Artificial Intelligence Worldwide: Statistics and Facts 2026

Artificial intelligence in 2026: the measured picture

Artificial intelligence has moved from a research field into a capital-intensive industry. The money now flows in three layers: investors fund the companies that build models and applications, cloud providers and large platforms build data centres, and chip makers supply the processors that train and run the models. Each layer reports its own numbers, and together they show how fast the build-out is happening.

Use is the second half of the story. Surveys of large companies, official business statistics and telemetry of AI tools all point the same way, but they measure different things, so this report keeps them apart.

Market size forecasts vary widely by method and are covered in our separate AI market size analysis. Here the focus is on measured spending, output, adoption, energy and policy.

AI investment: 581.7 billion dollars in 2025

The Stanford Institute for Human-Centered AI tracks corporate investment in AI companies across four channels: private investment, mergers and acquisitions, minority stakes and public offerings. In its 2026 AI Index, total corporate investment rose from 14.57 billion dollars in 2013 to a record 581.69 billion dollars in 2025. The 2025 total was 129.9 percent higher than the 253.02 billion dollars recorded in 2024.

Private investment, which captures venture and growth funding rounds, was the largest channel at 344.66 billion dollars, up 127.5 percent from 151.48 billion dollars. Mergers and acquisitions added 214.44 billion dollars. The remaining 22.59 billion dollars came from minority stakes and public offerings. Generative AI companies alone raised 170.87 billion dollars of private money, nearly half the private total and more than three times their 2024 level.

Private money leads: venture and growth rounds supplied 59 percent of the 581.7 billion dollars invested in AI companies in 2025.

The AI Index also groups private funding by focus area, and the biggest category in 2025 was AI infrastructure, models, research and governance at 143.22 billion dollars, larger than any application area.

Where AI money goes: the United States takes 83 percent

Private AI investment is concentrated in one country to a degree seen in few other industries. The United States attracted 285.88 billion dollars in 2025, which is 82.9 percent of the global private total. China was second at 12.41 billion dollars, the United Kingdom third at 5.90 billion dollars, followed by France at 4.36 billion dollars, Canada at 4.28 billion dollars and Germany at 4.09 billion dollars. Europe as a whole raised 20.92 billion dollars.

One country dominates: US companies raised 285.9 billion dollars of private AI money in 2025, more than the next ten countries combined many times over.

Counting companies rather than dollars gives a broader picture. In 2025, 3,499 AI companies received a new round of funding worldwide. The United States accounted for 1,953 of them, or 55.8 percent, but the next places were much closer together, led by the United Kingdom, China and India.

For the companies that dominate this spending, see our profiles of Nvidia, Amazon and the wider big tech revenue comparison.

Who builds the models: 59 from the US, 35 from China

The AI Index counts notable models, meaning models that set a new state of the art, were widely cited or had large real-world impact. In 2025, organisations based in the United States released 59 notable models, China 35 and South Korea 8. Industry produced more than 90 percent of notable models, a share that has risen steadily as training runs have become too expensive for most universities.

A two-country race: the United States and China released 94 of the notable AI models counted in 2025, with South Korea a distant third.

Model development rests on computing power. The AI Index estimates that global AI compute capacity has grown 3.3 times a year since 2022 and reached about 17.1 million H100-equivalent processors at the end of 2025, up from roughly 50,000 in early 2022. Nvidia supplies more than 60 percent of that capacity. AI data center power capacity reached 29.6 gigawatts, and the United States alone hosts 5,427 data centers, more than any other country. Our data centers statistics page covers the facilities themselves.

Patents tell a different story from models. The AI Index reports that China holds 74.2 percent of AI patents granted worldwide. For generative AI specifically, the World Intellectual Property Organization counted about 54,000 patent families filed between 2014 and 2023, more than a quarter of them published in 2023 alone. Inventors based in China filed 38,210 of these, six times as many as the United States.

The chip bottleneck: Nvidia data center revenue of 193.7 billion dollars

The clearest single measure of AI spending on hardware is the data center revenue of Nvidia, the main supplier of AI accelerators. In fiscal 2022, which ended in January 2022, the segment earned 10.61 billion dollars. Revenue rose to 15.01 billion dollars in fiscal 2023, then jumped to 47.5 billion dollars in fiscal 2024 after the launch of ChatGPT set off a race to build training clusters.

The increase continued: 115.2 billion dollars in fiscal 2025 and 193.7 billion dollars in fiscal 2026, which ended in January 2026. Data center sales were 89.7 percent of total company revenue of 215.9 billion dollars. In the second quarter of fiscal 2027, reported in August 2026, data center revenue was 89.0 billion dollars in a single quarter, up 117 percent from a year earlier.

Eighteen times in four years: Nvidia data center revenue grew from 10.6 billion dollars in fiscal 2022 to 193.7 billion dollars in fiscal 2026.

