Large-block colocation in 2026 is a scale-and-delivery market: Equinix, Digital Realty, CyrusOne, QTS, Vantage, CoreSite, Switch, Aligned, Iron Mountain and DataBank are identified among the largest U.S. operators, while vacancy held at 1% for a third consecutive year [S2][S5].
Wholesale and hyperscale colocation usually means whole buildings or multi-megawatt phases rather than small retail racks, with operators trading unit pricing for scale, long leases and faster capacity absorption [S4][S6].
Large-block leasing definition and market scale
Large-block colocation is wholesale leasing at campus scale, generally covering whole buildings or multi-megawatt development phases. One 2026 capacity review reported that 7,481 MW of hyperscale colocation was under construction and preleased [S6]. This is a materially different procurement exercise from retail colocation, where smaller deployments can cost almost twice as much per megawatt as hyperscale blocks [S4].
The larger blocks are sold as contracts for guaranteed access to power, cooling, physical capacity and network connectivity. Hyperscale tenants can absorb an entire campus under leases extending beyond ten years, while wholesale operators accept lower per-megawatt pricing in exchange for long-term commitments and quicker absorption [S4].
Frontier markets accounted for 77% of all data center capacity under construction in North America [S2]. In the established North American hubs, CBRE tracked a 33% year-over-year inventory increase across Northern Virginia, Atlanta, Dallas-Fort Worth and Chicago in Q1 2026 [S1].
Operator comparison: scale, network density and delivery
Equinix leads the scale comparison through carrier-neutral interconnection and dense network ecosystems, while Digital Realty combines a wholesale footprint with enterprise and hyperscale capacity. CyrusOne focuses on wholesale and hyperscale powered-shell campuses, QTS develops large contiguous hyperscale campuses, and Vantage specializes in hyperscale and build-to-suit delivery [S5].
This ranking is an operator scale screen, not a verified table of leased MW by company. The public material identifies the leading names but does not publish a consistent operator-by-operator total for large-block leased capacity [S5][S7]. A defensible ranking requires four fields: commissioned MW, MW under construction, MW preleased and MW available for a specific market.
Why footprint rank does not equal the best lease
Historical scale provides a shortlist, not a procurement decision. Live IT capacity across the listed operators does not reveal which megawatts are committed, when they will be energized, or whether one tenant dominates the lease [S5]. Available capacity in North America was limited to small, fragmented blocks [S2].
Large footprints can also conceal fragmented ownership. A national operator may hold separate campuses without a contiguous 20 MW or 40 MW powered block available for one tenant. By contrast, a regional operator may control the exact land, utility allocation and campus configuration required. The relevant unit is therefore the deliverable block at one site, not the sum of a company’s national assets [S4].
Power infrastructure is the harder constraint. CBRE identified global power availability and grid limitations as factors extending development schedules in Northern Virginia, Chicago, London and Frankfurt. Local opposition also constrained new supply in established hubs [S1]. A nominal block should therefore be treated as unleaseable until utility delivery, cooling design and construction status are documented.
Lease structure and economics for multi-megawatt space
Wholesale operators sell larger power blocks on longer commitments, while retail operators divide a facility into smaller units and maximize revenue per kilowatt. Smaller deployments can carry nearly double the per-megawatt price of hyperscale capacity, but the hyperscale tenant exchanges price for scale and duration [S4].
The commercial comparison should include committed MW, lease tenor, rent commencement, expansion rights and power reimbursement. Electricity reimbursement can increase reported revenue without creating equivalent profit, so contracted revenue alone is not an adequate measure of lease quality [S4]. A long agreement can also increase concentration risk when a single tenant carries much of the project economics.
Preleasing reduces development exposure by allowing construction to begin with much of the future revenue committed. In constrained markets, customers may contract months or years before opening [S4]. That makes a prelease credible only if the operator controls the site and can demonstrate a realistic energization path.
