MMM · 3M Company

Conglomerates · mkt cap $79.7B · calls: Q1 FY2026 vs Q4 FY2025

42.0 conviction · conf-adj 39

conf 3/10 partial

enthusiasm:24.0 · trend:8 · quantifies:5 · impact:0 · under_radar:0 · credibility:5 · confirmation:0

Enthusiasm latest 8 / prev 6 (rising)

3M's AI thesis became more concrete in Q1 FY2026: management linked AI to sales coaching, churn reduction, customer solution discovery, predictive attrition models, automated visual inspection, and data center demand. The strongest quantified AI-adjacent business impact is in data centers and EBO, with approximately $600 million of current revenue, a $1 billion-plus addressable market, and capacity expansion to support AI demand. Credibility is moderate to high because management provided specific use cases and some hard figures, though most productivity and cost claims are bundled with broader automation and operational excellence rather than isolated AI contribution.

GROUNDED NEXT-FY IMPACT vs CONSENSUS

Grounded on actual base — revenue $24.9B · net income $3.2B · net margin 13.0% · diluted EPS 6

These are next-fiscal-year annual uplift estimates, not next-quarter numbers.

Aggregate next-FY est. rev uplift: 0.0% · next-FY EPS uplift: % · vs analysts: inline · priced in: high · confidence: 3/10

Claim	Figure	Arithmetic	Next-FY Rev %	Next-FY EPS %
Cost of quality 5.4% (2026) → <4% over time cost	5.4% COQ in 2026, <4% over time	COQ as % of revenue (base disclosed → HARD). Disclosed over-time delta 5.4%→4.0% = 1.4pts × $24,948M = $349.3M pretax; @21% tax → $275.9M after-tax = 5.84% of FY26 adj NI ($4,726M). But the improvement is explicitly 'over time' (multi-year) and 2026 is already the 5.4% year with NO 2025 COQ baseline disclosed, so the NEXT-FY contribution is not sizable (X phased ~1/3→1.95%; Y left next-FY null — consensus uses null). eps_uplift_pct here records the full over-time prize.	0	5.84
Automation investment >$250M over 3 yrs cost · soft	>$250M over 3 years (~$83M/yr)	Cash investment/use of capital (~$83.3M annual), not a quantified P&L saving; no saving figure attached and the resulting labor/productivity benefit is unquantified. Cannot translate spend into EPS uplift.
Manual labor pool: 600 inspectors + 500 slitters + 7,000 handlers productivity · soft	~8,100 manual roles automatable (>600 inspection operators disclosed)	Headcount disclosed but NO fully-loaded wage/cost per role disclosed anywhere in the inputs. Cannot size $ saving without inventing a wage base. Large theoretical pool but unanchored.
Automated slitting +30% sq-yds/hr (Novato) productivity · soft	30% productivity at one facility (~500 manual slitters base)	Single-site proof point; 30% on ~500 slitters ≈ 150-slitter-equivalent capacity, but no revenue/slitter, labor/slitter, or network rollout timing disclosed → cannot extrapolate to a $ figure. Directionally supports the COQ/automation thesis, not separately sizable.
Data center + power utility revenue ~$600M (current) revenue	$600M ($100M inside-DC, $500M power)	SUPPLIER-side, CURRENT run-rate already in the base: $600M/$24,948M = 2.4% of revenue (no growth rate disclosed → no next-FY incremental uplift). $500M is low-margin power passthrough (2.0%); only the $100M inside-DC piece (0.4%) is genuine AI-tech content. As current revenue it is 0 incremental uplift (Y instead reported its 2.4% rev / 1.8% EPS share).	0	0
EBO 'more than double' capacity / $1B+ TAM / order in hand revenue	more than double capacity; $1B+ TAM	SUPPLIER-side. $1B+ is a TAM (TAM≠revenue; 4.0% of rev as scale only — no share/timing). 'More than double capacity' on the ~$100M inside-DC proxy implies up to +$100M potential ≈ 0.4% of rev @13.0% margin ≈ 0.30% EPS, but capacity≠revenue and the timeframe is unspecified (not necessarily next FY). Real optionality that mostly lives beyond next FY; low confidence.	0.401	0.301
EBO install 80% less time productivity · soft	80% less install time	Customer-side install efficiency driving take-rate; no install labor cost, price premium, conversion, or volume base disclosed → cannot translate to revenue/EPS. Supports EBO demand, not separately sizable.
Ask 3M AI assistant (reach broader customer base) engagement · soft	qualitative ('broader population')	No quantified reach, conversion, or revenue figure. Unanchored engagement claim.

