return pair.

The eleventh integrated pass used exact SEC filing snapshots and one read-only
independent review for Clorox, Coherent, Fox, Lam Research, ResMed, Roper,
Sandisk, and S&P Global. Coherent is `OUT_AI_TECH_VENDOR` under `M2`: its
Datacenter & Communications segment represented approximately 74% of revenue,
sells specifically named AI-datacenter interconnect products, and has a
multi-year NVIDIA supply agreement. The segment's $5.275 billion of revenue,
NVIDIA's $2 billion common-equity investment, R&D, CapEx, PP&E, and purchase
commitments remain unallocated to AI. The other seven stay in the adopter
denominator. Lam, Sandisk, and S&P Global illustrate the recurring supplier
boundary: AI-driven wafer-fab demand, storage demand, and debt issuance are not
internal AI benefits. No company in the batch disclosed dedicated AI dollars,
a current AI-attributed bottom-line amount, AI-specific 2027 guidance, or a
spend-return pair.

The twelfth integrated pass used exact SEC 10-Q snapshots retrieved through
public source archive and one read-only independent review for Universal Health Services,
Equinix, Gen Digital, The Trade Desk, Jacobs Solutions, Goldman Sachs, Omnicom,
and Generac. All eight remain in the adopter denominator. UHS's approximately
$115 million is the fair value of a minority investment, not purchase cost or
internal AI spend. Trade Desk's $2 million quarterly and $3 million six-month
technology-and-development increases are overlapping mixed increments
primarily attributable to AI software tools, not a dedicated AI budget.
Equinix and Generac illustrate the supplier-demand boundary: customer
AI-workload and data-center investment is not internal adoption. No company in
the batch disclosed dedicated AI dollars, a realized AI-attributed bottom-line
benefit, AI-specific 2027 guidance, or a spend-return pair.

The thirteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for CDW, Alphabet, PNC, Iron
Mountain, BNY Mellon, Microchip, Tapestry, and FIS. Alphabet is a core provider
under `M4` and `C2/C3`; its Cloud, TPU, Gemini, shared R&D, infrastructure,
CapEx, revenue, and operating-profit totals remain unallocated to AI. The
other seven stay in the adopter denominator. Tapestry is the only new issuer
with a dedicated internal AI mechanism (`S2`) and a qualitative operating
result (`B1`), but no amount or P&L bridge. PNC's AI-related CapEx is
macroeconomic context, BNY's language is threat/non-adoption risk, and CDW,
Iron Mountain, Microchip, and FIS provide supplier, customer-demand, or
embedded-product signals rather than internal financial results. No issuer
disclosed dedicated AI dollars, a realized AI-attributed bottom-line benefit,
quantified AI-specific 2027 guidance, or a spend-return pair.

The fourteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for Align Technology, PTC, Vertiv,
Amgen, Vistra, Moderna, Citigroup, and Cintas. All eight remain in the adopter
denominator. Cintas supplies real multi-function deployment evidence
(`D2/S1/B1`) but no measured operating or financial bridge. Vertiv and Vistra
illustrate the supplier-and-investment boundary: AI-driven data-center demand
and a $1 billion external-fund commitment are not internal AI spending.
Moderna's future cash-needs language is blended, while Align, PTC, Amgen, and
Citigroup disclose activity, embedded features, or risk without a separable
amount. No issuer disclosed dedicated AI dollars, a realized AI-attributed
bottom-line benefit, quantified AI-specific 2027 guidance, or a spend-return
pair.

The fifteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for AppLovin, Motorola Solutions,
Procter & Gamble, eBay, Robinhood Markets, DoorDash, Uber, and Teradyne. All
eight remain in the adopter denominator; Motorola is an incidental AI vendor.
AppLovin's AI-powered advertising solution fails the frozen purchased-layer
and materiality/core tests, while Teradyne's AI-compute and memory exposure is
test-equipment customer demand rather than internal adoption. eBay, Robinhood,
and DoorDash provide qualitative efficiency or productivity claims without a
measured financial bridge. P&G supplies unquantified activity, Uber's $899
million Aurora gain is an external investment mark, and Motorola's AI-product
acquisition prices are not internal AI spend. No issuer disclosed dedicated AI
dollars, a realized AI-attributed bottom-line benefit, quantified AI-specific
2027 guidance, or a spend-return pair.

The sixteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for FedEx Freight, Pinnacle West,
Baxter, Cardinal Health, Johnson Controls, Moody's, ServiceNow, and Blackstone.
ServiceNow is provider-side under `M2/M3` and `C1`; its platform revenue, RPO,
margin, R&D, cloud, and acquisition totals remain unallocated to AI. The other
seven stay in the adopter denominator, with Johnson Controls and Moody's
treated as incidental vendors. FedEx Freight and Pinnacle West disclose actual
but unquantified operational use; Baxter is exploring AI; Cardinal Health
identifies initiatives without economics. Blackstone's AI references are
portfolio and market context. No issuer disclosed dedicated AI dollars, a
realized AI-attributed bottom-line benefit, quantified AI-specific 2027
guidance, or a spend-return pair.

The seventeenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for Cadence, Incyte, Zebra, PPL,
State Street, Verizon, T-Mobile, and Entergy. Cadence is provider-side under
`M3`; Zebra remains an incidental vendor. Incyte supplies the batch's only
high-impact claim: an $80.0 million prepayment for Genesis Molecular AI R&D
services and $2.7 million of Q2 expense, both accepted as `S3`. The separate
$40.0 million equity investment and all contingent drug-development milestones
are excluded. PPL and Verizon disclose unquantified operational use, while
State Street discloses unquantified integration. T-Mobile and Entergy provide
risk or customer-demand context. No issuer disclosed a realized
AI-attributed bottom-line benefit, quantified AI-specific 2027 guidance, or a
spend-return pair.

The eighteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for American Electric Power,
Corning, GE HealthCare, F5, Ares, Keurig Dr Pepper, Marsh McLennan, and C.H.
Robinson. All eight remain in the adopter denominator; F5 is an incidental
vendor because its sold AI-security layer fails materiality. AEP's $150
million Gigawatt AI stake is an equity-method investment. Corning's optical
economics and GE HealthCare's medical-product economics are supplier-side.
Ares's amounts are fund or portfolio economics. Marsh's approximately $500
million Thrive cost and approximately $400 million savings target and C.H.
Robinson's $50 million–$75 million restructuring program are blended and have
no AI allocation. No issuer disclosed dedicated AI dollars, a realized
AI-attributed bottom-line benefit, quantified AI-specific 2027 guidance, or a
spend-return pair.

The nineteenth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for Amcor, Veralto, WEC Energy,
Wells Fargo, Aptiv, Brown & Brown, Domino's, and Globe Life. All eight remain
in the adopter denominator and none passes provider materiality. Amcor
supplies only qualitative internal use. Veralto and Aptiv supply product
references that fail the purchased-layer and materiality tests. WEC is
evaluating tools; Wells Fargo, Brown & Brown, and Domino's provide risk
language. Globe Life says AI-driven consumer search reduced paid-search
volume, but its 4% DTC sales decline to $54 million has no isolated AI share
and is a market-channel effect rather than internal adoption. No issuer
disclosed dedicated AI dollars, a realized AI-attributed bottom-line benefit,
quantified AI-specific 2027 guidance, or a spend-return pair.

The twentieth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for CenterPoint, Danaher, PSEG, Bank
of America, Cincinnati Financial, Henry Schein, Colgate-Palmolive, and Philip
Morris. All eight remain in the adopter denominator and none passes provider
materiality. Utility-load references, ordinary diagnostic and dental product
features, and competitor or customer demand are not internal adoption.
CenterPoint, PSEG, Bank of America, Cincinnati Financial, Colgate, and Philip
Morris provide qualitative activity or intended-benefit language without an
amount. No issuer disclosed dedicated AI dollars, a realized AI-attributed
bottom-line benefit, quantified AI-specific 2027 guidance, or a spend-return
pair.

The twenty-first through twenty-third integrated passes added Amphenol, TE
Connectivity, Qnity, Trimble, Biogen, Regeneron, Gilead, Disney, HCA, Morgan
Stanley, Labcorp, Revvity, Quest Diagnostics, Carvana, IFF, Becton Dickinson,
Huntington, Hasbro, Truist, ON Semiconductor, Emerson, Gallagher, Expeditors,
and BXP. All remain adopter-headline eligible; Trimble and Emerson are
incidental vendors. IFF's $9 million quarterly and $18 million half-year
strategic-initiative costs are blended rather than AI-only. ON, Expeditors,
Amphenol, TE, and Qnity provide customer or supplier-demand signals, while the
remaining issuers provide unquantified activity, intended benefits, or risk
language. None adds a qualifying `S3+`, `B2+`, strict 2027 amount, or
spend-return pair.

