Iron Mountain: How much data-center capacity should it commit next?
Download case PDF (4 pages)Request teacher’s pack
Decision date: May 2, 2024
Information cutoff: May 2, 2024
Iron Mountain Incorporated (IRM) built its business around storing and managing physical and digital information. Its global data-center business had become an increasingly visible growth platform. On May 2, 2024, after reporting first-quarter results, IRM’s leadership faced a capital-allocation question: continue the planned pace of data-center expansion, put additional commitments behind project gates, or slow new commitments while completing construction already underway.
This was not a decision about whether customers used data centers. It was a decision about how quickly to commit capital to long-lived, specialized capacity when projects differed in construction stage, customer commitment and readiness. Existing operating sites, signed leases and projects under construction had to be considered separately from future capacity held for development. The company’s public disclosures did not provide project-level returns or enough information to calculate the economics of the next site.
A growing platform with several stages of capacity
IRM reported signing 30 megawatts (MW) of new and expansion data-center leases during the first quarter of 2024. At March 31, its operating portfolio had 261.4 MW of leaseable capacity and was 94.0% leased. The company also reported 233.7 MW under construction and 365.7 MW held for development, for 860.8 MW of total potential capacity. IRM defines leaseable MW as critical power available for customer use; total potential MW includes existing and planned capacity at full build-out. Thus, the three categories describe materially different stages, not capacity that was all operating, powered, funded or contracted.[1]
The supplemental development table reported 233.7 MW under construction, 232.8 MW pre-leased and 365.7 MW held for development. IRM reported that the listed development activity was 99.6% pre-leased. The metric applies to the table’s development activity; it does not mean that 99.6% of all 860.8 MW of potential capacity was leased. Pre-leased means a lease was signed before construction was complete.[1]
The same development table showed $175.6 million invested during the quarter, $550.6 million invested cumulatively through March 31, and $1,359.3 million of total expected investment across the listed activity. Those figures do not disclose remaining spend, contingency or expected return by project. Nor do they establish how much capital IRM would need to commit to the 365.7 MW held for development.[1]
Capital, funding and execution
The company’s first-quarter supplemental reported $12.612 billion of net debt and net lease-adjusted leverage of 5.1 times. It also reported adjusted EBITDA to interest expense of 3.2 times. These are company-defined measures; the leverage calculation includes lease capitalization and adjusted EBITDAR. They provide context for capital allocation but do not establish IRM’s marginal borrowing capacity or whether any individual data-center project creates value.[1]
IRM’s 2023 Form 10-K gave full-company 2024 capital-expenditure guidance of $1.35 billion for growth and $150 million for recurring investment, excluding spend associated with potential future acquisitions. In the first quarter, the company reported $365.9 million of total capex: $337.1 million growth and $28.7 million recurring. Of the growth total, $276.6 million was data-center growth capex. Annual guidance was company-wide, and the company’s growth category also covered real estate, innovation and other activity.[2]
The 10-K described data centers as capital intensive and identified exposure to capital requirements, inflation and high interest rates, supply-chain disruption, changes in customer requirements, construction cost and timing, power cost, and service interruptions. These disclosures identify risks, but do not quantify their likelihood or their effect on any particular project.[2]
Power and delivery readiness mattered alongside customer demand. A site might appear attractive but still depend on utility connections, permits, construction sequencing, equipment availability and customer-specific requirements. IRM’s public materials did not provide project-by-project power milestones, remaining construction schedules or customer credit and cancellation terms. A general study of UK electricity networks described how data-center concentration could strain grid capacity and delay connections; it was a system-level UK analysis, not evidence about IRM’s individual sites or U.S. power access.[3]
The choice before management
The alternatives below are decision paths for analysis, not options IRM disclosed to shareholders. Each would need to distinguish finishing existing signed and under-construction projects from committing to new land, starts or capacity increments.
