Iron Mountain Incorporated (NYSE: IRM) — A Wide-Moat Records Annuity Wearing a Data-Center Multiple, Funding the Build With Debt

Report date: 2026-07-10 · Coverage: Initiation Price (2026-07-09 close, AZI adj): ~$119.48 · Market cap: ~$35.3B · Enterprise value: ~$54B (net debt + NCI ~$19.2B) FY2025: revenue $6,901.7M · Adjusted EBITDA ~$2,573.9M · AFFO $1,541.2M ($5.17/sh) · GAAP dil. EPS $0.49 (not meaningful) · net lease-adj. leverage 4.9x

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information — not investment advice. The analysis that follows takes no position and carries no price target; this block is the single exception.

Verdict: HOLD / own-the-annuity-but-not-here — the data-center re-rating is fully priced and then some. Accumulate only into a rate-or-AI scare toward ~$90–100 (≈16–17x forward AFFO), with a real table-pound in the low-$80s (where it traded in mid-2025). I’d fade/trim above ~$130. Not a short — you don’t short a 4x thematic winner with a genuine wide-moat annuity, a contracted data-center backlog, and accelerating fundamentals. Conviction: medium.

Tag: “The box that never leaves — now priced like the data center it’s building with borrowed money.”

Iron Mountain is two businesses wearing one multiple. The first is genuinely wonderful: a global records-storage annuity with a real, wide moat — the “box that never leaves” (~98% retention, ~15-year average box life, the customer bears the cost and risk of moving millions of cartons), ~740M cubic feet stored for ~240k customers including ~95% of the Fortune 1000, 37 consecutive years of organic storage-rental growth now driven by pricing (mid-single-digit revenue management on flat volume) rather than paper. It throws off ~$2.36B of segment EBITDA at a ~2%-of-revenue maintenance-capex intensity — a cash machine around a slowly-shrinking pond. The second is a debt-funded, subscale data-center build: 488 MW operating (~97% leased), 852 MW under construction/development, a ~$3.3B contracted-but-not-commenced backlog, and ~$1.75B/year of growth capex — a genuinely fast-growing but commodity-landlord business where Iron Mountain has no scale or cost edge against Equinix and Digital Realty, in the hottest capital cycle in the industry.

Here is the tension the price ignores: data centers are only ~12% of revenue and ~15% of EBITDA, yet the whole enterprise now trades at a data-center multiple — ~21.9x EV/EBITDA and ~23x trailing (~20.7x forward) AFFO, the 93rd percentile of Iron Mountain’s own price-to-sales history, near a June-2026 all-time high, up ~46% year-to-date. The factor tape reads it exactly right: this is idiosyncratic AI-thematic alpha (no systematic momentum loading), a crowded thematic trade carrying full REIT rate-beta. And the accounting flatters the story in a way that matters: AFFO deducts only the ~$147M of maintenance capex while adding back the entire ~$2.07B growth build, so it “covers” the dividend 1.7x — but the consolidated enterprise does not self-fund. In FY2025 operating cash flow of ~$1.34B did not even cover total capex of ~$2.2B, let alone the ~$0.92B dividend; the ~$1.8B gap was plugged with ~$2.4B of new debt at a rising ~6.25% marginal coupon. Total debt grew ~$2.7B in one year to ~$19B, interest coverage is a thin ~2.9x, book equity is negative, and consolidated ROIC (~5.9%) sits below the cost of capital. The 4.9x “net lease-adjusted leverage” the company markets is real but flattering; true net-debt/EBITDA is ~6.6x.

Framing: a wide-moat annuity re-rated to a data-center growth multiple it only half earns, on a levered balance sheet, near record highs. It is the DLR thesis — “own the AI landlord, just not at top-of-cycle rent” — with two extra twists: Iron Mountain’s data-center arm is subscale (~1/7th of DLR’s revenue), and the moat lives entirely in the boring records segment the market is no longer paying attention to. The sum-of-the-parts confirms it: value the records annuity at a storage-REIT multiple, the data center at a data-center multiple, ALM at a services multiple, net the ~$19B of debt, and you land at ~$67–97/share on trailing numbers — below the ~$119 quote. The ~$6–15B gap is the market crediting the growth plan and the development pipeline as already delivered. The business is genuinely excellent and accelerating (Q1-2026 organic revenue +17%, the fastest in 25 years); the price simply leaves no margin of safety and a lot of rate/execution risk.

What flips me bullish: a ~20% rate-or-AI-scare de-rate into the ~$90s, or proof the enterprise turns self-funding — retained cash flow after the build and dividend going durably positive as data centers stabilize. What flips me bearish: data-center lease-up or renewal spreads fading as record supply floods the segment, ALM’s memory-price-driven surge reversing, or long rates backing up against 2.9x coverage — any of which de-rates a 20x+ AFFO name hard.

📈 Stock Price Action — Five-Year Event Map

Factual price history — no recommendation, no price target. Price moves are Fact; attributed drivers are Interpretation.

Over the trailing ~60 months Iron Mountain compounded roughly 4x — from a five-year low of ~$34.6 (Jul-2021) to an all-time high of ~$133.1 (June-2026), and trades at ~$119.5 now, ~10% below the peak. The 52-week range is ~$77.7–$133.1; the 200-day EMA sits ~$108. Beta is ~0.99 and the stock carries meaningful REIT rate-sensitivity — the re-rating has been powered by the AI/data-center demand narrative layered on a steadily-growing records annuity.

Cycle narrative. (1) Iron Mountain recovered off its 2021 low on reopening volumes. (2) It rode out the 2022 rate-shock REIT selloff roughly flat and launched Project Matterhorn — the growth strategy that reframed the equity story. (3–4) 2023–2024 were the re-rating: Matterhorn execution, the data-center build, and the AI-demand narrative drove the stock from the $40s to $100, at one point +94% in a year. (5) An April-2025 tariff-shock crash briefly halved the momentum (to ~$74) as a levered REIT sold off hard, then rebounded. (6) A rate/higher-for-longer wobble pulled it back to ~$78 in H2-2025. (7) The dominant recent move is the ~62% run from ~$82 to a June-2026 all-time ~$133 on accelerating AI data-center demand and beat-and-raise prints, before (8) a ~10% consolidation to ~$119.5. The stock sits ~10% off a record, having quadrupled in five years. (Price moves are FACT from the AZI five-year series; attributed drivers are INTERPRETATION cross-referenced to earnings dates, 8-K events, and the news feed. Note the stock’s high rate-sensitivity — the two sharp drawdowns were rate/macro events, not company events.)

