You want one AI number and find five different ones.
That's rarely down to bad sources. Each source simply counts something else.
DataReportal estimates 2.42 billion active users of generative AI platforms in April 2026. It explicitly leaves open whether that means as many unique people. GWI asked about any form of AI instead. In its sample, 81.2% of online adults had used AI in the previous month.
In this article, you'll find the key AI statistics for 2026 on users and the market, each with its source.
Here are the key numbers first:
- DataReportal estimates 2.42 billion active users of generative AI platforms in April 2026, and warns that this may not be a unique-person count.
- In a separate GWI survey, 81.2% of online adults in a 54-market sample said they had used some form of AI in the previous month during Q4 2025.
- Providers publish individual subscription and coding-tool figures, but there is no reliable deduplicated global total for either group.
- Grand View Research estimates the AI market at $539.5 billion in 2026 and projects $3,497.3 billion by 2033.
- The OECD says AI companies received $258.7 billion, or 61% of global venture capital, in 2025.
- In McKinsey's 2026 survey, nearly nine in ten respondents reported regular AI use; 44% reported enterprise-wide scaling.
1. How many people use AI?
OpenAI, Meta, and Google publish large user figures for their AI products. These metrics can indicate reach, but they do not all count the same kind of use.
How widespread AI is also depends on whether a survey asks about generative chatbots or AI features across any software.
1.1 Global Usage Estimates and Their Limits
DataReportal’s April 2026 report estimates 2.42 billion active users of generative AI platforms such as ChatGPT, Gemini, and Doubao. That equals 29.2% of the world population.
This chart shows how the world population splits up overall:
gradually.aiEach person counts only once, in their most advanced group. Someone who pays for an AI subscription does not also appear among the free users.
The ITU offline figure and DataReportal's 2.42 billion active users are sourced. DataReportal does not rule out that some people are counted more than once, so that share is closer to an upper bound.
The two smallest groups are my estimate. OpenAI alone reports more than 59 million paying users, and other providers publish no totals. That points to roughly 80 million paying people. For coding agents, the figures for OpenAI Codex (20 million active users), GitHub Copilot (4.7 million paid subscriptions) and Cursor (more than 1 million daily users) overlap, so I estimate around 30 million. That estimate is uncertain, because the Codex figure of 20 million comes from an employee post without a time window. Only more than 5 million weekly active users is official, and with that floor the group would be considerably smaller.
A separate GWI survey asked more than 240,000 people aged 16 and over in 54 markets about recent AI use. In Q4 2025, 81.2% of online adults in that sample said they had used at least one form of AI in the previous month. The report says this includes AI features inside software, not just chatbots.
1.2 How Many People Use AI Chatbots?
The answer is that nobody knows exactly. What exists are the figures from individual providers:
The table shows three published provider metrics with different time windows. ChatGPT reports weekly active users, while Meta AI and Gemini report monthly active users.
You can find more detail on the individual services in our ChatGPT statistics. Section 10 shows how web traffic splits across chatbot websites.
1.3 AI Adoption by Country
The country comparison uses only Eurostat values with one consistent definition:
Germany sits at 32.3% in the Eurostat measurement, just below the EU average of 32.7%. Broader definitions produce higher values. According to Bitkom, 58% of people aged 16 and over in Germany use any kind of AI (2026).
Two patterns run through every study:
First, age. In the EU, 64% of 16- to 24-year-olds use generative AI, but only 7% of 65- to 74-year-olds (Eurostat, 2025). Second, wealth: richer countries like Norway sit far ahead, poorer ones like Romania trail far behind, and you can see the exact numbers in the chart above.
1.4 Generative AI Use by Age Group
The age bands in detail, each showing the share of people who used generative AI in the last three months (EU-27, 2025):
About 57 percentage points separate 16- to 24-year-olds from 65- to 74-year-olds. The share falls with every age band.
1.5 AI Use Based on Microsoft Telemetry
A third measurement method complements the surveys: Microsoft derives AI use from anonymized telemetry and extrapolates it to the population aged 15 to 64. For June 2026, the United Arab Emirates and Singapore lead by a wide margin. Germany ranks 23rd in the report at 32.3%, just below the USA (33.0%) and far above the worldwide figure. The almost identical Eurostat value for Germany (32.25% in 2025) is a coincidence, because the two measurements are independent.