Assumptions: Revenue denominator $24,948M (2025, latest FY). EPS uplift sized against the forward consensus adjusted base the estimates use — FY26 adj net income $4,726M / EPS 8.69 (NOT 2025 $4.327B/8.032, and NOT depressed GAAP $3.25B/6.0). Current net margin ~13.0% used for incremental flow-through; cost savings tax-effected at 21%. Supplier-side AI (data center, power, EBO) is excluded from the adopter headline; COQ, automation, slitting, labor, Ask 3M are adopter. COQ base (revenue) is disclosed → HARD; only the 5.4%→<4% delta (1.4pts) is computable as a multi-year 'over time' prize ($276M after-tax = 5.84% of adj NI), but 2026 is already the 5.4% year with no 2025 baseline, so next-FY phasing is unsizable → est_eps_uplift_pct=null. No bookings-to-revenue conversion needed (claims disclose current revenue, TAM, capacity, or targets, not bookings).

Top line: Adopter side: no quantified revenue uplift — the COQ/automation/slitting program is a margin story and Ask 3M's customer-reach claim is unanchored (soft). Supplier side: real but mostly already in the base — ~$600M current data-center/power run-rate (2.4% of $24,948M), of which $500M (2.0%) is low-margin power passthrough and only ~$100M (0.4%) is genuine inside-DC AI-tech content. EBO's 'more than double capacity' against a $1B+ TAM with an order in hand is genuine optionality (~+0.4% if the inside-DC piece doubles), but TAM≠revenue, capacity≠revenue and timing is unspecified — sizable supplier upside lives beyond next FY. Net next-FY supplier rev uplift ~0.4%, low confidence.

Bottom line: The one anchored, hard impact is cost of quality: 5.4%→<4% of revenue is a 1.4pt swing = $349M pretax / $276M after-tax = 5.84% of FY26 adjusted net income — the multi-year prize. But it is explicitly 'over time' and 2026 is already at 5.4% with no disclosed 2025 baseline, so the defensible NEXT-FY contribution is unsizable (X phased ~1/3≈1.95%; consensus leaves it null). The >$250M automation spend and ~8,100-role manual pool are real but unsizable (no wage base disclosed). Bottom-line AI signal: modest-but-credible margin expansion over time, not a next-FY step-change.

Consensus already models FY26 revenue ~$25.08B (+3.5% over 2025) and FY26 EPS ~8.69 (+8.2%). The $600M data-center/power figure is current supplier revenue already in that base; hard adopter-side next-FY revenue uplift is 0 and hard next-FY EPS uplift cannot be calculated from the disclosed claims (COQ's over-time prize lands beyond next FY). The AI program is thus largely priced in and inline with consensus, with genuine but unsized multi-year optionality in COQ margin and EBO capacity.

MODEL CONSENSUS (impact)

partial

Margin story, mostly priced in; COQ over-time prize (5.84% adj NI) and EBO optionality are real but land beyond next FY, so next-FY EPS uplift left null.

Conflicts reconciled

est_eps_uplift_pct: X=1.95 vs Y=null -> used null because 2026 is explicitly the 5.4% COQ year with no 2025 baseline, so the next-FY delta is unsizable
COQ/EPS base: X=$4,726M (FY26 fwd) vs Y=$4.327B (2025) -> used $4,726M, the forward consensus adjusted basis the task requires
supplier_rev_uplift_pct: X=0.4 vs Y=2.405 -> used 0.4; $600M is current/in-base, only EBO capacity doubling is incremental
data-center rev_uplift_pct: X=0 vs Y=2.405 -> used 0; current run-rate already in base, not an uplift
vs_analyst_expectations: X=inline vs Y=unclear -> used inline; priced_in high and growth already modeled
confidence: lowered to 3 (Y=4) due to the central next-FY EPS conflict