1. Executive Summary

Iron Mountain is a ~$35B-market-cap specialized REIT built on a global records and information management (RIM) annuity, now re-rated on a fast-growing data-center platform and a cyclically-surging asset-lifecycle-management (ALM) business. It reports two segments — Global RIM (records/data management, digital/DXP, secure shredding; FY25 revenue $5.29B, Adjusted EBITDA ~$2.36B, 44.7% margin) and Global Data Center (FY25 revenue $803M, Adjusted EBITDA $416M, 51.8% margin) — with ALM, fine art, and corporate in “Corporate & Other” (revenue ~$807M). Recurring storage-rental revenue is ~59% of the total ($4.05B) and the economic core.

The records business is a genuine wide-moat annuity. With ~740M cubic feet stored, ~240k customers across 61 countries (~95% of the Fortune 1000), and ~98% retention, the moat is textbook Greenwald customer-captivity plus scale: the customer bears the cost and risk of moving records, so the “box that never leaves” compounds. Volume is flat-to-slightly-down, but Iron Mountain has now delivered 37 consecutive years of organic storage-rental growth, entirely price-led (mid-single-digit revenue management). Maintenance capex is ~2% of revenue — a cash machine.

The growth engine is data centers and ALM — real, but different in kind. Data centers (488 MW operating, ~97% leased; 852 MW under construction/development; ~$3.3B contracted backlog; 10.3-year WALE) grew revenue ~30% in FY25 and are guided >$1.0B (+25%) for FY26 — but this is a commodity-landlord business where Iron Mountain is subscale (~1/7th of Digital Realty) and competes in the hottest capital cycle in the industry. ALM (IT-asset disposition + component recycling) surged ~63% in FY25 and is guided to ~$950M in FY26, but ~40% is hyperscale and memory/HDD component pricing is 40–50% of that — a cyclical, commodity-exposed tail with a durable enterprise leg underneath.

The financials are strong on the surface and levered underneath. FY2025 AFFO was $1,541.2M, or $5.17/share (+13.9%), guided to ~$5.79 (+11%) for FY2026; Adjusted EBITDA was ~$2,574M (37.3% margin). But AFFO flatters: it deducts only ~$147M of maintenance capex while adding back the entire ~$2.07B growth build (84% data centers), so its 1.7x dividend “coverage” is an artifact. On a true cash basis the enterprise does not self-fund — FY25 operating cash flow (~$1.34B) did not cover total capex (~$2.2B), let alone the ~$0.92B dividend; the ~$1.8B gap was funded with ~$2.4B of new debt. Total debt rose ~$2.7B to ~$19B, net lease-adjusted leverage is 4.9x (true net-debt/EBITDA ~6.6x), interest coverage is ~2.9x, book equity is negative, and consolidated ROIC (~5.9%) is below the cost of capital.

The valuation prices the growth as delivered. At ~$119.48, Iron Mountain trades at ~21.9x EV/EBITDA, ~23x trailing / ~20.7x forward AFFO, and the 93rd percentile of its own price-to-sales history, near an all-time high. A sum-of-the-parts (records annuity at a storage multiple, data center at a data-center multiple, ALM at a services multiple, less ~$19B net debt) yields ~$67–97/share on trailing numbers — below the quote; the gap is the market crediting the multi-year growth plan and the development pipeline in full. The business is excellent and accelerating (Q1-2026 organic +17%, the fastest in 25 years); the price simply carries no margin of safety and full rate/execution risk. The forward question is not whether Iron Mountain is a good business — it is — but whether a ~15%-data-center enterprise deserves a data-center multiple funded with debt. No recommendation or price target appears below; valuation is treated as embedded expectations and scenarios.

2. Business Overview (§7.1)

What Iron Mountain is. Founded in 1951 (a converted mushroom-farm records depository) and a REIT since 2014, Iron Mountain stores and manages information and physical assets for ~240,000 customers across 61 countries. It now reports two segments: Global RIM — physical records storage, data management (tape/media), secure shredding, consumer storage, fine art, and the fast-growing digital solutions business (the InSight/DXP intelligent-document platform) — and Global Data Center. ALM (asset lifecycle management), fine art, and corporate items sit in “Corporate & Other.”

How it makes money. The economic core is recurring storage-rental revenue — ~59% of FY25 revenue ($4.05B) — high-margin, contractual, and sticky, complemented by service revenue (~$2.85B: handling, shredding, digitization, ALM transactions, data-center power/connectivity). Data-center revenue is ~99% storage-rental (lease/power); ALM is ~91% service (transactional, low-recurring).

Segment economics [FACT — FY2025 10-K, segment note]:

The records annuity. ~740M cubic feet stored; ~240k customers; ~95% of the Fortune 1000; ~98% retention; ~15-year average box life. Physical volume is flat-to-slightly-declining, but Iron Mountain has delivered 37 consecutive years of organic storage-rental growth through pricing (revenue management, “value not CPI”). Digital solutions (document digitization, the DXP platform, a five-year up-to-$714M U.S. Treasury/IRS contract) is the effort to convert the physical customer base into a recurring-software relationship — real, early, and unproven at scale.

Data center. 31 sites across 21 markets; 488 MW operating at ~97% leased; 852 MW under construction or held for development; 1,340 MW total potential; 10.3-year WALE; ~$3.3B contracted-but-not-yet-commenced backlog. Iron Mountain pre-leases before building (not a spec developer) and has deliberately chosen inference/cloud hyperscale over speculative LLM-training campuses. Recent leasing re-accelerated (13 MW in Q3-25 → 43 MW in Q4-25 → 32 MW in Q1-26), with FY26 leasing guided “over 100 MW.”

