AFCOM’s 2026 State of the Data Center survey reports average rack density rising from 16 kW to 27 kW—a roughly 69% year-over-year increase—while average facility size approaches 38 MW. The figures describe respondents’ averages, not a universal rack specification. Their significance is strategic: AI is pushing operators to design power, cooling, networking and expansion as one system, often at hall or pod scale rather than rack by rack.
AFCOM’s 2026 findings at a glance
| Measure | Reported result | Qualification |
|---|---|---|
| Average rack density | 27 kW, versus 16 kW previously | AFCOM respondent average; approximately 69% growth |
| Average facility size | Approaching 38 MW, versus about 32 MW | Respondent-reported facility average |
| AI-capable infrastructure plans | 74% | Share of respondents planning deployment |
| AI capacity impact | 72% | Expect AI workloads to increase capacity requirements |
| Liquid cooling | 36% deployed; 28% planning adoption in 12–24 months | Survey responses; methods vary |
| On-site generation | 25% deployed; 23% planning adoption within 12 months | Reported in coverage of the AFCOM survey |
| Expected density growth | Nearly 70% | Expect increases over the next 12–36 months |
AFCOM identifies the publication as its 10th annual edition. Its public executive summary describes a survey of data-center professionals, but does not publish every methodological detail, including respondent count, facility mix, regional split and the precise denominator for “average rack density.” Treat these as AFCOM respondent findings rather than a statistically complete census of all global facilities (AFCOM executive summary).
What the 27 kW average does—and does not—mean
“Average rack density” could describe current operating load, designed capacity, a respondent estimate or another measure; the public summary does not fully define it. It also does not tell you whether the figure is an IT-load measure, a nameplate value, or a facility-wide average across every deployed rack.
That distinction matters. A facility can report a 27 kW average while a small AI zone contains 80 kW, 100 kW or higher racks alongside storage, networking and conventional enterprise equipment. AFCOM’s January 22, 2026 webinar listing says racks are pushing beyond 80 kW, but that is an attributed event-description statement, not a universal average (AFCOM webinar listing). The useful interpretation is a rising design baseline, not a claim that every rack now draws 27 kW.
#1 Best Overall
- 【Powerful Load-bearing】12U Network Rack Open Frame is constructed from durable cold rolled steel; Rack shelf supports enhance stability, wall-mounted capacity of 130lbs, the ground-mounted up to 260lbs
- 【Considerate Designs】Open-frame layout, including a top panel adding space, anti-slip shelf stops fixing devices and compatible racks for stack and expansion to meet requirements of home server rack
- 【Complete Accessories】A 12U open frame server rack, two ventilated shelves, four shelf stops, four velcro straps and a set of equipment mounting screws
- 【Versatile Application】Ideal for space-efficient multi-device setups in warehouses, retail, classrooms, offices and more; Excellent choices as AV Rack/IT Rack
- 【Effortless Setup】 Network Rack includes hardware, a comprehensive manual, mounting hole drilling template and an online assembly video to simplify setup
Why AI produces denser racks
Accelerators concentrated in one chassis
Training and some inference systems place multiple GPUs or other accelerators in a server, then combine many servers in a tightly coupled cluster. NVIDIA’s GB200 NVL72 illustrates the pattern: the manufacturer describes a liquid-cooled platform with 72 Blackwell GPUs and 36 Grace CPUs (NVIDIA GB200 NVL72).
High utilization and fast interconnects
AI training can keep accelerators busy for long periods, producing a more persistent thermal load than lightly utilized general-purpose servers. High-speed switches, optical transceivers and dense cabling add both power demand and physical constraints. Integrated pods also make it impractical to distribute equipment casually across a conventional hall.
Reference designs show the upper end
A Vertiv reference design describes a 1.2 MW GB200 NVL72 superpod using eight racks at approximately 132 kW each, plus lower-density racks (Vertiv 360AI reference design). That example is not a market-wide average; it shows why a facility average can conceal radically different local requirements.