For the third quarter of fiscal 2027, Nvidia guided to total revenue of about 108.0 billion dollars. Growth rates are slowing as the base gets larger, but the absolute increase each year is still rising. How the company compares with other chip and technology firms is covered in our ranking of the biggest companies in the world by market value.

Data centre spending: big tech capex heads past 650 billion dollars

The other side of Nvidia revenue is the capital spending of its largest customers. Four companies, Alphabet, Amazon, Meta and Microsoft, spent about 384 billion dollars on property and equipment in their latest full fiscal years, most of it on servers and data centres. All four have guided to much higher spending in 2026.

Spending roughly doubles: guidance implies about 710 billion dollars of combined 2026 capex from four companies, up from about 384 billion dollars.

Guidance has been raised during 2026: Alphabet moved from 175 to 185 billion dollars to 195 to 205 billion dollars, and Meta from 115 to 135 billion dollars to 130 to 145 billion dollars. Microsoft lowered its figure from about 190 to about 175 billion dollars only because it reclassified some leases.

These are company figures with different definitions, so the totals are indicative rather than exact. The direction, though, is not in doubt. The IEA counted more than 400 billion dollars of capital spending by five large technology companies in 2025 and expects a further increase of about 75 percent in 2026. Revenue trends for the same firms can be followed in our Alphabet annual revenue and Facebook statistics reports.

AI and electricity: data centres to use 950 TWh by 2030

Every new data centre needs power, and AI servers use far more of it than conventional ones. The International Energy Agency estimates that data centres consumed about 415 terawatt hours of electricity in 2024, around 1.5 percent of global electricity use. Consumption grew 17 percent in 2025 to about 485 terawatt hours, and electricity use in AI-focused data centres grew 50 percent.

Doubling by 2030: data centre electricity use is set to rise from 485 terawatt hours in 2025 to about 950 terawatt hours, roughly 3 percent of global demand.

Demand is concentrated in a few places. In 2024, the United States accounted for about 45 percent of data centre electricity use, China for about 25 percent and Europe for about 15 percent. In the IEA base case, US consumption rises by about 240 terawatt hours to 2030, an increase of 130 percent, and Chinese consumption by about 175 terawatt hours, or 170 percent. Europe adds about 45 terawatt hours, a rise of 70 percent.

Accelerated servers, the category that includes AI processors, are growing about 30 percent a year against 9 percent for conventional servers and account for roughly half of the projected increase. For grid operators, the issue is less the global total than local concentration: data centre clusters in northern Virginia, Dublin or Singapore already take a large share of regional supply. National context is available in our reports on the global electricity market and electricity in the United States.

Business adoption: 88 percent of organisations use AI

McKinsey has asked executives about AI use every year since 2017, which makes its survey the longest consistent view of corporate adoption. In 2017, 20 percent of respondents said their organisation used AI in at least one business function. The share peaked at 58 percent in 2019, dipped to 50 percent in 2022, then climbed quickly once generative AI arrived: 55 percent in 2023, 72 percent in early 2024, 78 percent by mid-2024 and 88 percent in the 2025 survey.

From one in five to nearly nine in ten: reported AI use rose from 20 percent of organisations in 2017 to 88 percent in 2025.

The Stanford AI Index, which draws on the same survey, notes that adoption was highest in Europe at 91 percent and North America at 90 percent in 2025. About 70 percent of organisations reported regular use of generative AI in at least one function, compared with about a third in 2023.

Using AI somewhere is not the same as using it at scale. The 2026 edition of the McKinsey survey, based on 1,719 respondents interviewed in May and June 2026, found that 44 percent of organisations say AI is now scaling across the enterprise, up from 38 percent a year earlier. Among large organisations, 40 percent report scaling AI agents, up from 27 percent, while smaller firms were flat at 22 percent.

Sector use cases are covered in more depth in our reports on AI in finance, AI in healthcare and AI-driven retail in China.

Europe: one in five firms uses AI, from 42 percent in Denmark to 5 percent in Romania

Official statistics give a more cautious reading. Eurostat asks a representative sample of EU enterprises with 10 or more employees whether they use any AI technology. The share was 7.65 percent in 2021 and 8.06 percent in 2023, then jumped to 13.48 percent in 2024 and 19.95 percent in 2025. In other words, the share of EU firms using AI rose by about 12 percentage points in two years, after barely moving before.

A Nordic lead: 42.0 percent of Danish firms used AI in 2025, eight times the 5.2 percent in Romania.

The gap between countries is wide. Denmark, Finland and Sweden lead, each above 35 percent, followed by Belgium, Luxembourg and the Netherlands at about a third. The large economies sit near or below the average: Germany at 26.0 percent, Spain at 20.3 percent, France at 18.2 percent and Italy at 16.4 percent. Romania, Poland and Bulgaria are below 9 percent. Denmark recorded the largest increase from 2024, at 14.5 percentage points, followed by Finland at 13.5 points.

Size explains much of the difference with private surveys. More than half of large EU firms used AI in 2025, against 17 percent of small firms, and small firms make up the great majority of enterprises in every country. The most common uses were analysis of written language at 11.8 percent of firms, generation of pictures, video or audio at 9.5 percent and generation of written or spoken language at 8.8 percent.

