Equinix: Where should the next capacity dollar go?
Download case PDF (3 pages)Request teacher’s pack
Decision date: 8 May 2024
Equinix has to serve two related but different markets. Its IBX data centers sell retail colocation and interconnection to a broad customer ecosystem. Its xScale facilities are designed for core deployments by a targeted group of hyperscale customers. A hyperscaler can use both, but the company’s disclosures do not show how much value one product creates for the other.
At the end of 2023, Equinix reported approximately 79% billed cabinet utilization across its consolidated IBX portfolio, compared with 82% at the end of 2022. The figure is global, considers power limits in its definition, and is not a measure of saleable capacity in any particular market. The utilization table includes KL1 and SL4, opened in January 2024, and excludes 18 unconsolidated data centers: 17 xScale centers and MC1 IBX. Equinix also warned that power may constrain growth even when physical cabinet space is available.¹
The Q1 2024 release presented a strong xScale pipeline. Equinix reported nearly 350 MW of cumulative xScale leasing, including 48 MW of new pre-leasing since its prior earnings call. Management said nearly 90% of operational and under-construction xScale capacity was leased, and attributed demand to cloud and AI activity. The company did not disclose site-level contract terms, tenant concentration, cancellation rights, lease commencement or project returns.²
Equinix reported multiple expansion paths underway: 14 xScale builds, more than 16,000 retail cabinets and over 50 MW of xScale capacity expected through the end of 2024. It also reported more than 468,000 platform interconnections, 6,200 Q1 interconnection additions, and over $1.3 billion of annualized retail revenue from hyperscalers across more than 60 IBX locations. These figures use different measures and do not establish the marginal return on either product.²
Capital structures also differ. In April, Equinix announced a $600 million U.S. xScale joint venture with PGIM for the planned SV12x facility. The project was to be built in two phases, with more than 28 MW of expected full-build power capacity; PGIM would own 80% and Equinix 20%. The first phase was scheduled for Q2 2024. The announcement projected the global xScale portfolio at more than $8 billion, 35-plus facilities and more than 725 MW when fully built. These were plans, not operating results or Equinix-only investment amounts.³
For FY2024, Equinix forecast total capex of $2.78–3.03 billion, including $2.57–2.80 billion of non-recurring capex and $210–230 million of recurring capex. It expected $12–52 million of xScale-related on-balance-sheet capex to be reimbursed by current or future joint ventures. Management also said more than 90% of current retail expansion capex related to owned land or buildings with long-term ground leases. These figures do not reveal a freely transferable capital pool or the full economic cost of xScale.²
Your decision
You are advising Equinix’s leadership as it considers the next increment of capacity development. Recommend how the company should pace its xScale JV pipeline relative to IBX retail, interconnection and existing-site power work. The public information does not establish that these alternatives draw on interchangeable capital, sites, power or teams.
Consider three paths:
- Maintain or accelerate xScale development to pursue reported hyperscale leasing and preserve market positions.
- Prioritize IBX, interconnection and existing-site power investments where customer demand may be constrained by usable capacity.
- Stage development across both using market-specific leasing, power and construction evidence before advancing each phase.
For your recommendation, state what evidence would change your view. Consider lease quality and timing, tenant concentration, local power readiness, cost to complete, construction milestones, JV funding and governance, and linked IBX demand. Do not infer a project return from portfolio-wide figures.
Research lens
Two pre-cutoff studies can help structure the decision, but neither supplies an Equinix forecast. Pindyck’s theoretical work explains why uncertainty and irreversible capacity can make waiting valuable; only indexed excerpts were accessible for this case preparation, so the cited mechanism is used without equations or calibration values.⁴ Mytton et al. discuss grid congestion, capacity reservations and limits on workload flexibility in data centers, drawing mainly on UK policy and case evidence; this does not establish power constraints at Equinix sites.⁵
Discussion questions
- What do the xScale leasing figures establish, and what do they leave unknown about future cash flows and delivery risk?
- Why can global IBX cabinet utilization not answer whether a particular market needs more power-ready capacity?
- How might IBX and xScale complement one another? What evidence would distinguish a real cross-product effect from separate demand?
- Which commitments are reversible, phaseable or dependent on joint venture and utility terms? What should management verify before advancing a phase?
- Which path would you choose now, and what specific evidence would cause you to change course?
Selected sources
- Equinix, FY2023 Form 10-K, filed 16 Feb 2024, printed pp. 21–22 and 47, 53; SEC filing.
- Equinix, Q1 2024 results release, 8 May 2024, official IR HTML lines 177–186 and 206–207; release. Related 8-K exhibit PDF locators: lines 121–150 and 199–203.
- Equinix and PGIM Real Estate, U.S. xScale JV announcement, 15 Apr 2024, lines 128–139; release.
- Robert S. Pindyck (1988), “Irreversible Investment, Capacity Choice, and the Value of the Firm,” American Economic Review 78(5):969–985, abstract, Table 1 (published p. 980), §§III–IV; MIT-hosted article PDF. Page numbers for the indexed section excerpts could not be verified.
- David Mytton et al. (2023), “Stretched grid? Managing data center energy demand and grid capacity,” Oxford Open Energy 2, oiad014, lines 181–216; full article.