- Metric
- Covered private employer firms reporting cloud-based computing use in production processes
- Scope
- All covered U.S. private sectors; firm-level and not weighted for nonresponse
- Period
- 2020–2022 use reported in the 2023 survey
- Status
- observed
Are data centers only for Big Tech?
Hospitals, banks, manufacturers, researchers, governments, schools, and communications networks all depend on digital compute and storage. The infrastructure is economy-wide even when a campus operator is a technology company.
Key facts, with scope and limits.
- Metric
- Covered hospital firms reporting cloud-based computing use in production processes
- Scope
- U.S. private employer firms in NAICS 622; does not cover every hospital or measure facility energy
- Period
- 2020–2022 use reported in the 2023 survey
- Status
- observed
- Metric
- Office-based physicians using any electronic health record
- Scope
- United States; digital workload evidence, not cloud or facility attribution
- Period
- 2024
- Status
- observed
- Metric
- Fedwire Funds Service transfers and total value
- Scope
- One Federal Reserve payment service with geographically dispersed data centers
- Period
- 2025
- Status
- administrative
- Metric
- Aurora compute delivered to science
- Scope
- One Department of Energy facility; node-hours are use, not electricity
- Period
- 2025 production year
- Status
- administrative
Cloud use appears across many industries
Cloud computing already appears across much of the private economy. This Census measure counts firms that reported cloud use in production processes. It does not identify the provider or facility.
2023 Annual Business Survey, reference years 2020–2022. Percentage of responding U.S. private employer firms in covered industries. The unit is the firm, not the establishment, and results are not weighted for nonresponse. Public institutions and several industries are outside the survey scope.
| Sector | Value (% of firms) |
|---|---|
| Manufacturing | 25.7 % of firms |
| All covered sectors | 31.4 % of firms |
| Health + social assistance | 36.6 % of firms |
| Telecommunications | 40.6 % of firms |
| Finance + insurance | 45.8 % of firms |
| Professional + technical | 50.8 % of firms |
| Information | 55.3 % of firms |
| Hospitals | 57.2 % of firms |
Many sectors share the same physical layer
Cloud and colocation campuses serve institutions throughout the economy. Technology companies may operate the facilities. The workloads span social media, AI, healthcare, finance, government, and more.
These functions are representative. The diagram provides no market-share estimate. Organizations use different mixes of on-site systems, colocation, cloud, edge facilities, and third-party software.
- Healthcare: clinical systems + imaging
- Finance: payments + fraud controls
- Government: public records + emergency systems
- Education: learning platforms + research
- Manufacturing: design + production systems
- Science: models + scientific data
Use the national baseline to read the local record.
National baseline
Data centers are a physical layer beneath many cloud, transaction, communications, research, and records systems. Institutions may use their own facilities, colocation, public cloud, edge infrastructure, or a mix; national reliance does not disclose one campus's workload.
Local case
Broad sector reliance cannot show that a proposed local campus serves the surrounding hospital, school, or bank. That claim needs tenant and workload disclosure. Without it, describe the facility only as part of a wider shared infrastructure system.
What the evidence supports.
Only Big Tech needs data centers.
Census reports cloud use across manufacturing, finance, healthcare, telecommunications, information, and professional services. Federal payment and scientific systems also document direct reliance on geographically dispersed data centers or computing facilities.
The cloud is virtual, so physical data centers are optional.
Cloud computing pools physical servers, storage, and networks behind an on-demand service. The delivery model still depends on buildings and equipment.
Every new campus directly serves its nearby community.
The economy-wide case is real. A specific facility's users may still be undisclosed or far away. A general hospital or bank example cannot become a local benefit claim without evidence about the tenant and workload.
Ask for these local records.
Without these inputs, a project-specific verdict is incomplete. Treat missing evidence as an open question.
- 01Disclosed tenants, workload classes, and data-sovereignty or latency requirements
- 02Whether the project is enterprise, colocation, cloud, edge, scientific, public-sector, or mixed
- 03Services hosted locally, distinguished from services merely used by local institutions
- 04Redundancy, failover, cybersecurity, and continuity requirements attached to each workload
- 05A ban on claiming local hospitals, schools, or agencies as beneficiaries without workload evidence
Sources used on this page.
- Agency dataGrade A2023 Annual Business Survey technology characteristics
U.S. private employer firms reporting cloud-based computing use in production processes during 2020–2022
These are firm-level responses from covered private employer industries. The measure does not identify a provider, hosting facility, workload size, or local data-center demand. Percentages are not weighted for nonresponse. - GuidanceGrade AThe NIST Definition of Cloud Computing
Cloud computing characteristics, service models, and deployment models
Defines cloud computing and its service models; it does not quantify facility energy, a user's marginal impact, or a particular sector's adoption. - Agency dataGrade AFedwire Funds annual statistics
Fedwire Funds Service transfers during calendar year 2025
Fedwire transfer volume and value demonstrate one critical digital workload; they do not measure the data-center footprint of the finance sector as a whole. - Official reportGrade AResilience of the Fedwire Services
Fedwire applications, redundant data centers, and failover testing
Describes Fedwire's geographically dispersed infrastructure. It cannot establish the architecture or local-serving role of a proposed commercial data center. - Agency dataGrade AOffice-based physician electronic health record adoption
U.S. office-based physicians using electronic health records
Electronic-health-record adoption demonstrates a digital healthcare workload; it does not reveal cloud use, facility location, or energy demand. - Agency dataGrade AWhat is a Medicare Administrative Contractor?
Medicare fee-for-service claims processed by contractors
Administrative-claim volume shows transaction scale. It cannot reveal hosting topology, data-center location, or the share served by commercial cloud providers. - Official reportGrade B2025 ALCF Year in Review
Aurora and Polaris scientific-computing use during 2025
This covers one Department of Energy computing facility. Node-hours measure allocated machine use; they do not measure electricity or the science sector's total infrastructure demand. - Official reportGrade BEnergy and AI
Global data-center and AI electricity demand, efficiency, and scenarios
This global analysis uses scenarios. Per-task energy varies by model, hardware, utilization, data-center efficiency, network, and device, so the report cannot serve as a universal personal-footprint calculator.