Project Matterhorn. The September-2022 growth strategy (a commercial/operating transformation) that reframed Iron Mountain as a double-digit grower; it cost ~$574M of cumulative “restructuring/transformation” opex and was declared complete in 2025 — a large adjusted-versus-GAAP add-back to watch.

Verdict (§7.1). A high-quality, high-retention records annuity (~59% recurring, ~2% maintenance capex) is the cash engine; data centers and ALM are the growth engines bolted on top. The parts are economically distinct — one a moaty annuity, one a capital-hungry commodity landlord, one a cyclical services business — which is why the sum-of-the-parts (§10) is the right analytical lens.

3. Industry Dynamics (§7.2)

Iron Mountain straddles three stacked industries with very different structures:

(A) Physical records storage — structurally good (for the incumbent). A consolidated oligopoly: Iron Mountain is the global #1, having acquired #2 Recall in 2016; Access Corp (private) is a distant second. This is a declining-supply annuity — physical box volume is flat-to-shrinking, but that helps the incumbent: no rational competitor builds new records warehouses into a declining-volume market, so the installed base is a defended, price-taking annuity. Switching costs are prohibitive. This is the Marathon capital cycle working for Iron Mountain — declining supply, disciplined pricing, no new entrants.

(B) Data centers — secular demand, but the hottest capital cycle in the market. Global data-center demand is genuinely strong (AI, cloud, inference), and wholesale/hyperscale landlords have enjoyed real pricing power on a power-and-land bottleneck. But the capital cycle is against new entrants: record private capital (~$45.7B into data centers in 2025, per the Digital Realty cross-read) is flooding exactly this segment. Iron Mountain is a subscale late entrant (~1/7th of Digital Realty’s data-center revenue) with no scale, cost, or connectivity edge versus Equinix or Digital Realty — its safety is pre-leasing (97% leased, contracted backlog), not a moat. When the power bottleneck eases and supply catches demand, development yields — a commodity return — mean-revert.

© ALM / IT-asset disposition — structurally weak. Fragmented, cyclical, and exposed to scrap/component (memory, HDD) pricing; it competes against customers’ own in-house IT teams and a long tail of local recyclers. The recent surge (+63% in FY25) is partly a memory/HDD price cycle (40–50% of the hyperscale portion), not a durable moat. The enterprise ITAD leg (360 Fortune-1000 logos) is the more durable part.

Synthesis (capital cycle). Iron Mountain funnels the cash from a declining-supply records annuity into a hot-supply data-center build. That is value-accretive while data-center leases are pre-let at good spreads and returns hold — and value-destructive if growth capex keeps flowing as supply floods the segment and yields compress. Consolidated ROIC (~5.9%) already sits below the cost of capital, a warning that the blended returns on the growth build have not yet proven themselves.

Verdict (§7.2): a good annuity industry, a hot/contested growth industry, and a weak services industry — blended. The records core is structurally excellent; the data-center and ALM growth engines are structurally more marginal and cyclical. The equity story rests on the growth engines; the durable value rests on the annuity.

4. Competitive Position (§4 / §7.3)

The records moat is genuine and wide. In Greenwald’s taxonomy it is customer captivity (switching costs) + scale: the 10-K itself describes moving records as “costly and time-consuming,” and the economics prove it — ~98% retention, ~15-year box life, and pricing power on a captive, flat-volume base. Iron Mountain’s global #1 scale (post-Recall) spreads a dense real-estate and logistics network no entrant can replicate into a declining market. This moat ties directly to ~$2.36B of Global RIM EBITDA that would erode without it — a real moat, doing real financial work, but wrapped around a flat pond (volume is not growing).

The data-center and ALM businesses have no comparable moat. The 10-K’s own description of data-center competition — “power, security, location, connectivity, rental rates” — is the language of a commodity landlord. Iron Mountain competes for hyperscale leases against Equinix, Digital Realty, and a wave of private capital, with no scale or connectivity advantage; its edge is executional (pre-leasing, existing power/land at some sites), not structural. ALM is a fragmented, cyclical services business with low switching costs.

The central tension. Data centers are ~12% of revenue and ~15% of reportable EBITDA, yet Iron Mountain trades at ~21.9x EV/EBITDA and the 93rd percentile of its own price-to-sales history — richer than storage REITs (Public Storage / Extra Space ~18–19x) and toward data-center territory (Digital Realty ~28x). The market is applying a blended growth/data-center multiple to an enterprise whose moat lives in the boring ~85%-of-EBITDA records-and-services base and whose growth lives in the no-moat, capital-hungry ~15%. Consolidated ROIC of ~5.9% (below WACC) is the quantitative expression of that mismatch: the annuity earns high returns on a tiny maintenance-capex base, but the enterprise’s incremental capital is being poured into a commodity build at unproven blended returns.

Verdict (§7.3): one genuine wide-moat annuity wrapped in a capital-hungry, no-moat data-center build on a levered balance sheet. The moat is real — but it is the records moat, not the data-center growth the price is celebrating. The right question is whether you are paying an annuity price or a data-center-growth price for the whole — and today it is the latter.

5. Growth History and Forward Opportunities (§7.4)

History. Revenue compounded from $4.15B (2020) to $6.90B (2025), an ~11% CAGR that has accelerated (+12.2% in FY25) — an unusually strong trajectory for a business rooted in physical records. Adjusted EBITDA grew to ~$2.57B and AFFO to $5.17/share. Q1-2026 organic revenue growth was +17% — the fastest in 25 years.

The three growth legs.

• Storage rental (the annuity, price-led): 37 consecutive years of organic growth, now ~mid-single-digit on flat volume via revenue management — durable and high-quality, the compounding base.
• Data centers (the re-rating driver): revenue +30% in FY25, guided >$1.0B (+25%) for FY26 and another 20%+ for 2027, with >$3.3B of contracted backlog “already in the bag,” >100 MW of FY26 leasing guided, and 400 MW energizing over 24 months. Real, contracted, and fast — but capital-intensive and commodity-priced.
• ALM (the cyclical surge): +63% in FY25, +92% in Q1-2026, guide raised to ~$950M for FY26; management calls it a future “multibillion-dollar business.” The enterprise leg (360 Fortune-1000 logos) is durable; the hyperscale leg (~40%, with memory 40–50% of that) is a component-price cycle.
• Digital solutions: the DXP/InSight platform plus the up-to-$714M U.S. Treasury/IRS digitization contract — the effort to monetize the physical customer base as recurring software; early and unproven at scale, but the highest-quality potential leg if it works.