The telemetry values mainly show how the countries rank against each other.
However: usage is not the same as routine. According to Pew Research, 49% of US adults have used AI chatbots (surveyed February 2026), but for most it stays occasional. Only 10% of people worldwide use AI chatbots weekly for news (Reuters Institute, 2026).
1.6 How Many People Pay for AI?
Published subscription figures cover individual providers. They are not enough to count unique people paying for AI worldwide:
In February 2026, OpenAI reported more than 50 million consumer subscribers and more than 9 million paying business users. Anthropic, Google, and Perplexity have not published comparable subscriber totals in the official sources reviewed here.
A comparison with the best-known subscription service puts these figures in context:
gradually.aiNetflix reported more than 325 million paid memberships for Q4 2025. That is more than six times as many as OpenAI's consumer subscribers.
I broke down what the subscriptions cost in our ChatGPT statistics and Claude statistics.
1.7 What Is Documented About AI Coding Tools?
Published figures describe different activity and subscription metrics. They do not add up to a global total of people using coding agents:
OpenAI reported more than 5 million weekly active Codex users in June 2026. In August, OpenAI employee Tibo Sottiaux separately cited 20 million active users without defining a time window. Anthropic said in February that weekly active Claude Code users had doubled since the start of the year, but gave no absolute figure.
Microsoft reported 4.7 million paid GitHub Copilot subscriptions in January 2026.
If you want the full picture on these tools, check the Claude Code statistics and Codex statistics.
1.8 The Blind Spot: 2.2 Billion People Are Offline
One detail often gets lost in AI debates. According to the ITU, 2.2 billion people were still completely offline at the end of 2025, around 27% of the world population.
Without internet access, cloud-based generative AI services are out of reach.
That also puts the hype line about AI being the fastest adopted technology in history into perspective. The available usage figures mostly describe people with internet access.
1.9 What Does This Mean for You?
I have been looking at AI numbers almost daily since early 2022. For global comparisons, I now pay particular attention to what a metric includes and whether it counts unique people.
In my work bubble, seemingly everyone uses AI. My newsletter readers use AI, my clients use AI, everyone on LinkedIn talks about agents. The survey and provider figures show that my work bubble is not automatically representative of everyone.
The good news:
If you sometimes feel like you are too late, you may be wrong. The data shows widespread AI use, but the measured share changes sharply with the definition and population.
DataReportal estimates that active GenAI users grew by 1.4 billion from April 2025 to April 2026. Future surveys will show how usage changes.
If you want to use that head start, our overview of the best AI tools is a good place to start.
2. How large is the AI market in 2026?
Grand View Research estimates the global AI market at $390.9 billion in 2025 and $539.5 billion in 2026. It projects $3,497.3 billion by 2033, a compound annual growth rate of 30.6% from 2026.
Generative AI is a segment of that market. Grand View Research's report, updated in June 2026, puts it at $22.2 billion in 2025 and an estimated $29.6 billion in 2026. It projects $324.68 billion by 2033, at 40.8% annual growth.
Gartner published a differently scoped measure on September 16, 2026. It forecasts $2.67 trillion in worldwide AI spending for 2026, up 49.5% from the prior year. This includes spending on AI services, software and infrastructure, among other categories.
Only the 2026 and 2033 values are published. The years in between in the chart are calculated from the growth rates.
3. Investment: Selected AI funding rounds
The OECD counted $258.7 billion in venture capital for AI companies in 2025, equal to 61% of $427.1 billion in global venture capital. The chart shows selected rounds announced in 2026:
OpenAI described its $122 billion as committed capital. xAI announced its Series E in January, before the merger with SpaceX became effective on February 2, 2026, according to the SEC merger document.
On February 9, Databricks said it was completing more than $7 billion in investments, including about $5 billion in equity financing at a $134 billion valuation and about $2 billion in additional debt capacity.
On August 13, Databricks announced the closing of a separate $5 billion strategic round at a $190 billion valuation. Cognition announced a Series E of more than $2 billion on September 8.
