Digital Realty Trust — How Much Capacity Should DLR Commit?
Download case PDF (5 pages)Request teacher’s pack
Decision date and information cutoff: 24 September 2026, 07:30 ET
Decision scope: A hypothetical policy for sequencing future development commitments
Case convention: The committee assignment and alternatives are hypothetical. Company disclosures are historical; no private deliberations are represented.
The decision
Digital Realty Trust (DLR) has expanded its development program while demand for data-center capacity is growing and uncertain. A hypothetical investment committee must set the policy for the next wave of commitments: how much capacity should DLR advance toward full construction before customers have signed leases, and what evidence should be required at each step?
One approach is to continue moving quickly into full construction to capture demand, powered sites and customer relationships before competitors do. Another is to stage more commitments and require stronger lease, power-delivery and cost evidence before authorizing the largest, least reversible outlays. A project-specific policy could combine early investment to preserve scarce positions with later gates before full construction.
The committee is not being asked to rank individual sites. The public record does not disclose the project-level customer commitments, power schedules, construction costs or returns needed to do that. It must recommend a policy, identify the evidence that would trigger a faster or slower pace, and explain what would change its view.
A platform built around data-center capacity
DLR owns, operates and develops data centers and related connectivity infrastructure. Its customers use the facilities to house computing equipment and connect to networks and other organizations. The company serves both large deployments and smaller colocation and interconnection needs. DLR's 2025 Form 10-K describes deployments below 1 MW as typically ranging from one cabinet to 75 cabinets, with deployment possible in days and terms generally lasting two to five years. Deployments above 1 MW can scale much larger, are described as executable in weeks, and generally carry terms of five to ten or more years. These are company descriptions, not guaranteed service times. [S1]
The different offerings matter for capital allocation. A portfolio of small deployments and interconnections can support network density and customer relationships without necessarily requiring a new large building. Hyperscale capacity can require larger facilities and longer planning horizons. The case evidence does not map bookings to particular sites or specify which demand segments a future project would serve.
A growing construction program—and a larger future pipeline
At 31 March 2026, DLR reported 1,169 MW of projects underway, 61% pre-leased, and more than 5 GW of additional future development capacity. At 30 June, it reported 1,402 MW underway—about 82% above year-end 2025—and 54% pre-leased. The change from Q1 to Q2 is an increase of 233 MW underway and a decrease of seven percentage points in the pre-leased share. Those are calculations from company-reported aggregate figures; they do not establish why the pre-lease percentage changed or whether the newer projects caused it. [S2] [S3]
The Q2 filing separately reported more than 7 GW of additional future development capacity. Of that amount, DLR classified 70% at locations with more than 100 MW of buildable capacity, 20% at locations with 25–100 MW, and 10% at locations below 25 MW. The Q2 investor presentation showed approximately 9 GW of future development capacity under a different definition: it includes land, shell and data centers under construction, and its portfolio totals use 100% share. The approximately 9 GW measure therefore includes construction; it is not interchangeable with the filing's more than 7 GW of additional capacity. Neither figure establishes that every MW is powered, permitted, funded or ready to build. [S3] [S4]
One dated project example provides context, but not a universal rule. On 14 September 2026, DLR announced a joint venture with Rönesans Infrastructure to develop and operate data centers in Türkiye. The company said the first Ankara campus was designed for more than 22 MW of carrier-neutral IT capacity, with land, power and permitting secured and early construction commenced. The first facility was scheduled for completion in 2028. The announcement did not disclose project economics, customer commitments, ownership split or a company-wide approval policy. It shows one JV project with stated readiness conditions; it cannot establish that other DLR projects can be staged in the same way. [S5]
Demand signals do not all mean the same thing
DLR reported Q2 2026 bookings of $208.5 million in annualized GAAP base rent at DLR's share. Of this, $87.8 million came from 0–1 MW leases and $20.5 million from interconnection, together about 52% of the total. DLR also reported a $1.4 billion share of signed-but-not-commenced backlog; at 100% share, the reported amount was $1.9 billion. The weighted-average lag between Q2 leases signed and contractual commencement was nine months. Bookings and backlog are management-reported measures of signed leases, not current revenue, cash collection or margin. [S6]
After quarter-end, DLR announced two hyperscale leases with $410 million of annualized GAAP base rent gross, or $205 million at DLR's share. These signings strengthen the case that demand can arrive in large increments. The disclosure does not give their commencement timing, cancellation provisions or a construction-cost bridge. They also do not identify a particular project that the committee could approve. [S6]
At 30 June, consolidated portfolio occupancy was 89.8% across 2,515 MW of white-space IT load, excluding active or held-for-development capacity. Occupancy was 95.2% in the Americas, 81.9% in EMEA and 85.9% in APAC. This variation cautions against treating global demand or capacity scarcity as uniform. These are portfolio occupancy figures, not market vacancy rates or site-level measures of future demand. [S3]
The capital and execution exposure
DLR estimated an 11.5% pre-tax stabilized cash yield for the 1,402 MW underway. The estimate used anticipated net operating income based partly on signed leases and partly on market assumptions. DLR cautioned that completion, costs and achieved yields could differ materially. The figure is management's estimate for the aggregate underway portfolio; it is not a realized return, project-level return or forecast for the additional future pipeline. [S3]