Matterhorn targets. Management guides multi-year revenue and EBITDA CAGRs above 12–13% and “double-digit top and bottom line across cycles,” with a long-term frame of ~$8.1B revenue / ~$3.0B EBITDA.

Verdict (§7.4): genuinely high growth, mixed in quality. The storage annuity’s price-led growth is high-quality; digital solutions could be excellent; but the two legs currently exciting the market — data centers and ALM — are the most capital-intensive and cyclical, and their returns are unproven. The growth is real and accelerating; the debate (Valuation) is what to pay for a mix that is part durable annuity, part hot-cycle commodity build.

6. Financial Quality (§7.5)

Read this as a REIT — GAAP is meaningless. GAAP diluted EPS was $0.49 in FY2025 and book equity is negative (~−$981M ex-minority; tangible book ~−$24/share) after years of distributions exceeding GAAP earnings on a levered balance sheet. The relevant metrics are AFFO, Adjusted EBITDA, and cash flow.

Headline REIT metrics look strong. FY2025 AFFO was $1,541.2M, or $5.17/share (+13.9%), guided to ~$5.79 (+11%) for FY2026; normalized FFO was ~$1,080M (+15.8%); Adjusted EBITDA was ~$2,573.9M at a 37.3% margin (+15.1%). Storage-rental organic growth of ~10% (price, not volume) and a ~2%-of-revenue maintenance-capex intensity are the fingerprints of a high-quality annuity.

But the adjustments and the capex tell a more sober story. Company Adjusted EBITDA sits ~$113M above “clean” EBITDA (~$2,461M) because it adds back a recurring ~$196M “restructuring/transformation” line (present every year through Matterhorn), ~$140M of stock-based comp (~2% of revenue and growing), and ~$20M of deal costs — treat the add-backs skeptically. More important is the capex split: FY2025 growth capex was ~$2,068M (data centers ~$1,747M, 84% of total) against maintenance capex of just ~$147M. AFFO deducts only the ~$147M and adds back the entire ~$2.07B build — which is why AFFO “covers” the dividend ~1.7x.

The enterprise does not self-fund. The true FY2025 cash bridge: operating cash flow of ~$1,340M minus total capex of ~$2,215M = ~−$875M before the dividend; add ~$919M of dividends paid and the gap is ~$1.8B, funded by ~$2,384M of net new debt. Operating cash flow did not cover capex, let alone the dividend — 100% of the dividend and the data-center build is externally financed. This is a defensible model while growth capex earns attractive, contracted returns and the balance sheet de-levers on rising EBITDA — but it is a debt-funded growth machine, not a self-funding cash cow, and it is fully exposed to rates and capital-market access.

Balance sheet. Total debt ~$19.05B (up ~$2.68B in one year); net lease-adjusted leverage 4.9x (covenant max 7.0x; managed to ~5x for years) — but true net-debt/EBITDA is ~6.6x and total-debt/EBITDA ~7.7x. Weighted-average rate ~5.6%, interest coverage a thin ~2.9x, and marginal funding cost is repricing up (a June-2026 upsized $1.5B of 6.25% notes due 2035 versus a legacy book of lower coupons).

Verdict (§7.5): a high-quality records annuity wrapped in a levered capex machine. The core economics are genuinely excellent (recurring, high-retention, price-led, capex-light); the consolidated entity is not self-funding and carries real leverage and rate risk. AFFO is a fair operating metric but flatters the cash reality — the honest test is retained cash flow after the build and dividend, which is barely positive.

7. Capital Allocation (§7.6)

A coherent but debt-financed growth strategy. Iron Mountain runs an atypical REIT capital model: it funds growth almost entirely with debt, not equity (share count rose just ~0.8% in FY2025; no meaningful ATM, no buybacks), while paying a growing dividend and building data centers. That keeps per-share dilution low but concentrates the risk in leverage and rates.

The build. ~$2.07B of growth capex in FY2025, 84% data centers (~$1.75B) — the defining capital-allocation choice. Management pre-leases (97% leased, ~$3.3B contracted backlog) and touts ~51–52% data-center margins and +9–19% renewal spreads, which mitigates but does not eliminate the Capital Returns concern: Iron Mountain is deploying ~$1.75B/year into the hot hyperscale cycle where its moat does not apply and it has no cost/scale edge against Equinix and Digital Realty. Returns are contracted near-term but unproven across a full cycle and exposed to supply flooding the segment.

Dividend. Grown ~10%/year ($2.51 → $2.69 → $3.11/share; now ~$3.46 annualized after four consecutive hikes), at a ~60% AFFO payout — but ~85% of normalized FFO and ~6x GAAP net income. Safe on the AFFO definition; less comfortable once the debt-funded nature of the enterprise is acknowledged.

Leverage management. Held to ~5x net-lease-adjusted for years, now 4.9x (the company frames this as the lowest since pre-REIT-conversion) — genuine discipline given the build, but on a heavier true-leverage base (~6.6x) with thin ~2.9x coverage and rising marginal rates.

Incentives (a real positive). Short-term comp is keyed to Revenue / Adjusted EBITDA / AFFO-per-share / strategic goals; long-term comp to ROIC / Revenue / absolute-and-relative TSR. The ROIC metric in the LTI is a genuine guardrail on the data-center build — the right metric for a company whose central risk is over-deploying into a commodity cycle. (FY2025 STI paid 90.2%, with the AFFO/share component below internal target at 85.9% — some honest signaling.)

M&A and insiders. Tuck-ins only (ITRenew 2022, Regency 2023, Premier Surplus/ACT 2025 in ALM; data-center brownfields including full ownership of Web Werks India) — no transformational deals, no buybacks. Iron Mountain is widely held with no founder control (CEO William Meaney since 2013); an anti-hedge/pledge policy is in place. Insider Form 4 detail was not available in the local corpus — no conviction open-market signal to report either way.

