Yes—demand for hyperscale data centers was accelerating, and the United States held the largest concentration in the historical data available for this report. Synergy Research Group counted more than 390 hyperscale facilities worldwide at the end of 2017 and found that U.S. sites represented 44% of the locations operated by 24 major cloud and Internet companies in its sample. At least 69 additional facilities were then in planning or construction.
Those figures were reported on February 22, 2018. They describe a dated market snapshot, not a current facility census. A later estimate from the National League of Cities places more than half of approximately 1,100 hyperscale data centers worldwide in the United States as of late 2024, but its definition and counting method are not established as identical to Synergy’s. The two data points should therefore be read as evidence of continued U.S. prominence, not as a precise growth curve.
As an Amazon Associate I earn from qualifying purchases.
What the 2017 figures actually measured
The original Building Design + Construction report summarized Synergy Research Group’s end-of-2017 analysis. Its headline number was more than 390 hyperscale data centers worldwide. That count had a clear cutoff: the end of 2017.
Synergy also reported that the United States contained 44% of major cloud and Internet data-center sites in its analysis of 24 large service companies. This is a share of sites in that defined sample. It is not 44% of every data center in the world, 44% of global power capacity, or 44% of hyperscale computing capacity. A site-count percentage can differ substantially from a capacity percentage because facilities vary in size, power density, workload and expansion stage.
#1 Best Overall
The same 2018 report said at least 69 more hyperscale facilities were in various stages of planning or construction. That was a pipeline observation available at the time, not a promise that all 69 projects would be completed and not a current pipeline count.
“Hyperscale growth goes on unabated, and we already have visibility of at least 69 more hyperscale data centers that are at various stages of planning or building,” said John Dinsdale, Synergy Research Group’s chief analyst and research director, in the 2018 report.
Dinsdale’s statement belongs to that 2018 reporting period. It should not be presented as a current assessment.
Free tools Windows power users keep installed
One-click scans. No signup required.
What “hyperscale” means in this context
The report attributed a working definition to IDC: a hyperscale facility has at least 5,000 servers and measures at least 10,000 square feet. Those thresholds are useful for understanding the article’s historical scope, but the available evidence does not establish them as a universal, current standard used by every operator or analyst.
Rank #2
In practice, hyperscale describes an operating model as much as a building size. Facilities are designed for very large, repeatable deployments, rapid expansion and centralized management of computing, storage and networking. A single campus can contain multiple buildings and phases, so a “facility,” “site,” “building” and “campus” may not be interchangeable in a count. That distinction is one reason headline totals from different studies require a methodology check before they are compared.
How to read the later U.S. estimate
A National League of Cities fact sheet published in 2025 estimated that roughly 1,100 hyperscale data centers existed worldwide in late 2024, with more than half located in the United States. The fact sheet presents this as an estimate and does not provide enough detail in the available material to prove that it counts sites in exactly the same way as Synergy’s 2017 dataset.
It is reasonable to say that both sources identify the United States as the leading national location in their respective snapshots. It is not methodologically sound to subtract 390 from 1,100, treat 44% as a directly comparable U.S. baseline, or calculate a compound annual growth rate from the two figures. One source reports a global count and a sampled-company site share; the other reports an approximate global count and a U.S. majority estimate. They may include different facilities, campuses or operator categories.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
| Measure | Value | Date and scope | How to use it |
|---|---|---|---|
| Global hyperscale facilities | More than 390 | End of 2017; Synergy Research Group, as reported by Building Design + Construction | Historical worldwide count |
| U.S. share of sites | 44% | 2017 analysis of sites operated by 24 major cloud and Internet companies | Sample-based site share, not national capacity |
| Projects in planning or construction | At least 69 | Reported in 2018; worldwide pipeline observation | Historical pipeline indication |
| Global hyperscale facilities | Approximately 1,100 | Late 2024 estimate reproduced by the National League of Cities in 2025 | Later estimate; method is not demonstrated to match Synergy’s |
| U.S. position | More than half of the estimated global total | Late 2024 estimate in the National League of Cities fact sheet | Indicates continued U.S. leadership, not a directly comparable trend line |
Why demand for large facilities kept rising
The later material identifies several demand drivers that help explain why the development pipeline continued beyond the 2017 snapshot.
Cloud adoption
Moving applications and data from private rooms into shared cloud infrastructure increases the need for large, standardized facilities. Hyperscale operators can add servers, storage and network capacity in repeatable blocks instead of expanding a small server room one rack at a time.
More data and richer applications
Higher volumes of business records, media, telemetry, backups and machine-generated information require additional storage and processing. Data-intensive services also need resilient network connections and carefully managed power and cooling systems, all of which favor purpose-built campuses when demand is large enough.
Artificial-intelligence workloads
AI training and inference can require dense clusters of accelerators, high-bandwidth interconnects and substantially more power per rack than many traditional enterprise workloads. The National League of Cities fact sheet reproduces a projection that 70% of data-center needs could be tied to advanced AI applications by 2030. That is a forecast quoted by the fact sheet; the original forecasting study was not verified in the available material, so it should be treated as a projection rather than a measured outcome.