The twenty-fourth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for MSCI, CBRE, Principal,
Eversource, AES, ADM, Avery Dennison, and Apollo. MSCI remains an incidental
vendor because Vantager passes the external-product test but fails every
provider materiality safe harbor. Eversource adds an unquantified strategic
investment mechanism (`D2/S2/B1/G1`), and ADM adds an intended transaction-cost
mechanism (`D2/S1/B1/G1`). Principal's $175 million refers to a pending
acquisition target's 2025 premiums; Avery's approximately $75 million is an external
minority investment; Apollo's $35 billion compute facility belongs to managed
funds and affiliates. CBRE and AES provide customer-demand or unseparated fund
context. No issuer adds dedicated internal AI dollars, a measured financial
benefit, quantified AI-specific 2027 guidance, or a spend-return pair.

The twenty-fifth integrated pass used exact filing and official transcript
evidence retained through public source archive and one read-only independent review for
AT&T, AbbVie, Chevron, Tyson Foods, Mastercard, Caterpillar, NextEra Energy,
and Intuitive Surgical. All eight remain adopter-headline eligible. AT&T adds
`D2/S2/B3`: its AI Gateway is already “saving millions” and can reduce scoped
AI costs by as much as 90%, but the vague magnitude, ceiling rate, and lack of
a financial-statement bridge prevent `B4`. Tyson supplies unquantified testing
and expansion evidence. AbbVie, Chevron, Mastercard, Caterpillar, NextEra, and
Intuitive provide bundled infrastructure, risk, intended productivity,
customer-facing product, or supplier-demand evidence without a qualifying
financial claim. No issuer adds `B4`, `S3+`, quantified AI-specific 2027
guidance, or a spend-return pair.

The twenty-sixth integrated pass used exact SEC filing snapshots from public source archive and one read-only independent review for Interactive Brokers, Darden,
CMS Energy, IDEX, Tyler Technologies, Intercontinental Exchange, Invitation
Homes, and Ventas. All eight remain adopter-headline eligible at
`D1/S0/B0/G0/J0`. IDEX supplies a real customer-demand signal, but management
combines AI-driven data-center and semiconductor demand with space, defense,
price, acquisitions, and rebate effects, so no isolated AI revenue or profit
amount is accepted. Tyler and ICE provide product or market-opportunity
language without a purchased AI layer or provider materiality. The remaining
five issuers provide market, governance, tenant, implementation-ability, or
risk language only. No issuer adds `S2+`, `B1+`, AI-specific 2027 guidance, or
a spend-return pair.

### Airbnb — realized direct expense effect

Airbnb's Q2 10-Q attributes a $17 million quarterly reduction in third-party
service-provider costs to fewer support contacts resulting from increased AI
use. The six-month reduction was $15 million. The parent operations-and-support
expense line nevertheless increased by $29 million in Q2 and $52 million over
six months. Airbnb therefore supplies a clean gross expense bridge, but not net
AI ROI.

Primary source:
[Airbnb Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1559720/000155972026000027/abnb-20260630.htm).

### Equifax — strongest forward savings target

Equifax doubled its 2026–2028 AI-driven cost-reduction target from $75 million
to $150 million. Management expects benefits to accelerate through 2027 and
2028, but the official investor presentation says the target combines AI with
overall process productivity, affects both expense and capital, is measured
against previously expected spending growth, and may be reinvested. It is
forward `B3/G3` evidence—not a realized AI-only $150 million benefit.

Primary source:
[Equifax Q2 2026 release](https://investor.equifax.com/news-events/press-releases/detail/1414/equifax-delivers-strong-11-growth-in-second-quarter-2026).

### Sysco — quantified 2027 bridge with mixed attribution

Sysco described approximately $100 million of FY2027 AI-related and
previously-announced cost-out, net of investment and flowing to the bottom
line. Its official deck splits the $100 million into approximately $55 million
of new benefits and $45 million of carryover. A secondary transcript cites
approximately $160 million of run-rate value for the same work; that amount is
not in the SEC exhibit or official slides and is not additive. The AI portion
is not separable, so the full amount cannot be classified as AI-only savings.

Sources:
[Sysco FY2026 release](https://www.sec.gov/Archives/edgar/data/96021/000009602126000026/syy2026q4pressrelease.htm)
and
[earnings slides](https://investors.sysco.com/~/media/Files/S/Sysco-IR/documents/events-and-presentations/q4-fy26-earnings-slides.pdf).