Verdict (§7.6): competent and coherent for a growth REIT, but it is a levered growth machine. Disciplined ~5x leverage, ROIC in the comp plan, a ~60% AFFO payout, and minimal dilution are all genuine positives. The reservation is structural: a ~$2.1B/year build (84% data centers) plus a ~$0.92B dividend funded almost wholly with rising-cost debt, against negative equity and ~2.9x coverage — a bet that the hot data-center cycle keeps paying. Well-governed execution of an inherently aggressive capital plan.

8. Changes and Headwinds — Last Two Years (§7.7)

Strategic milestones.

• Project Matterhorn (launched Sep-2022) drove the transformation to double-digit growth; its ~$196M/year “restructuring/transformation” charges ended in 2025 (a tailwind to reported metrics going forward, and one fewer add-back).
• Data-center scale-up to ~1.3 GW of potential via brownfield deals (IO, EvoSwitch, full ownership of Web Werks India), plus large N. Virginia (195 MW dev capacity) and Chicago/Miami/Amsterdam leasing.
• ALM build-out via ITRenew (2022, hyperscale) and Regency (2023, enterprise), with 2025 tuck-ins — now guided to ~$950M.
• Digital/government: FedRAMP High achieved in Q1-2026 (the only vendor at that stage), and the U.S. Treasury/IRS digitization contract expanded (Sep-2025) to a five-year, up-to-$714M deal (~$45M FY26 ramp, >$100M/year from 2027).
• Capital markets: a near-continuous drumbeat of debt offerings (€1.2B 4.75% 2034 notes; a June-2026 upsized $1.5B of 6.25% 2035 notes) and four consecutive ~10% dividend hikes.

Leadership. CEO William Meaney (since 2013) and CFO Barry Hytinen are stable; a new IR head was appointed — no leadership shock.

Headwinds / watch-items. Rising marginal funding cost (6.25% vs. a lower legacy book) against 2.9x coverage; the ALM surge’s dependence on memory/HDD component pricing (~40–50% of the hyperscale portion) and hyperscale demand; data-center supply flooding the industry; and — the overarching one — a valuation that has migrated the risk from “is the business OK?” to “is the all-time-high multiple already pricing the AI/data-center/ALM optionality?”

Verdict (§7.7): the changes strengthen the operational thesis but tighten the price. Q1-2026 organic +17% (fastest in 25 years), de-levering, restructuring charges ending, a contracted data-center backlog, and durable enterprise-ALM and government legs are all genuine positives. But the bull case is now consensus, two cyclical dependencies (memory pricing, hyperscale demand) underpin the raised guide, and the equity discounts the growth as delivered. Operationally strengthening; risk/reward tightening.

9. Risk Analysis (§7.8)

Overall risk read: the dominant risks are valuation, rate/leverage, and external-financing dependence — a materially higher-risk profile than a net-cash compounder. The records annuity makes a permanent impairment unlikely, but the equity carries full REIT rate-beta, a data-center multiple, and a debt-funded build near an all-time high. The realistic bad outcome is a sharp de-rating on a rate or thematic shock, not a business collapse.

10. Valuation Discussion (§7.9)

Frame it as a REIT sum-of-the-parts — GAAP P/E (~130x+) and P/B (negative equity) are meaningless. At ~$119.48 (295.8M shares, market cap ~$35.3B; net debt + minority interest ~$19.2B → EV ~$54B), Iron Mountain trades at ~21.9x EV/EBITDA, ~23x trailing / ~20.7x forward AFFO ($5.17 → ~$5.79), and the 93rd percentile of its own price-to-sales history — near an all-time high. (Note: a stale third-party feed showed ~17.8x EV/EBITDA using a ~$84 market cap; the current-price figure is ~21.9x.)

Comps — a blended premium.

Iron Mountain sits above storage REITs and below pure data-center — a blended premium that credits the ~15%-of-EBITDA data-center arm across the whole enterprise.

Sum-of-the-parts (segment Adjusted EBITDA and multiples = ASSUMPTION; trailing):

The trailing SOTP (~$67–97/share) sits below the ~$119 quote. The ~$6–15B gap between the market’s ~$54B EV and the trailing SOTP is the market (a) applying a growth multiple to the whole (records pricing power + Matterhorn cross-sell) and (b) paying for the data-center development-pipeline optionality — the 852 MW of leased/under-construction capacity not yet in trailing EBITDA. Only the bull case, with a ~$3–5B pipeline-NAV credit, reaches the current price.

Embedded expectations. To justify ~$54B EV / ~$2.5B Adjusted EBITDA (~21.7x trailing) at a fair ~15x forward, Adjusted EBITDA must reach ~$3.6B (~13% CAGR over three years) — i.e., the market is pricing Matterhorn’s ~10%+ plan plus the data-center lease-up as delivered. On AFFO, holding a ~24x multiple requires ~10% compounding ($5.17 → ~$6.6 by 2028) even to fade to ~18x forward.

Scenarios (AFFO/share + exit multiple; analytical ranges only, NO price target):

• Bear: records pricing ceiling + data-center lease-up disappoints + ALM component-cycle reverses; AFFO CAGR ~4% to ~$5.6 (2028); multiple de-rates to storage-REIT ~16–17x → material downside (~$90–95).
• Base: growth continues but decelerates; AFFO CAGR ~8% to ~$6.25 (2028); multiple fades from ~24x toward ~20x → roughly flat-to-modestly-below spot (~$115–125).
• Bull: data-center surge + Matterhorn deliver + digital scales; AFFO CAGR ~11% to ~$6.8 (2028); sustains ~24x plus a pipeline-NAV credit → upside (~$150+).

Verdict (§7.9): priced as a data-center REIT on a records-annuity base — richest in its own history, near record highs, with the growth already in the multiple. The trailing sum-of-the-parts sits below the quote; the premium is the market crediting the multi-year plan and the development pipeline in full. The valuation requires the growth to be delivered and rates to behave — little margin of safety.