Edge computing alongside centralized capacity
Edge computing is also named as a demand driver. Processing some information closer to users or devices can reduce latency and network traffic, while hyperscale campuses continue to handle centralized storage, large-scale analytics and model workloads. Edge sites and hyperscale campuses serve different functions, so growth in one does not make the other unnecessary.
Energy is becoming the defining construction constraint
The National League of Cities fact sheet reproduces a projection of global data-center energy demand rising from 60 GW in 2024 to 219 GW in 2030. Because the underlying forecasting publisher was not verified in the available material, the figures should be attributed to that fact sheet and understood as projected demand, not installed capacity already committed.
For a construction team, the implication is practical: electrical service can determine a project’s schedule before the building is fully designed. Early due diligence should establish the utility’s available capacity, interconnection process, substation requirements, transmission constraints, outage plan and timetable for energization. A site that has inexpensive land but no credible path to power is not development-ready.
What U.S. leadership means for project planning
A national lead does not mean every U.S. location is suitable. Hyperscale development is a site-selection problem involving several interdependent approvals and physical systems.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minutePower and resiliency
- Confirm firm power availability and the date it can be delivered, rather than relying on a nearby transmission line alone.
- Define redundancy targets, backup generation, fuel storage, switchgear, transformers and maintenance access during the concept stage.
- Model the electrical load by phase so the first building does not consume the capacity reserved for later expansion.
Land, water and cooling
- Verify that the parcel can accommodate the data halls, utility yard, cooling plant, security setbacks, stormwater controls and future buildings.
- Assess water rights, municipal supply, drought conditions, discharge limits and the operating consequences of water-efficient or air-cooled systems.
- Account for the heat rejection equipment and noise controls required by high-density computing, not just the white-space floor area.
Connectivity
Hyperscale sites need diverse, high-capacity fiber routes and physically separated entrances. A single carrier path or a route that shares a common trench can undermine resilience even when nominal bandwidth is ample. Network design should be reviewed with the same rigor as the utility plan.
Permitting, zoning and community planning
Large campuses can affect land-use patterns, tax bases, roads, water systems and local noise conditions. The National League of Cities material specifically highlights electricity, water, permitting, zoning and community planning as local infrastructure questions. Developers should identify the responsible authorities, required studies, hearing dates, construction-traffic conditions and any limits on generator operation before committing to a schedule.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.A practical due-diligence sequence
- Define the counting and delivery unit. Decide whether the project is being described as a building, campus, site or phased program. Use the same term in the business case, approvals and public communications.
- Set the technical envelope. Document initial and ultimate IT load, rack density, cooling approach, redundancy objective, floor area and expansion phases.
- Validate utility feasibility. Obtain written information on available and future power, interconnection milestones, substation work and energization dates.
- Screen water and environmental conditions. Check supply reliability, discharge requirements, flood risk, stormwater design and applicable environmental reviews.
- Map fiber and physical access. Require diverse routes, construction access, heavy-haul logistics and room for secure maintenance operations.
- Open the entitlement path early. Confirm zoning, conditional-use requirements, building permits, noise limits, generator rules and public-meeting obligations.
- Plan the community interface. Explain expected construction traffic, permanent staffing, utility work, water use and emergency-generation testing in terms residents and officials can evaluate.
- Stage the campus. Tie each building phase to verified power, cooling, network and permitting milestones instead of assuming the full ultimate campus is immediately deliverable.
How to compare countries or markets without misleading readers
Before comparing two locations, identify the metric. A facility count answers a different question from a share of sampled operators’ sites, critical IT load, total power capacity or projects in the pipeline. The date and geography must be stated alongside the number.
| Comparison question | Metric to standardize | Common error |
|---|---|---|
| Which country has more facilities? | Count of facilities, with a shared definition of facility or campus | Mixing campus counts with building counts |
| Where is operator presence concentrated? | Share of sites in the same operator sample | Calling a sample share a share of global capacity |
| Where is more computing installed? | Critical IT load, rack capacity or another capacity measure | Inferring capacity from the number of sites |
| Where is construction accelerating? | Pipeline count or megawatts, with a stated stage definition | Counting announced projects as completed facilities |
| Which site is easier to build? | Power, land, water, connectivity, permitting and community conditions | Ranking markets from one headline statistic |
What the historical evidence can—and cannot—prove
The evidence supports three careful conclusions. First, hyperscale development was already expanding rapidly by the end of 2017. Second, the United States was the leading national location in Synergy’s sampled-site analysis and remained the leading location in the later National League of Cities estimate. Third, growth creates infrastructure and approval challenges that must be addressed at the project level.
The evidence does not establish a single authoritative, current worldwide facility count using a method directly comparable with the 2017 Synergy dataset. Nor does it prove that every planned project became operational. Readers seeking a present-day market total should insist on a dated source that defines its unit of count, operator coverage and treatment of campuses, expansions and projects under construction.
Bottom line for builders and owners
Hyperscale demand has been a durable construction driver, with cloud adoption, data growth, AI and edge computing all adding pressure for new capacity. The United States’ lead is clear in both the 2017 sampled-site analysis and the later late-2024 estimate, but the figures use different descriptions and should not be plotted as a seamless time series. For any new campus, reliable power, cooling and water strategy, diverse connectivity, realistic permitting and early community planning matter as much as the building itself.
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.