### Booking Holdings — quantified spending share and unit-cost effect

Booking said AI remained a low-single-digit percentage of technology spending.
Management connected AI to lower customer-service cost per booking, while
prepared remarks described a double-digit decline from combined AI and human
capabilities. Management also said AI and technology cost per merge request
was declining. These are current direct-unit-cost effects, but neither has an
aggregate dollar amount. A directional statement that AI ROI was positive
does not qualify as quantified capital justification. The $650 million
transformation target and approximately $100 million primarily-2027 increment
were not attributed to AI.

Primary sources:
[Booking Q2 2026 transcript](https://s25.q4cdn.com/383369491/files/doc_financials/2026/q2/BKNG-Q2-2026-Earnings-Call-Transcript.pdf)
and
[10-Q](https://www.sec.gov/Archives/edgar/data/1075531/000107553126000037/bkng-20260630.htm).

### AT&T — quantified unit economics, incomplete P&L bridge

AT&T reported approximately 45 billion AI tokens processed daily, AI-cost
reductions of as much as 90%, and savings described as "millions." It did not
provide an exact amount, measurement period, baseline, net implementation cost,
or reported-expense reconciliation.

Primary source:
[AT&T, The Tokenomics Equation](https://about.att.com/blogs/2026/the-tokenomics-equation.html).

### CVS — material but inseparable savings

CVS reported more than $1 billion of OpEx savings through technology
efficiencies and AI. Its disclosed workflow results are operationally material,
including a greater-than-20% claims-processing-time target. The company did not
isolate AI's share of the savings, so the skeptical review retained the
operational results but rejected strict `B4`. The greater-than-$20 billion
decade technology commitment and $770 million of annual Caremark technology
spending are broad envelopes, not AI budgets. The preliminary 2027 EPS floor is
enterprise-wide.

Primary source:
[CVS Q2 2026 transcript](https://s206.q4cdn.com/752775519/files/doc_financials/2026/q2/2Q-Transcript.pdf).

### Apple — AI is a named R&D driver, not a quantified budget

Apple's quarterly R&D expense rose from $8.866 billion to $11.729 billion, and
nine-month R&D rose from $25.684 billion to $34.035 billion. The filing says
the increases were primarily driven by higher infrastructure costs, including
AI investment, and headcount. It does not isolate AI's share of the $2.863
billion quarterly or $8.351 billion nine-month increase. Management separately
described rising AI-related OpEx and COGS without amounts. External Siri AI and
customer outcomes are product/provider evidence, not internal-adopter savings.
No realized AI P&L result, quantified return, payback, or 2027 AI guide was
disclosed.

Primary source:
[Apple Q3 FY2026 10-Q](https://www.sec.gov/Archives/edgar/data/320193/000032019326000020/aapl-20260627.htm).

### FedEx — qualitative cost mechanism, not a measured result

FedEx says AI-enabled workflows are producing faster responses, fewer
exceptions, and lower cost to serve. It does not disclose a baseline,
measurement period, amount, rate, expense line, or isolated AI contribution.
The evidence therefore remains `B1`, not a realized financial result. Broader
company savings targets, the fiscal-2027 commitment outlook, and $3.9 billion
of planned CapEx lack AI attribution.

Primary source:
[FedEx FY2026 10-K](https://www.sec.gov/Archives/edgar/data/1048911/000104891126000105/fdx-20260531.htm).

### WTW — quantified transformation plan, inseparable AI contribution

WTW's Propel plan expects approximately $625 million of cash costs and
$25 million of noncash charges through 2028, with approximately $400 million
of gross and $350 million of net annualized run-rate savings by the end of
2028. The program combines AI-system implementation with automation, process
redesign, severance, vendor termination, and legacy-system changes. WTW did
not disclose an AI-only amount, a realized AI P&L effect, a quantified 2027
amount, or a comparable AI spend-return pair. The result is forward `B3/G3`
evidence, not ROI.

Primary source:
[WTW Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1140536/000119312526325374/wtw-20260630.htm).

### Axon — quantified provider economics, excluded from adopter rates

Axon's filing attributes year-over-year technology-license expense increases
of $4.4 million in quarterly SG&A and $5.1 million quarterly/$6.5 million
six-month in R&D to AI initiatives. Those figures overlap across periods and
do not distinguish internal tooling from customer-facing AI products. Axon's
AI Era revenue grew nearly 700%, but the company did not disclose AI revenue
dollars, profit, margin, quantified 2027 economics, or return. Axon is
classified `OUT_AI_TECH_VENDOR`; the evidence is retained in the provider
appendix and excluded from adopter numerators.