11. Variant Perception (§7.10)

Consensus. The sell-side is bullish and rising (targets in the ~$138–143 range), crediting an accelerating growth story: 37 years of storage growth, a contracted data-center backlog, an ALM surge, de-levering to 4.9x, and Matterhorn “delivered.” The factor tape shows a crowded thematic/alpha trade near record highs — strong relative strength (RS +46% YTD), positive alpha, positive DividendYield loading, negative Value loading (expensive), and a large negative interest-rate beta — but, tellingly, no systematic momentum or tech factor loading: the outperformance is idiosyncratic AI-data-center alpha, not a mechanical momentum exposure. The closest common-stock peer is Digital Realty.

Strongest bull case. Iron Mountain is a rare compounder that pairs a genuinely wide-moat, price-inelastic records annuity (37 straight years of growth, ~98% retention, ~2% maintenance capex) with two secular growth engines — data centers (contracted ~$3.3B backlog, 97% leased, >100 MW/year of leasing, AI-driven demand) and ALM/digital — funded by that annuity’s cash. Q1-2026 organic growth of +17% is the fastest in 25 years; the business is de-levering on rising EBITDA; ROIC sits in the comp plan as a guardrail; and the digital/government leg (FedRAMP High, the $714M IRS contract) is optionality the market barely credits. Deliver ~11% AFFO growth and the ~24x multiple is defensible.

Strongest bear case. The market has re-rated a ~15%-data-center enterprise to a data-center multiple, near an all-time high, on a levered balance sheet that does not self-fund — operating cash flow doesn’t cover capex, and 100% of the dividend and build is debt-financed at a rising ~6.25% marginal cost against ~2.9x coverage. The records moat is real but wraps a flat pond (volume declining; growth is a pricing lever with a ceiling), consolidated ROIC (~5.9%) is below WACC, AFFO flatters by excluding the ~$2B build, and the two exciting growth legs are the most cyclical/commodity (data-center yields the capital cycle will compress; ALM riding a memory-price cycle). The trailing sum-of-the-parts (~$67–97) is below the quote — the premium is pipeline optionality and delivered-growth assumptions. A rate back-up or thematic rotation de-rates it hard.

The 3–5 assumptions that matter most:

1. Does the data-center build earn attractive blended returns across a full cycle, or do yields compress as supply floods the segment?
2. Does the enterprise turn self-funding (retained cash flow after build + dividend durably positive) as data centers stabilize?
3. Does the records annuity’s price-led growth have a ceiling as volume declines?
4. Do rates behave against ~2.9x coverage and a debt-funded model?
5. Is the ALM surge durable (enterprise ITAD) or a memory-price cycle that reverses?

What would falsify each side. Bull falsified: data-center lease-up/renewal spreads fade, or the enterprise keeps out-borrowing its cash flow with no self-funding inflection — the growth is real but value-neutral because it is bought with ever more debt. Bear falsified: retained cash flow inflects durably positive as the data-center book stabilizes and the enterprise de-levers on rising EBITDA without new debt — proving the build creates value, not just revenue.

Net variant view. Consensus is right that the business is excellent and accelerating; the variant point is that the equity now prices a data-center growth story that is ~15% of the enterprise, funded with debt, near an all-time high — with the durable value (the records moat) in the segment the market is no longer paying for. The disagreement is about what you pay, not whether Iron Mountain is a good company.

12. Fact vs. Interpretation

13. Open Questions

1. Self-funding inflection — when (if ever) does retained cash flow after the build and dividend turn durably positive?
2. Data-center blended returns — what stabilized yield does the 852 MW pipeline actually earn across a cycle, net of rising funding cost?
3. Records pricing ceiling — how much longer can ~mid-single-digit price-led storage growth run on declining volume?
4. ALM durability — how much of the +63%/+92% surge is durable enterprise ITAD vs. a memory/HDD price cycle?
5. Digital solutions scale — does DXP/InSight + the IRS contract become a material recurring-software leg, or stay a rounding error?
6. Leverage trajectory — does 4.9x actually fall as the build continues, or does the debt keep growing with capex?
7. Add-back normalization — with Matterhorn “complete,” does the ~$196M restructuring add-back genuinely disappear from Adjusted EBITDA?

14. What Must Be True (§14)

Bull case — what must be true:

1. The data-center build earns attractive blended returns across a full cycle (lease-up and renewal spreads hold as supply grows), and the 852 MW pipeline converts to high-margin EBITDA.
2. The enterprise turns self-funding — retained cash flow after build and dividend goes durably positive — and de-levers on rising EBITDA without ever-more debt.
3. The records annuity keeps compounding ~mid-single-digit on price, and digital solutions scales into a real recurring-software leg.
4. Rates behave, and the market sustains a ~20–24x AFFO multiple.

Falsification test: durable positive retained cash flow with de-levering without new debt confirms the build creates value; continued out-borrowing of cash flow with no self-funding inflection falsifies it.

Bear case — what must be true:

1. Data-center yields compress as record supply floods the segment, and/or ALM’s component-price surge reverses.
2. The enterprise keeps funding its dividend and build with rising-cost debt against ~2.9x coverage, and rates back up.
3. The market re-rates a ~15%-data-center, levered REIT from a data-center multiple back toward a storage/blended multiple (~16–18x AFFO).

Falsification test: a rate/thematic shock de-rating the stock toward the trailing SOTP (~$67–97) confirms the bear; a durable self-funding inflection with sustained double-digit AFFO growth falsifies it.

Synthesis. Both cases agree the records annuity is a genuine wide-moat cash machine and that the growth is real; they disagree on whether the debt-funded data-center build creates value and whether the price already assumes it does. Because the trailing sum-of-the-parts sits below the quote and the enterprise does not self-fund, the asymmetry favors patience — own the annuity, but demand the data-center optionality at a discount, not at an all-time-high premium. The realistic bad outcome is a rate/thematic de-rating (toward the ~$90s), not a business failure; the realistic good outcome requires both delivery and a sustained premium multiple.

15. Source Appendix

(Primary sources below.)

