Iron Mountain's survey says 86% of AI projects stalled; it signed 110 megawatts anyway

The company's own survey says 86% of AI projects stall on data debt, while its data-center arm books 110 megawatts of leases on borrowed money. On November 5 the market finds out which half of Iron Mountain the numbers pay.

Vincent JiangVincent Jiang · 3 min read
Share
William Meaney, chief executive of Iron Mountain, speaking at the Horasis China Meeting in Interlaken
1 / 7Slide 1 of 7
William Meaney, chief executive of Iron Mountain, at the Horasis China Meeting in Interlaken in November 2016.

Iron Mountain+1.47% — Iron Mountain, up 1.47 percent as of its last quote spent the weekend telling 240,000 customers that their AI projects are stuck, and its own stock is priced as if those projects will soon need somewhere to live 1. On October 19, at the Gartner+0.47% — Gartner, up 0.47 percent as of its last quote IT Symposium in Orlando, the records-storage REIT released The Information Supercycle, a survey finding that 86% of organizations had an AI project delayed or canceled over data concerns in the past 12 months 1.

The same ticker also holds 110 megawatts of new data-center signings from the first seven months of 2026, 75 of them in July alone 2. One company, two businesses, and the survey describes both.

The stall is the product. Four in ten organizations burn at least 41% of an AI project's timeline on manual data preparation, only 39% keep data "digitized and structured," 56% blame disconnected systems and silos, and 53% still point to legacy formats or physical archives, the boxes Iron Mountain has warehoused since 1951 1.

"AI readiness is about much more than deploying the latest technology," said Varun Gupta, the company's senior vice president of global product and solutions marketing 1. The cure has a seller, and the seller wrote the diagnosis.

Most enterprises cannot trace the data their AI models use

020406080Cannot track data lifecycle70Disconnected systems or silos56Legacy formats or physical archives53Spend 41%+ of timeline on prep40Data digitized and structured39
Data
Value
Cannot track data lifecycle70
Disconnected systems or silos56
Legacy formats or physical archives53
Spend 41%+ of timeline on prep40
Data digitized and structured39
Most enterprises cannot trace the data their AI models use1

The market pays for growth, not the cure

At $113.88 on October 20, Iron Mountain trades at 81.34 times earnings on a $33.9 billion market cap, with Jefferies−1.12% — Jefferies, down 1.12 percent as of its last quote cutting its target to $135 from $147 on October 12 3. The growth half is what carries that multiple: data-center, digital and asset-lifecycle businesses grew more than 50% in the second quarter, revenue hit $2.03 billion, and full-year guidance sits at $7.94–8.01 billion 4.

The growth half also borrows. Debt has climbed from $14.1 billion in late 2023 to $19.6 billion, capital spending has outrun operating cash flow in 11 of the last 12 quarters, and the June notes cost a fixed 6.25% to 2035 4.

Debt is up $5.5 billion in three years while capex outruns operating cash in 11 of 12 quarters

  • Total debt
  • Capital spending
  • Operating cash flow
$0B$5B$10B$15B$20BQ4 '23Q2 '24Q4 '24Q2 '25Q4 '25Q2 '26$19.62B
Data
Total debtCapital spendingOperating cash flow
Q3 '23$14.14B$0.35B$0.22B
Q4 '23$14.5B$0.37B$0.45B
Q1 '24$15.23B$0.38B$0.13B
Q2 '24$15.39B$0.4B$0.38B
Q3 '24$15.82B$0.4B$0.25B
Q4 '24$16.05B$0.61B$0.43B
Q1 '25$17.14B$0.68B$0.2B
Q2 '25$17.85B$0.55B$0.38B
Q3 '25$18.48B$0.52B$0.27B
Q4 '25$18.73B$0.52B$0.5B
Q1 '26$19.39B$0.51B$0.34B
Q2 '26$19.62B$0.58B$0.55B
Quarterly figures from Iron Mountain's SEC filings, retrieved 21 October 2026. Debt in USD billions at period end; capex and operating cash flow in USD billions per quarter.5

The survey cuts both ways, and Iron Mountain published it

The 86% stalled projects are, in the bulls' telling, future demand for the storage-and-governance half. They are also the tenants for the 110 megawatts. If four in ten companies cannot finish preparing data, the leases signed in July fill slowly, and the $19.6 billion of debt waits for revenue that arrives late.

The November 5 print sorts it

Consensus expects $1.99 billion of revenue against $1.75 billion a year earlier, and EPS of $0.6120 against $1.32, a comparison the debt service makes unforgiving 3. Watch the data-center leasing line, and whether 75 megawatts in a month repeats while 86% of the market reports its projects frozen 3.

The other side is real

The bulls have a fair case: enterprises that cannot deploy AI still must fix the data underneath it, and 95% call data a top priority 1. Deferred is not canceled. But a company cannot charge 81 times earnings for growth that its own research says is deferred, and collect a 6.25% coupon on the money building it, without the November numbers doing the talking 1.

Get Data Center by email, free

Who is building AI's data centers, and who is paying for them. In our daily email, every weekday.

Loading IRM…

Reader comments

0 comments

    Get the newsletter

    The day's AI and market news, in your inbox every weekday.