United States: about one in five firms uses AI

The US Census Bureau asks a large sample of employer businesses every two weeks about AI in its Business Trends and Outlook Survey. Under the original question, which asked about AI used to produce goods or services, the share rose from 3.7 percent in September 2023 to 5.4 percent in February 2024.

In November 2025 the Census Bureau broadened the question to ask about AI use in any business function. Under the new wording, overall use hovered between 17 and 20 percent from December 2025 to May 2026, and reached 37 percent among firms with 250 or more employees. The Federal Reserve Board puts US firm adoption at about 18 percent at the end of 2025. The two series cannot be joined because the question changed.

Taken together with Eurostat, the official data suggest that a fifth of firms in both the EU and the United States now use AI in some form. The much higher McKinsey figure reflects a sample tilted towards large companies, where adoption is close to universal.

AI users by country: 70.1 percent in the UAE, 31.3 percent in the US

Microsoft measures how many people use generative AI tools, based on anonymised device telemetry adjusted for market share. In the first quarter of 2026, 17.8 percent of the global working-age population, people aged 15 to 64, used generative AI, up from 16.3 percent in the second half of 2025. The share rose again to 18.8 percent in the second quarter of 2026.

Small, connected economies lead: seven in ten working-age adults in the UAE use generative AI, against fewer than one in three in the United States.

The ranking does not follow the investment ranking. The United States, home to most leading AI developers, was 21st at 31.3 percent, while small, highly digital economies such as the UAE and Singapore lead. China reached 16.4 percent on this measure, below the world average. In the second quarter of 2026, the UAE rose to 73.3 percent and Singapore to 64.3 percent, while Ireland at 49.9 percent and France at 49.6 percent moved ahead of Norway.

The divide between richer and poorer countries is large. Microsoft estimates usage at 27.5 percent in the Global North and 15.4 percent in the Global South in the first quarter of 2026, so a working-age adult in a richer country is almost twice as likely to use generative AI.

ChatGPT: 900 million weekly users

The most widely used single AI product is ChatGPT. OpenAI said 500 million people used it every week at the end of March 2025, 700 million weekly active users by September 2025 and more than 800 million users in October 2025. In February 2026, the company reported more than 900 million weekly active users, more than 50 million consumer subscribers and more than 9 million paying business users.

400 million added in 11 months: ChatGPT weekly users rose from 500 million in March 2025 to more than 900 million in February 2026.

Only about one weekly user in 18 pays for a consumer plan. For context on how quickly other platforms grew, see our data on the benefits of generative AI in social media marketing.

Jobs: 170 million created, 92 million displaced by 2030

The World Economic Forum Future of Jobs Report 2025, based on a survey of more than 1,000 large employers, expects technology and other structural trends to create 170 million jobs and displace 92 million between 2025 and 2030, a net gain of 78 million. About 22 percent of current jobs will be affected. AI and big data top the list of fastest-growing skills.

Employers plan to respond in two ways at once. 77 percent intend to upskill their workforce for AI, while 41 percent expect to reduce headcount where AI automates tasks. The McKinsey 2026 survey points the same way, with 39 percent of respondents expecting AI to reduce total employment at their organisation in the coming year, up from 32 percent.

The PwC 2026 Global AI Jobs Barometer, based on job advertisements and company financial data, finds that workers with AI skills command an average wage premium of 62 percent, up from 57 percent a year earlier. Companies in the most AI-exposed sectors recorded productivity growth of 34 percent between 2018 and 2025, against 24 percent for those least able to use AI, and the number of jobs requiring AI skills grew 69 percent against 9 percent for all jobs.

Regulation: 58 US federal AI rules in 2025

Governments are catching up. The AI Index counts AI-related regulations issued by US federal agencies, which rose from one in 2016 to 58 in 2025; the Executive Office of the President accounted for 28 of them. At state level, the number of AI-related bills passed rose from fewer than 10 in 2020 to 150 in 2025. The number of witnesses called to US congressional hearings on AI went from 5 in 2017 to 102 in 2025.

In Europe, the AI Act entered into force on 1 August 2024 and is the first comprehensive AI law in a major economy. Its obligations apply in stages, starting with bans on certain uses and rules for general-purpose models, followed by requirements for high-risk systems.

Outlook: what the numbers point to for 2026 and beyond

The build-out is still accelerating: capex guidance implies another large step up in 2026, Nvidia guided to about 108 billion dollars of revenue for a single quarter, and the IEA expects data centre electricity use to double by 2030.

Adoption is following more slowly: about one firm in five in the EU and the United States uses AI, and scaling across the enterprise is under way in only a minority of firms. Three questions will decide the pace: whether revenue from AI products grows fast enough to justify the capital spending, whether power grids can connect new data centres in time, and how regulation evolves in Europe, the United States and China.

Artificial intelligence in ten numbers

Questions people ask about artificial intelligence data