Rank #2
- ADJUSTABLE DEPTH: 4-Post 42U open frame server rack with 4 vertical rails and adjustable mounting depth 22" to 40" (56,0cm to 101,7cm); Compatible with various servers / switches / data / AV and other IT equipment; EIA/ECA-310-E Compliant
- EASY ASSEMBLY: Mobile network rack with easy-to-follow assembly instructions and online video; Compact flat-pack shipping to avoid damage and facilitate installation; Total product height of 80.3in (204 cm) with casters, 78in (198cm) without casters
- COLD ROLLED STEEL: Durable 4 Post 19in open frame rack designed for ventilation with 42U mounting height and 1320lb (600kg) weight capacity (stationary); 3 install options included: casters, levelling feet, or base-plate to secure rack to the floor
- HARDWARE INCLUDED: Rolling computer/data rack includes cage nuts and screws to mount equipment, easy to read Units (U) and depth adjustment markings, cable management hooks for organization, and required assembly tools
- THE IT PRO'S CHOICE: Designed and built for IT Professionals, this 42U rack is backed for 2-years, including free lifetime 24/5 multi-lingual technical assistance
Why larger facilities are being planned from the start
A move from roughly 32 MW to 38 MW is not simply a decision to install more servers. Utility interconnections, substations, transformers, generators, cooling plants and network pathways often take years to coordinate. Adding them after construction can require major civil work, outages and new permits.
Dell’Oro’s Alex Cordovil told Data Center Knowledge that the minimum viable increment is increasingly a complete hall or pod, with networking, power and cooling engineered together (Data Center Knowledge). Building too large creates stranded land, power or cooling capacity if demand forecasts fail; building too small can leave an operator unable to serve a committed AI customer.
Power is the binding constraint
Liquid cooling cannot compensate for unavailable electrical capacity. High-density deployments require adequate medium-voltage service, transformers, switchgear, busway, UPS systems, generators and distribution redundancy. Operators must distinguish permitted capacity, connected capacity and firm power available on the date equipment arrives.
Rank #3
- Adjustable Depth: 23-40'' adjustable depth is used for servers and network equipment, ensuring enough space for AV equipment, components, and cabling, while allowing you to access ports and equipment from multiple sides.
- Strong Load Capacity: Ground-Mounted Load Capacity: 500 lbs, Wall-Mounted Load Capacity: 150 lbs. The av rack is made of carbon steel for better weldability performance and can help save space while meeting your need to place multiple devices.
- User-friendly Design: Ergonomic design makes the open frame av rack easier to use. The additional top panel is able to place other items with more available space. Roller design moves anywhere and anytime, is convenient, and is more energy-saving.
- Complete Accessories: We provide the accessories you need, including 2 x Pallets, 145 x M5*10 Cross Head Screws, 4 x Casters, 4 x M10*50 Expansion Screws,10 x M6*12 Cage Nuts, 1 x Grounding Wire, 1 x User Manual.
- Wide Application: The server rack wall mount maximizes the use of available space, suitable for retail venues, classrooms, offices, and other places where space is limited.
Utility delays are encouraging interest in on-site generation. AFCOM coverage reports 25% of respondents already using it, up from 19%, with another 23% planning implementation within 12 months. AFCOM describes data-center and power-sector planning as increasingly connected (Power Is Not a Line Item; The Gigawatt Era Begins).
Behind-the-meter generation can mitigate a grid constraint, but it introduces fuel supply, emissions, noise, maintenance, permitting, interconnection and community-impact issues. It may be attractive in one jurisdiction and uneconomic in another; it is not an automatic solution.
Cooling choices at 27 kW, 80 kW and beyond
Air cooling
Air remains practical for conventional servers, storage and networking zones with available airflow and chilled-water headroom. As density rises, however, fan energy, containment, balancing and hot-spot control become more demanding.
Rank #4
- Universal 19” Rack Mount Compatibility – Perfect for pro audio, video, IT, and network gear. Compatible with mixers, routers, patch panels, servers, power amps, and more.
- Heavy-Duty Load Capacity – Built to support up to 550 lbs. Ideal for studio gear, DJ setups, server equipment, and AV components that demand serious stability.
- Robust Steel Frame & Design – Made with 1.5mm thick steel and weighs 36 lbs for maximum durability, reduced vibration, and long-term reliability in any setting.
- Mobile & Secure – Preinstalled with 3” industrial-grade caster wheels (lockable), making it easy to move and position your rack exactly where you need it.
- All-In-One Setup Kit Included – Comes with 34 rack screws (5mm & 6mm), a 1U blank spacer, and an assembly tool—ready for fast installation out of the box.
Rear-door heat exchangers
These capture exhaust heat at the rack and can preserve much of an air-based room design. They add rack-level mechanical complexity, coolant distribution and maintenance requirements, and do not remove electrical or facility-level constraints.
Direct-to-chip liquid cooling
Cold plates and manifolds remove heat from GPUs and CPUs while some components remain air-cooled. The design requires compatible servers, coolant distribution units, leak detection, fluid-quality controls, service procedures and suitable heat rejection. Retrofitting piping and CDUs into an occupied room can be harder than deploying a purpose-built pod.