DLR reported approximately $4.1 billion of open construction-contract commitments at 30 June, including about $320.1 million that was reimbursable. It expected $2.8–$3.3 billion of capital expenditures for the rest of 2026, including its share of joint-venture contributions and net of development-program partner contributions. These are distinct measures and should not be added as if they were a single remaining-cost estimate. Guidance could change with demand, leasing and debt or equity availability. [S3]
The company reported net debt to last-quarter annualized adjusted EBITDA of 4.7 times and approximately $6 billion of liquidity. Its stated average target for that leverage measure was 5.5 times. The target is an average, not a hard covenant or a disclosed development budget; the leverage and liquidity measures use company-defined bases. [S4]
Growth opportunity and bottlenecks
The International Energy Agency's April 2026 analysis projects global data-center electricity demand to rise from 485 TWh in 2025 to 950 TWh in 2030. It also describes uncertainty in AI uptake and efficiency, and constraints involving power infrastructure, chips, financing and local acceptance. These are industry scenarios, not a DLR demand forecast. Global electricity use cannot be converted directly into DLR's rentable IT capacity. [S7]
Competitors are investing as well. Equinix reported 52 projects underway across 33 markets, nine new projects since April, and higher guidance citing stronger demand, bookings and presales. Its product mix and reported measures differ from DLR's, so these figures are a signal of rival investment—not a like-for-like performance comparison. [S8]
DLR's 22 September announcement of a planned Los Angeles cable landing station with Assured Communications adds another dimension to the capacity proposition: connectivity can complement powered floor space. The project remained under development at the cutoff, with cable support expected in 2028. The release did not disclose project economics or expected revenue. [S9]
Evidence and unresolved questions
| Evidence available by the cutoff | What it indicates | What remains unknown |
|---|---|---|
| Underway capacity grew from 1,169 MW in Q1 to 1,402 MW in Q2; pre-leased share moved from 61% to 54%. [S2] [S3] | DLR was advancing a larger program while a smaller share of the aggregate underway capacity was pre-leased. | Which projects drove the change, customer terms, lease commencement, cancellation rights, site mix and power delivery. |
| More than 7 GW of additional future development capacity; a separate approximately 9 GW measure includes construction. [S3] [S4] | The potential development platform is large, with definitions that must be kept distinct. | Readiness, project sequence, cost, power and whether capacity can be delayed, resized or abandoned. |
| $208.5 million of Q2 bookings at DLR share, a $1.4 billion DLR-share backlog and two July hyperscale signings. [S6] | Demand includes both smaller/interconnection activity and large hyperscale leases. | Which sites and products the demand supports, margin, commencement risk and its fit with new construction. |
| Aggregate 11.5% estimated pre-tax stabilized cash yield; large open commitments and future capex guidance. [S3] | Management expects returns on the underway program while execution and funding exposure remain material. | Project-level cost-to-complete, realized returns, financing cost and yield for future projects. |
| IEA scenarios and Equinix's reported expansion. [S7] [S8] | Industry demand can grow substantially, while infrastructure constraints and competitors create both uncertainty and urgency. | DLR-specific site demand, competitor overlap, customer switching and market-level power availability. |
| One Ankara JV with land, power and permitting secured before early construction. [S5] | A project-specific example of stated readiness and partner structure. | Whether the same sequence applies elsewhere, what rights are separable and the project's economics. |
The choice
The committee can recommend faster full construction to capture demand and scarce powered capacity; a more staged policy that requires stronger lease, power and cost evidence before full construction; or a portfolio policy that makes early commitments selectively and sets project-specific gates for later spending. For any recommendation, specify what evidence should permit a project to advance, what would cause a pause or a change in pace, and how the company should weigh the cost of waiting against the cost of committing too early.
Do not assign a numeric pre-lease threshold or project return from the public information in this case. If you introduce a threshold for policy design, label it as your proposed governance assumption and explain how it would be tested. Distinguish company-reported facts and estimates from your inferences, and identify the strongest evidence that could overturn your recommendation.
Sources and conventions
All company information is limited to the 24 September 2026, 07:30 ET cutoff. This fixed pre-market boundary follows DLR's 22 September release and precedes the 24 September trading day; the 14 September Türkiye JV announcement is included. No later company outcome is used. DLR presentations, forecasts and estimates are attributed to management. MW and dollar figures retain the source's definitions and share basis.
[S1]: Digital Realty Trust, Inc. 2025 Form 10-K, filed 13 February 2026, Item 1, Business, pp. 4–5. [S2]: Digital Realty Trust, Inc. Q1 2026 Form 10-Q, filed 1 May 2026, MD&A, “Development Projects,” pp. 49–50. [S3]: Digital Realty Trust, Inc. Q2 2026 Form 10-Q, filed 31 July 2026, MD&A, “Revenue Base,” p. 47; “Future Uses of Cash,” pp. 56–57; “Development Projects” and “Capital Expenditures,” p. 57. [S4]: Digital Realty Q2 2026 earnings presentation, furnished 23 July 2026, slides 3–4 and 21–22, with notes on slide 32. [S5]: Digital Realty, “Digital Realty Enters Türkiye Through Formation of New Joint Venture,” 14 September 2026, paragraphs 1–5. [S6]: Digital Realty Q2 2026 earnings release, issued 23 July 2026, “Leasing Activity” and “Backlog.” [S7]: International Energy Agency, Key Questions on Energy and AI, 16 April 2026, executive summary and sections “AI and energy nexus,” “bottlenecks,” “updated demand outlook,” and “social concerns.” [S8]: Equinix Q2 2026 earnings release, issued 29 July 2026, highlights and outlook. [S9]: Digital Realty, “Digital Realty to Develop New Cable Landing Station in Los Angeles,” 22 September 2026, paragraphs 1–5.