AI Center › AI Infrastructure, Compute, Data & Economics
1ArtificialIntelligence · The Global AI Ecosystem
AI Infrastructure, Compute, Data & Economics
AI Infrastructure, Compute, Data & Economics maps the sources of record for the physical and economic substrate of artificial intelligence — the chips and clusters models are trained on, the electricity data centres draw, the export controls governing advanced computing, and the official measurement of AI investment, patents and adoption — each linked to the authority that publishes it.
This is a switchboard: it attributes and points to the primary sources. It does not rank them or add an assessment of its own. Every entry names its publisher, links to that publisher's own page, and records when it was verified. Figures are deliberately not reproduced here, because they change continuously — follow each link for the current data.
The organisations below produce the measurements this field runs on. The AI Center gathers them in one neutral, sourced place and points you to each; it does not re-originate or re-estimate their work. Coverage in this release is concentrated in OECD-aligned institutions — the EU, the United States, the OECD, WIPO and industry consortia — which reflects where comparable official measurement is currently published, not the whole world's activity.
Compute, hardware and performance measurement
Where the numbers on AI hardware, clusters and training compute actually come from.
Data on Machine Learning Hardware
A database of AI accelerators recording per-chip attributes including computational performance in different numeric formats, memory, power draw, release price and release date, published as an interactive table with a downloadable CSV.
Data on GPU Clusters
A database of large GPU clusters and supercomputers used for AI training and inference, recording chip type and quantity, aggregate performance, power capacity, hardware cost, owner and operational status, with a downloadable CSV.
Data on AI Models
A database of machine learning models spanning the field's history that records training compute in FLOP alongside parameters, training dataset size, training hardware, training duration, estimated training cost and publication date, with a downloadable CSV.
Live data: AI compute
An interactive live-data area presenting indicators on the availability of AI compute, including share of cloud zones by AI capability and GPU availability by country and AI capability.
Historical Billings Report
Free spreadsheet and PDF tables of monthly worldwide semiconductor billings and three-month moving averages, broken out by the four WSTS reporting regions (Americas, Europe, Japan, Asia Pacific).
TOP500 List
A list issued twice a year, in June and November since 1993, ranking submitted supercomputer installations by measured performance on the High Performance Linpack benchmark, with per-system records of cores, accelerators, vendor, site, country and power.
MLPerf Benchmarks
An engineering consortium that defines the MLPerf benchmark suites — including MLPerf Training, MLPerf Training HPC and MLPerf Inference for datacenter, edge, mobile and tiny systems — and publishes the submitted, reviewed results for named hardware and software configurations.
Energy, electricity demand and data centres
What official bodies publish on the electricity and energy footprint of data centres.
Energy and AI (World Energy Outlook Special Report)
An IEA World Energy Outlook special report analysing the relationship between artificial intelligence and the energy system, covering data centre electricity demand, grid and supply-chain constraints, and applications of AI within the energy sector.
Key Questions on Energy and AI (World Energy Outlook Special Report)
An IEA World Energy Outlook special report revisiting the energy and AI relationship after the earlier Energy and AI report, covering data centre electricity demand growth, investment, and grid and supply-chain response.
Energy performance of data centres
The Commission's page on the data centre energy performance monitoring and reporting obligation introduced by the Energy Efficiency Directive, linking the European database that collects and publishes energy performance and water footprint data for data centres with significant energy consumption, together with reporting guidance and work on a rating scheme.
Commission Delegated Regulation (EU) 2024/1364 on the first phase of the establishment of a common Union rating scheme for data centres
The delegated act of 14 March 2024 setting out the information and key performance indicators — including total and IT-equipment energy consumption, installed IT power demand, floor area, redundancy levels, renewable energy, waste heat re-use and water usage — that in-scope data centres report to the European database.
United States Data Center Energy Usage Report: 2025 Update (LBNL-2001758)
A DOE-funded technical report estimating historical and projected electricity consumption of data centres in the United States and expressing it as a share of national electricity use across a range of scenarios, updating the December 2024 edition of the same series.
Today in Energy: Data center server energy use grows across the commercial building stock
An EIA analysis article presenting Annual Energy Outlook 2026 projections of electricity consumed by data centre servers and its share of U.S. commercial-sector electricity consumption over the projection period.
Data Centres Metered Electricity Consumption
An official national statistics series reporting metered electricity consumption by data centres in Ireland, published with data tables and methodology notes.
Export controls and compute sovereignty
The legal instruments controlling advanced computing exports, and the public programmes building sovereign compute.
Export Administration Regulations (EAR), 15 CFR Parts 730–774
The regulations administered by BIS that govern U.S. export, reexport and in-country transfer of dual-use commodities, software and technology, presented on the issuing agency's own site and sourced from the Electronic Code of Federal Regulations.
Commerce Control List, Supplement No. 1 to Part 774 of the EAR (including advanced-computing ECCNs 3A090 and 4A090)
The list of commodities, software and technology subject to BIS export licensing, organised by Export Control Classification Number, including the Category 3 and Category 4 entries 3A090 and 4A090 that define the performance thresholds used to control advanced computing integrated circuits and the computers containing them.
Regulation (EU) 2021/821 — Union regime for the control of exports, brokering, technical assistance, transit and transfer of dual-use items (recast)
The EU regulation establishing the Union's dual-use export control regime, requiring authorisation for the export of items on its Annex I control list and setting rules on brokering, technical assistance, transit and intra-Union transfers.
