Data Centre Statistics: The 2026 Numbers
A statistics roundup compiled by Brian Dean for Programs dot com, tracking the global data centre build-out and its power, water and land costs. It aggregates figures from Data Center Map, McKinsey, the Electric Power Research Institute and others into one referenced page. The host is a commercial site that compares AI and cybersecurity degree programmes, so treat it as a well-sourced aggregator rather than primary research.
What it is
A single-page compilation of data centre statistics, put together by Brian Dean and published on Programs dot com, a commercial education site that compares AI and cybersecurity degree programmes.
It is an aggregator, not original research, but each figure is attributed, which makes it usable as a jumping-off point to the underlying sources.
The headline figures
- 12,259* data centres across *179 countries (Data Center Map)
- The US holds 4,767* of them, roughly *nine times more than any other single country (Data Center Map)
- Demand predicted to nearly triple by 2030 (McKinsey)
- $7 trillion in investment expected through 2030 (McKinsey)
- Data centres projected to consume 17% of total US electricity by 2030 (Electric Power Research Institute)
- A typical facility runs around 100,000 sq ft*. Hyperscale sites reach up to *10 million sq ft
- 38 US states offer tax incentives to attract data centre construction (National Conference of State Legislatures)
- Large facilities can draw up to 5 million gallons of water per day (Environmental and Energy Study Institute)
Why it matters
The AI ethics conversation tends to stay in the interface: bias, output, consent, provenance. This page is about the concrete part. The buildings, the grid connections, the aquifers.
Three things fall out of the numbers. The 17% of US electricity* projection is the one to sit with, because it reframes AI from a software question into an energy-policy question. The *38 states with tax incentives figure explains why these facilities land where they land, and who is subsidising them: the local area absorbs the grid load, the water draw and the construction traffic, while the tax revenue that would normally offset it has been signed away to win the build. And the water figure matters most where it is least reported, because siting decisions rarely follow water availability.
The commercial context is worth naming. The host site sells attention to degree programmes in AI and cybersecurity, so a page arguing that this sector is enormous and growing is not a neutral artefact. That does not make the numbers wrong, since they come from McKinsey, EPRI and the NCSL, but it does mean you should click through to the source rather than cite the roundup.
What to look at first
- The country-level counts, for a sense of how concentrated the infrastructure actually is.
- The electricity projection, then go straight to EPRI's own publication.
- The tax incentive count, if you care about who pays for this.
Key takeaways
- 12,259* data centres worldwide. The US has *4,767, about nine times the next country.
- Projected to reach 17% of US electricity* by 2030, with *$7 trillion of investment through 2030.
- 38 states* subsidise construction through tax incentives. Large sites can use *5 million gallons of water a day.
- Compiled by Brian Dean on a commercial degree-comparison site. Good sourcing, but follow the citations.
Photo by Geoffrey Moffett on Unsplash
· End of dispatch ·
About the correspondent
Dead Good Club
The House Curator brings together stories worth knowing about. Our editor selects and rounds up useful resources for the library, while summaries (usually AI-generated) offer a quick overview and helpful context. Every post links to the original article or video, so you can explore the full story for yourself. If you'd like to write for The Dead Good Club and see your work featured here, get in touch by email.