Large-block demand from hyperscalers and AI
AI, neocloud and hyperscale demand is the principal force expanding available large blocks. CBRE attributed global Q1 2026 inventory growth largely to hyperscale and AI requirements, with Latin American inventory rising 41.3% year over year and North American inventory increasing 33% across four major markets [S1].
Hyperscale colocation removes the distinction between self-built cloud infrastructure and leased wholesale capacity for some tenants, but it does not remove the need for a power strategy.
Operators are consequently targeting markets where land and utility capacity can support large campuses. CBRE highlighted Querétaro, Johor, Batam, Tennessee and West Texas as emerging locations with scalable deployment conditions. Querétaro recorded 450.2% inventory growth and reached 298.2 MW, while São Paulo remained Latin America’s largest tracked market at 536.7 MW [S1].
Market fit and operator selection
Large-block colocation is for hyperscalers, AI operators and enterprises requiring multi-megawatt campuses, long terms and custom power envelopes. It is not the right procurement model for a workload needing a small rack, rapid monthly expansion or a retail metro interconnection footprint [S4]. The comparison between large-block and smaller colocation formats is covered in Hyperscale vs Colocation: 2026 Growth and Selection.
Equinix ranks first in the supplied 2026 provider comparison for global footprint, Digital Realty for hyperscale scale, and QTS is also placed among the leading operators. Another market ranking lists Equinix, Digital Realty, NTT, QTS, CoreSite, CyrusOne and Iron Mountain among the largest names [S3][S7]. These rankings converge on the same screening principle: use recognized operators to form the shortlist, then verify live MW at the required site.
For Qu Data Centres, the more specific availability case is 17 MW of live capacity across nine Tier III-grade facilities in five Canadian markets. The provider is 100% Canadian-owned, with SOC 1/2, ISO 27001, HIPAA and PCI certifications, plus carrier-neutral access to more than 15 networks [S3]. That does not place Qu ahead of the U.S. operators by total leased MW, but it demonstrates why jurisdiction, deployable capacity and certification can outweigh headline footprint.
Power economics and delivery constraints
CBRE recorded European rents of $235–$265 per kW per month in Frankfurt and $330–$475 per kW per month in Singapore. Northern Virginia vacancy was 0.3%, while Atlanta stood at 1% in Q1 2026 [S1]. These figures explain why an apparently straightforward lease can become expensive when the requested block is fragmented or requires power that is not yet energized.
Financing terms also differ by tenant quality. Top-tier credit hyperscalers achieved credit spreads in the low-200-basis-point range and leverage up to 85% loan-to-cost. Non-credit tenants were generally priced about 200–300 basis points wider than investment-grade construction loans, with leverage in the 70–80% loan-to-cost range [S2]. These are project-finance observations rather than direct colocation rent quotes, but they show how tenant credit affects the provider’s delivery economics.
Funding availability remained firm during the first half of 2026. SASB/CMBS and ABS data center volume reached $17 billion, a 29% increase from the prior year [S2]. That supports continued development, but it does not remove utility, permitting or community-acceptance constraints, which remain central to project schedule risk.
Decision criteria for a leaseable MW ranking
Rank operators on seven measurable fields: commissioned MW, construction MW, preleased MW, uncommitted MW, contracted rent per kW per month, committed power by date and tenant concentration. A high total portfolio does not compensate for missing four of those fields. At the building level, require evidence of utility service, switchgear, cooling, generators or uninterruptible power, and commissioning status. [S4]
Leasing records should also distinguish campus commitments from letters of intent and reserved capacity. The 7,481 MW figure for preleased hyperscale colocation under construction provides a useful scale measure [S6], but a published aggregate cannot be translated into an operator ranking without operator-level disclosures and consistent definitions.
Two trackable signals will shape the next ranking cycle: whether the 1% North American vacancy rate begins to move as more than 66 GW of construction reaches operation, and whether tenants continue signing for 2028 delivery because large contiguous blocks remain scarce [S2].