Field	Opus 4.8	GPT-5.5
Rev uplift %	0	0
EPS uplift %	1.95	–
Priced in	high	high
vs analysts	inline	unclear
Confidence	5	4
Top line	On the ADOPTER side there is essentially no quantified revenue uplift — the productivity/COQ program is a margin story, and Ask 3M's customer-reach claim is unanchored (soft). Supplier-side AI revenue is real but mostly already in the base: ~$600M current data-center/power run-rate (2.4% of $24,948M rev), of which $500M (2.0%) is low-margin power passthrough and only ~$100M (0.4%) is genuine inside-DC AI-tech content. EBO's 'more than double capacity' against a $1B+ TAM with an order in hand is genuine optionality, but TAM≠revenue and the timeframe is unspecified — sizable supplier upside lives beyond next FY. Net next-FY supplier rev uplift ~0.4% (doubling the inside-DC piece), low confidence.	No hard adopter-side AI revenue uplift is disclosed. The quantified AI-linked revenue is supplier-side: current data center/power revenue of $600M equals 2.405% of 2025 revenue, split between $100M inside data centers and $500M power-to-facility.
Bottom line	The one anchored, hard impact is cost of quality: improving from 5.4% to <4% of revenue is a 1.4pt swing = $349M pretax / $276M after-tax = 5.84% of FY26 adjusted net income (the multi-year prize). But that move is explicitly 'over time' and 2026's own step-down lacks a disclosed baseline, so the defensible NEXT-FY contribution is the phased ~1/3 ≈ $92M after-tax ≈ 1.95% of adjusted EPS. The >$250M automation spend and the 8,100-role manual pool are real but unsizable (no wage base disclosed). So the bottom-line AI signal is modest-but-credible margin expansion, not a step-change.	The clearest adopter-side dollarizable claim is cost of quality: moving from 5.4% of revenue to less than 4% over time would be at least $349.3M pre-tax and $275.9M after tax, or 6.377% of the adjusted NI base, but management did not disclose the current baseline or next-FY phasing.
Reasoning	Consensus already models margin expansion: FY26 EPS growth +8.2% vs revenue +3.5%, FY27 EPS +9.2% vs revenue +3.3% — EPS compounding ~2.5x revenue means analysts already assume the operating-margin lift that the COQ/automation 'integrated operating company' program is meant to deliver. The hard COQ contribution (~1.95% adj EPS next FY, ~5.8% over time) fits comfortably inside that embedded margin-expansion path rather than exceeding it. Supplier-side (~0.4% next-FY rev, $1B+ EBO TAM) is the only genuinely under-modeled optionality, but it is too small and too unanchored on timing to move the headline. Net: the AI story corroborates consensus rather than beating it.	Consensus already models 2026 revenue of $25.084B, up $844.2M or 3.483% from 2025 consensus revenue, and 2026 EPS of $8.69263, up 8.219% from 2025 consensus EPS. The $600M AI data center/power figure is current supplier-side revenue already in the base, while hard adopter-side next-FY revenue uplift is 0 and hard next-FY EPS uplift cannot be calculated from the disclosed claims.

Rows highlighted where the two models disagreed.

QUANTIFICATIONS

automation investment: more than $250 million (over the next 3 years, bottomline)
“At the same time, we're investing more than $250 million over the next 3 years in standard, easy-to-replicate automation across our plants and distribution centers.”

manual inspection labor pool: over 600 operators (current network, bottomline)
“To illustrate the opportunity, we have 7,000 material handlers and over 600 operators performing manual visual inspections across our network and about 500 manual slitters.”

automated slitting productivity: 30% increase (late last year, bottomline)
“When we automated the slitting operation at our [ Novato ] facility late last year, we achieved a 30% increase in square yards per hour productivity.”

data center and power utility revenue: approximately $600 million (current, topline)
“I also want to highlight our growing data center and associated power utility business with current revenue of approximately $600 million, $100 million inside the data center and about $500 million bringing power to the facility.”

inside data center revenue: $100 million (current, topline)
“I also want to highlight our growing data center and associated power utility business with current revenue of approximately $600 million, $100 million inside the data center and about $500 million bringing power to the facility.”

power-to-facility revenue: about $500 million (current, topline)
“I also want to highlight our growing data center and associated power utility business with current revenue of approximately $600 million, $100 million inside the data center and about $500 million bringing power to the facility.”

EBO addressable market: $1 billion-plus (not specified, topline)
“With hyperscaler validation, a significant order in hand and $1 billion-plus addressable market, we're investing to more than double our capacity to support growing AI demand.”