• Iron Mountain Incorporated FY2025 Form 10-K (filed 2026-02-12, year ended 2025-12-31) — segment note (Global RIM / Global Data Center), storage-rental vs. service revenue, data-center MW/leasing/backlog, capex split, leverage, AFFO/FFO reconciliation, Project Matterhorn.
• Iron Mountain FY2024 / FY2023 Form 10-K — multi-year revenue/EBITDA/AFFO trend, segment recast, ALM build.
• Q1-2026 / recent 10-Qs — organic growth (+17%), data-center leasing, ALM, guidance, leverage.
• Iron Mountain earnings-call transcripts — Q1-2026, Q4-2025, Q3-2025 (public earnings calls) — AFFO guidance, data-center leasing/MW, storage organic, ALM, Matterhorn targets, leverage.
• DEF 14A proxy — incentive metrics (STI: AFFO/share; LTI: ROIC + TSR), ownership, anti-hedge policy.
• Form 8-Ks — earnings releases, debt offerings (€1.2B 4.75% 2034; $1.5B 6.25% 2035), dividend increases, data-center deals.
• ROIC.ai — income statement, cash flow, enterprise value, valuation multiples (IRM, and comps DLR/PSA/EXR); reconciled to filings.
• AZI — five-year adjusted price CSV (event map), valuation-index own-history percentiles (P/S 93rd, composite 87th), news feed.
• FactorsToday — factor loadings (Real Estate, DividendYield, negative Value, large negative InterestRate beta; no momentum factor), leaderboard (beta 0.99, y3 Sharpe 0.99), related-stocks (DLR + REIT ETFs).
• Business-quality / capital-cycle frameworks — Greenwald & Kahn, Competition Demystified; Marathon, Capital Returns (investment-research-frameworks skill).
• Third-party press: data-center leasing signings; debt offerings; dividend increases; analyst actions.

Facts are cited to primary filings where possible; interpretations and assumptions are labeled as such throughout. Management commentary is treated as hypothesis and validated against filings, financials, and external data.

APPENDIX A — Standard Diligence Questionnaire — Iron Mountain Incorporated (NYSE: IRM)

Supplemental to the analysis. Fact / Interpretation / Assumption labeled where it matters.

General

What thoughtful questions have other investors asked? Is Iron Mountain a data-center REIT or a records-storage REIT, and which multiple should it carry? Does the enterprise self-fund, or is the dividend + data-center build entirely debt-financed? How much of AFFO growth is real cash vs. a growth-capex add-back? Is the records business in secular decline (volume) or a durable price-led annuity? Is the data-center arm a subscale late entrant into a flooding capital cycle? How cyclical is the ALM surge (memory/HDD pricing)? Is the all-time-high multiple already pricing the AI/data-center optionality?

Cyclicality & Earnings Nature

Cyclical high or low? Interpretation: mixed. The records annuity is non-cyclical (contractual, price-led). Data centers are secular-growth but at a cyclical valuation peak. ALM is cyclically high (memory/HDD price cycle + hyperscale refresh). Blended, earnings are healthy and the multiple is near a cyclical high.

External or internal? Storage growth is internal (pricing/revenue management); the re-rating and data-center/ALM surge are partly external (AI demand, component prices, rate environment).

How stable are revenues? The ~59% storage-rental base is very stable; service revenue (ALM, projects) is more variable.

Outlook for products/services? Records: durable but flat-volume, price-led. Data center: strong secular demand, contracted backlog. ALM: growing but cyclical. Digital: early optionality.

How big is the market? Records storage is mature/consolidated (IRM #1). Data centers are a massive, fast-growing but capital-flooded market. ALM/ITAD is large and fragmented.

Business Quality & Competitive Moat

More or less competitive? Records: stable oligopoly (less competitive, favorable). Data centers: intensely competitive and capital-flooded. ALM: fragmented/competitive.

How profitable (ROIC/ROE)? GAAP ROE is meaningless (negative equity). Consolidated ROIC ~5.9% — below WACC — the records annuity earns high returns on tiny maintenance capex, but the debt-funded growth build dilutes blended returns.

How profitable is the industry / barriers? Records: high-margin, high-barrier (scale + switching costs). Data centers: high-margin but low-barrier for a subscale entrant. ALM: low-margin, low-barrier.

Easily understood? Moderately — the records annuity is simple; the consolidated REIT (AFFO add-backs, growth-vs-maintenance capex, lease-adjusted leverage, negative equity) requires careful adjustment.

Undermined by foreign low-cost labor? No — it is a real-estate/logistics/service business; the relevant disruption is digitization of paper (slow) and data-center commoditization.

Do brands matter? Modestly — Iron Mountain’s brand signals security/compliance/trust (valuable for regulated records and government), less so in commodity data-center leasing.

Switching costs? Very high in records (the “box that never leaves”); moderate in data centers (long leases, but re-tenanting possible); low in ALM.

Financial Condition & Balance Sheet

Assets not on the balance sheet? The records customer relationships/annuity and the data-center land/power bank are worth more than book (book equity is negative).

Off-balance-sheet liabilities? Operating leases (many warehouses are leased, hence “lease-adjusted” leverage); data-center power commitments; earn-outs on tuck-ins.

How conservative is the accounting? Aggressive-leaning on adjusted metrics — recurring ~$196M “restructuring/transformation” add-back to Adjusted EBITDA, and AFFO adds back the entire ~$2B growth build. GAAP is clean but uninformative (negative equity).

How capex-hungry? Extremely on a growth basis (~$2.1B/year, 84% data centers), though maintenance capex is only ~2% of revenue. This is the crux — the enterprise is capital-hungry and debt-funded.

Capital Allocation & Management

How much FCF, how used? Negative after total capex and dividend — the ~$1.8B FY25 gap was debt-funded. Priorities: data-center build > dividend > tuck-in M&A.

Significant acquisitions? Tuck-ins only (ITRenew 2022, Regency 2023, ALM 2025 deals, Web Werks India full ownership); no transformational M&A.

Buying back shares? No — and no meaningful ATM issuance either (shares +0.8%). Funds growth with debt, not equity — atypical for a REIT.

Issuing stock to insiders? SBC ~$140M (~2% of revenue, growing), added back in AFFO and Adjusted EBITDA.

Compensation policy? Well-aligned — STI on Revenue/Adjusted EBITDA/AFFO-per-share/strategic; LTI on ROIC/Revenue/absolute+relative TSR. ROIC-in-LTI is a genuine guardrail.