- Continue the planned pace. Maintain the expansion path to preserve customer access and delivery credibility. IRM signed 30 MW of leases in Q1, and almost all capacity in the listed development activity was reported pre-leased. Long construction lead times may make delay costly if customers need capacity on firm schedules or suitable power-ready sites are scarce. The public information does not establish the delivery schedule, margin, customer quality or returns attached to those leases.
- Gate or stage incremental commitments. Complete work already underway while linking later land and construction commitments to evidence such as signed customer obligations, credit quality, utility milestones, construction readiness and project-level returns. Staging could preserve choices where projects are separable, but the company did not disclose the cost of pausing, the options available at each site or whether waiting could cause IRM to lose customers, sites or power access.
- Slow new commitments while completing current builds. Prioritize execution and stabilization of operating and under-construction capacity before adding discretionary projects. This could preserve capital flexibility if marginal economics or readiness are weak. It could also sacrifice growth opportunities if demand remains strong and competitors secure customers or scarce sites first. The reported leverage ratio alone does not settle the trade-off.
The decision also has two dimensions: when to invest and how much capacity to build. A theoretical capacity-choice model shows why uncertain demand and irreversible investment can make the ability to defer a commitment valuable. A separate model finds that, under its assumptions, greater uncertainty can delay investment while increasing the preferred capacity conditional on investing. These results do not estimate IRM’s investment threshold, future demand or optimal MW.[4]
Empirical work on manufacturing firms finds that uncertainty can make investment less responsive to demand shocks when investment is partly irreversible. Its estimate is specific to that sample and setting, not a parameter for a data-center REIT. Research on data-center grid constraints likewise identifies a possible development bottleneck but does not establish IRM’s site readiness.[5]
Questions for discussion
- Which path would you recommend for IRM’s next data-center commitments, and what facts drive your choice?
- Which capacity is already operating, under construction, pre-leased or merely held for development? How should those stages affect the decision?
- What does the 99.6% pre-leased figure establish, and what does it leave unanswered?
- How should leverage, funding costs and the company-wide capex plan enter the decision without being mistaken for project-level economics?
- What customer, construction, power and financial evidence would you require before the next commitment? Which evidence would cause you to accelerate, stage or slow?
- How would you distinguish the value of waiting from the value of learning? What information could actually arrive before an irreversible commitment?
Selected sources
- Iron Mountain, Q1 2024 Supplemental Financial Information, published May 2, 2024, pp. 3, 17–20, 23–24: Q1 release; Data Center Customer Lease Expiration and Leasing Activity Summary; Data Center Operating Portfolio and Total Potential Capacity; Data Center Expansion and Development Activity; Capitalization and Debt Maturity Profile; Data Center Business Definitions. SEC-filed supplement.
- Iron Mountain, 2023 Form 10-K, filed Feb. 22, 2024, Part I, Item 1A, data-center risk factors (printed p. 13); Item 1, “Global Data Center” (printed p. 36); Item 7, “Capital Expenditures” (printed pp. 49–50). SEC filing; Q1 2024 Supplemental Financial Information, p. 21, “Quarterly Capital Expenditures and Investments,” link above.
- David Mytton, Masaō Ashtine, Scot Wheeler and David Wallom, “Stretched grid? Managing data center energy demand and grid capacity,” Oxford Open Energy 2 (2023), oiad014, abstract, introduction and grid-capacity discussion. Article.
- Robert S. Pindyck, “Irreversible Investment, Capacity Choice, and the Value of the Firm,” American Economic Review 78(5) (1988), pp. 969–985; Thomas Dangl, “Investment and Capacity Choice under Uncertain Demand,” European Journal of Operational Research 117(3) (1999), pp. 415–428. Dangl was reviewed through the publisher abstract and accessible preview only; the full text was not accessed. Pindyck full paper; Dangl publisher record.
- Nick Bloom, Stephen Bond and John Van Reenen, “Uncertainty and Investment Dynamics,” Review of Economic Studies 74(2) (2007), pp. 391–415, journal abstract/result. Journal record.