EBO capacity expansion: more than double (not specified, topline)
“With hyperscaler validation, a significant order in hand and $1 billion-plus addressable market, we're investing to more than double our capacity to support growing AI demand.”

EBO installation efficiency: 80% less time (not specified, both)
“And we can put that together at 80% less time with a less strain technician; better reliability, can operate in a dusty environment, which is why it's gotten some good take rate.”

cost of quality target: 5.4% cost of quality in 2026 and less than 4% over time (2026 and over time, bottomline)
“We are leveraging Kaizen events, visual inspection systems, automation solutions, and AI-enabled models to optimize changeovers to improve quality, with a target of 5.4% cost of quality in 2026 and less than 4% over time.”

PAST (realized)

Q1 FY2026, William Brown: "When we automated the slitting operation at our [ Novato ] facility late last year, we achieved a 30% increase in square yards per hour productivity."
Q4 FY2025, Bill Brown: "We are leveraging Kaizen events, visual inspection systems, automation solutions, and AI-enabled models to optimize changeovers to improve quality, with a target of 5.4% cost of quality in 2026 and less than 4% over time."

CURRENT (now)

Q1 FY2026, William Brown: "We've introduced AI tools to drive growth, reduce churn and automate manual work, including an agent that analyzes our sales and opportunity pipeline data to develop customized coaching plans for sales managers to help reps meet their targets."
Q1 FY2026, William Brown: "I also want to highlight our growing data center and associated power utility business with current revenue of approximately $600 million, $100 million inside the data center and about $500 million bringing power to the facility."
Q1 FY2026, William Brown: "They have the same predictive models tailored for TEBG into that business."

FORWARD (guidance)

Q1 FY2026, William Brown: "And we believe digital tools like Ask 3M, a new AI-powered digital assistant that helps customers find solutions to design challenges using 3M products, will allow us to reach a broader population of customers."
Q1 FY2026, William Brown: "With hyperscaler validation, a significant order in hand and $1 billion-plus addressable market, we're investing to more than double our capacity to support growing AI demand."
Q4 FY2025, Bill Brown: "Like we previewed at Investor Day, this phase includes reengineering the structural cost base that underpins our supply chain network and business processes, simplifying and standardizing core activities, and embedding an AI-first mentality as we shift from a holding company model to an integrated operating company."

TRACK RECORD — PROMISE vs DELIVERY

60/100 track record delivers 6 calls reviewed

3M's quantified AI-related promises are mostly embedded in broader operational and commercial targets rather than framed as standalone AI goals. Using the broader AI-substance standard, the judgeable items show delivery or positive tracking, though several larger targets remain too early and some linkage to AI is indirect.

Reduce cost of poor quality to 5.4% in 2026 and toward ~4% over time, with AI-enabled changeover/quality models among the levers — promised Q2 FY2025
partial Later calls showed improvement and reiterated progress, but the 2026 and longer-term targets had not fully matured within the transcript set.

Automated slitting/inspection deployment at Novato would deliver a 30% productivity gain — promised Q1 FY2026
delivered Management reported the Novato automated slitting line achieved a 30% increase in square yards/hour productivity after going live.

Invest more than $250M over three years in standardized automation including automated visual inspection and slitting/material-handling automation — promised Q1 FY2026
too-early The program had just been announced, with only early proof points available, so the three-year target was not yet judgeable.

Use predictive churn analytics to support the $100M/50-pair cross-sell goal through 2027 — promised Q1 FY2025
partial Later calls cited deployment and win-backs, with cross-sell progress rising toward the 2027 target but not yet complete.

PRICED-IN (REFINED)

UNKNOWN

Est. revisions unknown · Fwd P/E 19.0 · EV/Sales 3.5x

LLM synthesis failed: codex: OpenAI Codex v0.136.0 -------- workdir: /tmp/aes-codex model: gpt-5.5 provider: openai approval: never sandbox: danger-full-access reasoning effort: medi

COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS

RECENT AI CATALYSTS & NEWS

BUSINESS IMPACT - QUALITATIVE MATERIALITY

-/10 immaterial unclear · soft evidence

LLM materiality judgment failed: codex: OpenAI Codex v0.136.0 -------- workdir: /tmp/aes-codex model: gpt-5.5 provider: openai approval: never sandbox: danger-full-access reasoning effort: medium reasoning summari

OPTIONS / MARKET STRUCTURE

option liquidity: good

ATM IV

—

TYPICAL BID-ASK

—

OPEN INTEREST

—

proxy inputs — dollar-ADV $603M · beta 1.09 · px $152.90

source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.