Immersion cooling
Immersion can support very high heat transfer and reduce dependence on room airflow, but requires compatible hardware, fluid handling, filtration, altered service procedures and clear OEM and warranty support. AFCOM’s public material discusses rear-door exchangers and immersion but does not establish one method as dominant (AFCOM density analysis).
Best Value
- Adjustable Depth: Depth adjustable from 23" to 40", this open frame server rack accommodates servers and network equipment while providing ample space for A/V gears and cable management. Enjoy easy access to ports and devices from multiple angles.
- High Weight Capacity: Supports up to 300 lbs on the floor (200 lbs when adjusted to maximum depth) and 200 lbs when wall-mounted (depth cannot be adjusted in wall-mounted mode). Made from carbon steel for superior welding performance and durability, this open frame rack is designed to save space while accommodating multiple devices.
- User-Friendly Design: Designed with your convenience in mind, this open frame server rack features an top shelf for extra storage and improved space utilization. The rolling casters let you move it effortlessly wherever you need it, making setup and movement a breeze.
- Widely Applicable: Maximize your space with this adaptable open frame server rack, designed to make the most of every inch. Ideal for retail spots, classrooms, offices, and any area where space is at a premium, it delivers practical solutions for your storage needs.
- Everything You Need: Our open-frame rack comes with fully equipped accessory kit for easy setup and secure installation: 2 x Trays, 4 x Casters, 1 x set of Screws, 16 x M6*12 Cage Nuts, 1 x Grounding Wire, 1 x Internal & External Hex Wrenches, and 1 x User Manual.
Retrofit or new build?
A retrofit may work when
- Spare utility and electrical-distribution capacity exists.
- The floor and structure can support the equipment and coolant infrastructure.
- Piping, rear-door exchangers or another viable liquid path can be installed.
- UPS, generators, busway and redundancy can be upgraded without unacceptable outages.
- AI equipment can be isolated in a high-density zone beside conventional workloads.
New construction is more likely when
- Target density exceeds the building’s electrical or thermal envelope.
- Structural loading, ceiling or plenum capacity is limited.
- There is no practical coolant route or utility expansion path.
- The business case depends on large halls, pods or superpods.
- Retrofit downtime would exceed the value of the added capacity.
Legacy facilities are not automatically unsuitable. Feasibility depends on electrical, thermal, structural, network and operational limits—not floor area alone.
The hidden constraints behind a high-density deployment
- Networking: Dense switching and fiber pathways can become the limiting infrastructure before floor space does.
- Structure and logistics: Floor loading, rack dimensions, delivery routes and service clearances must be verified.
- Operations: Specify leak detection, automatic isolation, CDU and pump redundancy, coolant monitoring, spare connectors and trained technicians.
- Commissioning: Test simultaneous electrical and thermal load, not merely individual racks.
- Security: AFCOM reports human threats as the leading security concern in its 2026 report, overtaking ransomware; concentrated AI systems also place more asset value in fewer locations (AFCOM security coverage).
- Water and sustainability: Cooling architecture, climate and accounting boundaries determine water use. Closed loops can reduce ongoing water consumption, but heat still must be rejected.
What this means for operators and colocation buyers
Operators should model current, committed and peak rack power; separate facility, hall, pod and rack averages; and reserve high-density zones rather than assuming uniform loading. Their procurement review should cover utility delivery dates, electrical topology, cooling compatibility, network capacity, staffing, expansion and stranded-capacity risk.
Enterprise customers should ask a provider:
- What rack power is guaranteed continuously, and what is merely a peak or design maximum?
- Is cooling included, and who owns the CDU, manifolds and coolant loop?
- Is the facility certified for the exact server or GPU platform?
- What lead time applies to a high-density deployment or later density upgrade?
- What are the provider’s power, water and heat-rejection assumptions?
- How are maintenance windows, liquid incidents and service parts handled?
- Is the site powered today, or only permitted and planned?
The financial and planning interpretation
For investors, infrastructure executives and technology buyers, the valuable asset is not a headline megawatt number. It is deliverable power with thermal headroom, network pathways, commissioning capability and an expansion plan that matches actual customer demand. A large campus can still lack rack-level capacity; a smaller site can be valuable if it has firm power and a workable high-density design.
AI also is not one workload. Training, inference, recommendation systems, analytics and HPC differ in utilization, latency, network and thermal behavior. Density assumptions should therefore be tied to a specific platform and workload profile rather than an “AI” label.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Bottom line
AFCOM’s 16-to-27 kW movement is best read as evidence that the average design baseline is changing. The durable shift is from incremental rack expansion to integrated power-and-cooling systems planned around high-density compute. Liquid cooling may be part of that solution, but utility capacity, electrical distribution, networking, structures and trained operations determine whether the deployment is actually feasible.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