Primary source:
[Axon Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1069183/000162828026053646/axon-20260630.htm).

### Cigna — quantified customer savings, not issuer P&L

Cigna reported approximately $2,000 of lower annual medical costs per engaged
customer and 42% fewer avoidable inpatient stays. Management separately
estimated $200 million of customer medical-cost savings over three years.
These outcomes combine AI, predictive models, clinicians, data, and pharmacy
capabilities. The disclosure does not isolate AI's contribution or establish a
Cigna income-statement benefit. The three-year estimate cannot be annualized
or assigned to 2027. No in-window AI spending amount or comparable return
measure was disclosed.

Primary sources:
[Cigna Q2 2026 transcript](https://s202.q4cdn.com/757723766/files/doc_earnings/2026/q2/transcript/2Q26-transcript.pdf)
and
[Cigna Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1739940/000173994026000065/ci-20260630.htm).

### Tesla — large AI asset balances, no isolated spend flow

Tesla reported $10.823 billion of gross AI-infrastructure PP&E at June 30,
2026, versus $6.816 billion at year-end 2025. The mechanical $4.007 billion
difference is not a disclosed additions or cash-spending flow. Tesla also
acquired an AI-hardware company for $1.95 billion of equity-funded
consideration. First-half cash CapEx of $8.282 billion and the greater-than-
$25 billion 2026 CapEx guide blend AI with factories, manufacturing, R&D,
fleet, and other infrastructure. Provider-side FSD economics are positive but
unquantified; no AI profit, 2027 financial amount, or return/payback was
disclosed.

Primary source:
[Tesla Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1318605/000162828026049270/tsla-20260630.htm).

### Intel — charge nonrecurrence is not an AI benefit

Intel's DCAI operating income benefited from $512 million of lower period
charges, primarily because prior-year Gaudi inventory charges did not recur.
A separate $294 million consolidated inventory-charge comparison is blended
with $226 million of lower 18A reserves and an offsetting current charge. The
amounts overlap and cannot be added or treated as value created by current AI.
Intel's $637 million Mentee Robotics acquisition and AI-driven business growth
remain provider-side. Directionally higher 2027 CapEx is company-wide and
unquantified.

Primary source:
[Intel Q2 2026 10-Q](https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/intc-20260627.htm).

## Provider comparison

Provider economics are excluded from adopter headlines. Microsoft reported
$41 billion of quarterly CapEx, approximately two-thirds for CPUs and GPUs
serving AI and non-AI workloads, and expects FY2027 CapEx to grow. Meta reported
$31.1 billion of quarterly CapEx and $130–145 billion of 2026 CapEx guidance.
Both reported commercial or operational AI outcomes; neither disclosed an
isolated AI profit contribution or an auditable payback schedule.

Tesla, Intel, and Axon reinforce the same pattern. All three disclose
quantified AI-linked assets, acquisitions, expenses, or product growth, but
none supplies an isolated AI profit or margin amount. Tesla's PP&E is a
balance rather than a spending flow, Intel's favorable charge comparison is
nonrecurrence rather than AI-created benefit, and Axon's disclosed expense
increases cannot be separated between internal tools and customer products.

Sources:
[Microsoft FY2026 Q4](https://www.microsoft.com/en-us/Investor/events/fy-2026/earnings-fy-2026-q4)
and
[Meta Q2 2026 transcript](https://s21.q4cdn.com/399680738/files/doc_financials/2026/q2/META-Q2-2026-Earnings-Call-Transcript.pdf).

## Final conclusion

Explicit AI discussion is widespread in filings, but strict financial evidence
is rare. Among 411 adopter-headline issuers, 47 disclose substantive internal
AI mechanisms, 18 disclose an investment or cost mechanism, three reach S3+,
and only Incyte supplies a total ring-fenced AI dollar-spend level. Thirteen
report a quantified operational or financial result. Airbnb and Booking are
the two direct-expense B4 cases; Cigna adds a quantified customer medical-cost
outcome without an issuer-P&L bridge. Broadridge is the sole strict,
independently confirmed AI-specific FY2027 financial-guidance case. No adopter
provides a complete comparable spend-to-return reconciliation. Provider
investment and revenue evidence is often larger, but it remains outside the
adopter denominator and generally lacks an isolated AI bottom-line return.