Motivations of management? CEO William Meaney (since 2013), CFO Barry Hytinen — stable, no founder control; anti-hedge/pledge policy. Executing an aggressive but disciplined growth plan.

Valuation & Market Data

ADR, MLP, or K-1? No — a U.S. REIT common stock (NYSE: IRM); distributions are 1099-DIV (REIT ordinary/capital-gain/return-of-capital components).

Dividend policy? ~$3.11/share (2025), grown ~10%/year; ~60% AFFO payout (85% of normalized FFO); ~2.8% yield.

How profitable? Very at the segment level (RIM 44.7%, DC 51.8% EBITDA margins); consolidated returns diluted by leverage and the growth build (ROIC ~5.9%).

Net income vs. cash flow? GAAP net income tiny ($145M); AFFO large ($1.54B) but flattered by capex add-backs; true post-capex-and-dividend cash flow is negative (debt-funded).

Risks & Downside

What would cause the stock to decline? A rate back-up (large negative rate beta), a data-center-multiple de-rating, an ALM component-price reversal, or evidence the enterprise cannot self-fund.

Catastrophic loss risk? Low but non-zero — historical warehouse fires; data-security/cyber on the digital side. Insured/managed.

Total loss? Very unlikely — the records annuity underpins value; the risk is a de-rating, not a wipeout.

Recent News & Events

Has the environment changed? Positively operationally (Q1-26 organic +17%, data-center backlog, de-levering, Matterhorn charges ended) but the multiple is near an all-time high.

Significant acquisitions? ALM tuck-ins (2025); full ownership of Web Werks India.

Change in accounting policies? Segment reporting simplified to two segments (RIM + Data Center; ALM into Corporate & Other). Matterhorn add-backs ending in 2025.

Recent changes? FedRAMP High (Q1-26); expanded U.S. Treasury/IRS digitization contract (up to $714M); $1.5B 6.25% 2035 notes (June-2026); four consecutive ~10% dividend hikes.

APPENDIX B — Source Appendix — Iron Mountain Incorporated (NYSE: IRM)

Primary sources prioritized. Facts cited to filings where possible; interpretations/assumptions labeled in the memo. Access date: 2026-07-10.

Primary — SEC Filings (Iron Mountain Incorporated, CIK 0001020569)

• FY2025 Form 10-K (filed 2026-02-12; year ended 2025-12-31) — segment note (Global RIM / Global Data Center; ALM in Corporate & Other), storage-rental vs. service revenue, cubic-feet/customer/retention disclosure, data-center MW/leasing/backlog/WALE, growth-vs-maintenance capex split, debt schedule and lease-adjusted leverage, AFFO/FFO reconciliation, Project Matterhorn. Local: output/IRM/sources/10-K/2026-02-12_irm-20251231.htm.
• FY2024 / FY2023 Form 10-K — multi-year revenue/Adjusted-EBITDA/AFFO trend, segment recast, ALM build (ITRenew, Regency). .../2025-02-14_irm-20241231.htm, .../2024-02-22_irm-20231231.htm.
• Q1-2026 / recent 10-Qs — +17% organic, data-center leasing (MW), ALM growth, raised guidance, leverage. output/IRM/sources/10-Q/.
• DEF 14A proxy — incentive metrics (STI: Revenue/Adjusted EBITDA/AFFO-per-share/strategic; LTI: ROIC/Revenue/TSR), ownership, anti-hedge/pledge policy. output/IRM/sources/DEF_14A/.
• Form 8-Ks — quarterly earnings releases; debt offerings (€1.2B 4.75% 2034 notes; upsized $1.5B 6.25% 2035 notes, June-2026); dividend increases; data-center and ALM deals. output/IRM/sources/8-K/.

Primary — Earnings-Call Transcripts (via ROIC.ai + saved corpus)

• Q1-2026 / Q4-2025 / Q3-2025 — AFFO/share guidance ($5.79–5.86 FY26), storage organic (+6% Q1-26, 37th year), data-center leasing (13→43→32 MW; >100 MW FY26 guide; >$3.3B backlog), ALM (+92% Q1-26, $950M guide, memory-price caveat), Matterhorn long-term targets, leverage to 4.9x, retained-cash-flow framing.
• (CEO William Meaney; CFO Barry Hytinen.)

Quantitative Data Sources

• ROIC.ai — income statement, cash flow, balance sheet, enterprise value, valuation multiples (IRM, and comps DLR/PSA/EXR); reconciled to filings (filings primary). Note: ROIC’s stated year-end EV used a stale market cap; current EV recomputed at spot.
• AZI — five-year adjusted price CSV (event map); valuation-index own-history percentiles (P/S 4.91x/93rd pctile, composite 87th; P/E meaningless; P/B null on negative equity); news feed. CSV local: output/IRM/2026-07-10/_scratch/IRM_price.csv.
• FactorsToday — factor loadings (Real Estate sector, DividendYield positive, Value negative, large negative InterestRate beta, no momentum factor), leaderboard (beta 0.99, y3 return +31.7%/Sharpe 0.99, y5 +28%/Sharpe 0.87, maxDD −39%), stock-info (RS +46% YTD), related-stocks (DLR + REIT ETFs), specific-vol.
• EDGAR / edgar.sh — corpus enumeration; EV recomputation (295.8M shares × $119.48 + net debt $18.89B + MI $0.34B ≈ $54B).

Secondary — Press & Third-Party

• Data-center leasing signings (N. Virginia, Chicago, Miami, Amsterdam) — company releases / trade press, 2025–2026.
• Debt offerings ($1.5B 6.25% 2035 notes, June-2026) — 8-K / press.
• U.S. Treasury/IRS digitization contract expansion (up to $714M) — company release, Sep-2025.
• Analyst actions — 2026 (targets not reproduced in the memo body per no-price-target rule).

Analytical Frameworks

• Greenwald & Kahn, Competition Demystified — moat taxonomy (customer captivity/switching costs + scale in records), market-share/ROIC tests, EPV vs. growth value.
• Marathon Asset Management (Edward Chancellor, ed.), Capital Returns — supply-side capital-cycle analysis (declining-supply records annuity vs. hot-supply data-center build), asset-growth anomaly.
• (via the repository’s investment-research-frameworks skill.)