CONFIRMATION — INSIDERS · 13F · LANGUAGE

Undercutting — insiders selling, institutions trimming, management language unknown.

INSIDERS selling 8 open-market sell(s) vs 0 buy(s) — net distribution

INSTITUTIONS (13F) trimming as of 2026-03-31: 145 new / 257 closed positions; 791 increased / 927 reduced; institutional ownership -3.41pp; -116 net 13F holders

MGMT LANGUAGE n/a unknown language analysis failed: codex: OpenAI Codex v0.136.0 -------- workdir: /tmp/aes-codex model: gpt-5.5 provider: openai approval: never sandbox: danger-full-access reasoning effort: medi

VERBATIM AI QUOTES

“We've introduced AI tools to drive growth, reduce churn and automate manual work, including an agent that analyzes our sales and opportunity pipeline data to develop customized coaching plans for sales managers to help reps meet their targets.”
— William Brown, Q1 FY2026

“And we believe digital tools like Ask 3M, a new AI-powered digital assistant that helps customers find solutions to design challenges using 3M products, will allow us to reach a broader population of customers.”
— William Brown, Q1 FY2026

“At the same time, we're investing more than $250 million over the next 3 years in standard, easy-to-replicate automation across our plants and distribution centers.”
— William Brown, Q1 FY2026

“By automating material handling in our warehouses, replacing manual slitters with automated systems and automating our current manual visual inspection processes, we are improving safety, reducing labor costs, increasing yield and putting ourselves in a better position to support demand as volumes recover.”
— William Brown, Q1 FY2026

“To illustrate the opportunity, we have 7,000 material handlers and over 600 operators performing manual visual inspections across our network and about 500 manual slitters.”
— William Brown, Q1 FY2026

“When we automated the slitting operation at our [ Novato ] facility late last year, we achieved a 30% increase in square yards per hour productivity.”
— William Brown, Q1 FY2026

“I also want to highlight our growing data center and associated power utility business with current revenue of approximately $600 million, $100 million inside the data center and about $500 million bringing power to the facility.”
— William Brown, Q1 FY2026

“With hyperscaler validation, a significant order in hand and $1 billion-plus addressable market, we're investing to more than double our capacity to support growing AI demand.”
— William Brown, Q1 FY2026

“I'm very pleased with the traction on the sales force, on pricing discipline, on cross-selling, on churn reduction and looking very hard at attrition with the predictive AI models that we have in place.”
— William Brown, Q1 FY2026

“They have the same predictive models tailored for TEBG into that business.”
— William Brown, Q1 FY2026

“We are leveraging Kaizen events, visual inspection systems, automation solutions, and AI-enabled models to optimize changeovers to improve quality, with a target of 5.4% cost of quality in 2026 and less than 4% over time.”
— Bill Brown, Q4 FY2025

“Like we previewed at Investor Day, this phase includes reengineering the structural cost base that underpins our supply chain network and business processes, simplifying and standardizing core activities, and embedding an AI-first mentality as we shift from a holding company model to an integrated operating company.”
— Bill Brown, Q4 FY2025

ANALYST QUESTIONS ON AI

Q (Q1 FY2026, Joe O'Dea): And then just on the transportation, electronics commercial excellence program, can you talk about where you are on that trajectory?
A: William Brown: "I'm very pleased with the traction on the sales force, on pricing discipline, on cross-selling, on churn reduction and looking very hard at attrition with the predictive AI models that we have in place."

Q (Q1 FY2026, Deane Dray): And I'd love to hear a bit more about the Expanded Beam Optics opportunity. There's a lot of focus on this. It's addressing the data transfer bottlenecks in AI processing. So just where do you stand competitively?
A: William Brown: "We've had at least a validation by at least one hyperscaler, a second one is in testing. I expect that will be positive as well. We had a fairly large order coming in, in Q1 relating to the hyperscaler that has certified it. We are in a ramp-up mode. We will double capacity towards the back end of the year."