Exporting dual-use items
The Commission's guidance page on the EU dual-use export control regime, explaining Regulation (EU) 2021/821, the Annex I common control list, EU General Export Authorisations and the items subject to intra-EU transfer controls.
AI Factories
The EuroHPC JU's programme page for AI Factories, ecosystems built around AI-optimised supercomputers that provide computing resources and support services to European industry, startups and scientific users for developing AI models, with a section listing AI Factory Antennas.
AI Factories (EU policy page)
The Commission's policy page describing the AI Factories initiative, its legal basis in an amendment to the EuroHPC Regulation, and the related AI Gigafactories concept and InvestAI Facility for large-scale AI model training infrastructure.
AI economics, investment and adoption
Official measurement of AI investment, patents, skills and take-up by firms.
Live data: Investments in AI and data
A set of interactive visualisations of venture capital investment into AI and data start-ups, built on Preqin transaction data and broken down by country, industry and sub-theme including generative AI and AI compute.
Live data: AI jobs and skills
An interactive live-data area presenting indicators on demand for AI-related skills and AI job postings across countries and occupations.
AI Index Report 2026 — Chapter 4: Economy
The AI Index chapter compiling indicators on AI private investment, corporate adoption, company revenue and compute costs, labour-market effects, productivity studies and robot installations, published as a web chapter with a downloadable PDF.
Top Generative AI Patent Trends in 2025 (WIPO Technology SPARK Report)
WIPO's patent-analytics page presenting its update to the generative-AI patent landscape, counting published patent families and scientific publications over a multi-year window and breaking them down by model type, output mode, country of origin and patent owner.
PATENTSCOPE Artificial Intelligence Index
A search index that maps AI techniques, functional applications and application fields to ready-made PATENTSCOPE queries and IPC/CPC classification symbols for retrieving AI-related patent applications.
Use of artificial intelligence in enterprises
Eurostat's Statistics Explained article reporting official EU survey results on the share of enterprises using AI technologies by enterprise size class, economic activity and country, together with the AI technology types used and the barriers to adoption reported.
Business Trends and Outlook Survey (BTOS) data, with AI supplement
The Census Bureau's data page for its high-frequency survey of employer businesses, which includes supplemental questions measuring whether businesses use AI to produce goods or services, which AI technologies they use, expected future use and reported effects on employment and business organisation, published by sector, state and metropolitan area.
How this switchboard is maintained
- Every entry names the organisation that publishes it and links to that organisation's own official page.
- Entries are described neutrally — what each source measures, not whether it is good. Nothing here is ranked, rated or recommended.
- Figures are not reproduced. Dataset sizes, electricity totals, investment sums and patent counts all change continuously, so this page describes what a source measures and links to it rather than quoting a number that would be stale by the time it is read.
- Several of these publishers estimate values that developers do not disclose, and say so on their own pages. Where a source labels its data as estimated, that characterisation belongs to the publisher, not to the AI Center.
- Each entry records the date it was last verified.
Where do the figures on AI training compute come from?
There is no official registry of AI training compute. Epoch AI, a research organisation, publishes figures assembled from published papers, developer announcements and its own estimates, and labels each record with a confidence tier where the developer has not disclosed the value.
This matters when reading any claim about how much compute a model used: for many systems the number is an informed estimate rather than a disclosed fact, and the publisher says so. Epoch AI publishes its model, hardware and cluster datasets openly with downloadable data.
Source: Epoch AI — Data on AI Models ↗
Who measures the electricity data centres use?
Measurement is national and sectoral rather than global. The International Energy Agency analyses data centre electricity demand within the energy system; in the European Union, Commission Delegated Regulation (EU) 2024/1364 establishes a reporting scheme for data centre energy performance; and national statistical offices publish their own figures where they collect them.
Ireland's Central Statistics Office is one of the few national statistics offices publishing metered data centre electricity consumption as an official series. In the United States, Lawrence Berkeley National Laboratory produces periodic data centre energy usage reports and the Energy Information Administration publishes related analysis. Because the collection basis differs between these sources, figures from one are not directly comparable with another.
Source: Commission Delegated Regulation (EU) 2024/1364 — data centre energy performance reporting ↗
Cite this directory
1BusinessWorld AI Center, "AI Infrastructure, Compute, Data & Economics." https://1businessworld.com/ai-center/ai-infrastructure-and-compute/ Version as of July 28, 2026.
The AI Center is informational only. It is provided by 1BusinessWorld strictly for general informational and educational purposes. Nothing in the AI Center constitutes, or should be construed as, legal, regulatory, compliance, technical, engineering, security, investment, financial, or other professional advice, or a recommendation, endorsement, solicitation, or offer regarding any technology, product, model, provider, framework, or course of action. 1BusinessWorld is not a law firm, regulatory authority, standards body, conformity-assessment or certification body, or investment adviser, and nothing in the AI Center creates any advisory, fiduciary, attorney-client, or other professional relationship with 1BusinessWorld. Although the AI Center references official materials published by legislatures, regulators, standards bodies, research organizations, and other named authorities, 1BusinessWorld makes no representation or warranty, express or implied, as to the accuracy, completeness, timeliness, or fitness for any purpose of any content, and, to the fullest extent permitted by law, disclaims all liability for any loss or damage of any kind arising directly or indirectly from the use of, or reliance on, any information presented. Laws, regulations, standards, technical practices, and AI capabilities change frequently and differ by jurisdiction; readers must verify all information against the current official text or source and consult qualified legal, compliance, technical, and other professional advisors before acting. Any decision relating to the development, deployment, procurement, or governance of AI systems is made solely at the reader's own risk. Last reviewed: July 28, 2026.