On September 29, Bloomberg and Reuters reported that OpenAI was discussing a new round of at least $30 billion at a valuation of about $1.4 trillion. According to a report on October 7, those talks were still ongoing.
4. Private AI company valuations
The chart shows selected, most recently publicized valuation markers:
Anthropic announced its post-money valuation on May 28, 2026, and OpenAI reported its on March 31. Cognition announced a $48 billion valuation with its Series E on September 8, and TechCrunch reported on June 11 that Prometheus raised $12 billion at a $41 billion valuation.
On October 26, 2026, Anthropic publicly filed its IPO prospectus (Form S-1 with the SEC). It reports revenue of about $16.3 billion for the first half of 2026, up from about $1.2 billion in the first half of 2025. The shares are to trade on the Nasdaq under the symbol “ANT.” The document does not yet give a price range.
This is how Anthropic's valuation developed, according to the company's own announcements:
gradually.aiIn September, Bpifrance announced a €3 billion Mistral round at a post-money valuation above €21 billion. xAI has not been an independent private company since its merger with SpaceX became effective on February 2.
More company data is available in our statistics on OpenAI, Anthropic and xAI.
5. AI adoption in organizations
McKinsey, Eurostat and the U.S. Census Bureau each measure AI use in organizations differently.
5.1 Organizations Using AI Since 2022
McKinsey's repeated surveys track the share of respondents whose organizations use AI in at least one business function; in 2025 McKinsey asked about regular use.
McKinsey's 2025 survey was in the field from June 25 to July 29, 2025, and included 1,993 respondents across 105 countries. High performers are respondents who attribute at least a 5% EBIT impact to AI and report significant value from it.
The 2026 Stanford AI Index reports a different measure. About 70% of organizations used generative AI in at least one function in 2025.
5.2 Scaling and High Performers in 2025 and 2026
McKinsey's latest survey ran from May 4 to June 8, 2026, and included 1,719 respondents across 97 countries. Nearly nine in ten reported regular AI use in at least one business function. The chart compares scaling, use in at least three functions and the high-performer share for 2025 and 2026:
EBIT impact was little changed. In 2026, 37% of respondents report any EBIT impact from AI.
5.3 Enterprises Using AI in the EU and Germany
McKinsey surveys organizations worldwide, while official statistics count enterprises. Eurostat records how many enterprises with 10 or more employees (excluding the financial sector) use at least one of eight AI technologies:
In 2025, about one in five EU enterprises with ten or more employees used AI, and about one in four in Germany. The EU value has more than doubled since 2023.
5.4 AI Use by Industry in the U.S.
By industry, there is a clear gradient. In the US, the Census Bureau regularly measures which businesses use AI in any business function. Information and finance lead, accommodation and agriculture trail:
The values refer to US businesses of all sizes.
How many people use AI overall is covered in section 1.
6. What people use AI for
An NBER working paper, which OpenAI presented in an accompanying announcement, analyzes a sample of about 1.1 million consumer ChatGPT conversations from May 15, 2024, through June 26, 2025. The chart shows shares of classified messages in that sample:
Practical guidance, seeking information and writing account for 77.1% of classified messages combined. Programming accounts for 4.2% of all messages.
6.1 OpenAI Signals v2: a separate June 2026 snapshot
OpenAI publishes a separate, more recent measure in Signals v2. These shares cover classified consumer messages from June 2026. The monthly sample contains 300,000 messages from individual ChatGPT accounts.
OpenAI separately reports that 30.4% of messages were work-related and 69.6% were not work-related.
6.2 AI in higher education
The UK HEPI survey puts the same questions to full-time undergraduates each year. Use rose from two thirds to almost everyone. The share who put AI-generated text directly into assessed work has quadrupled too, but stays small.
7. AI at work
The figures come from different surveys and populations:
Gallup asked U.S. employees how often they use AI at work. The Glean Work AI Index surveyed 6,000 full-time digital workers in the U.S., U.K. and Australia. Pew surveyed 1,458 U.S. teens ages 13 to 17 and their parents online. More language-model data is in our LLM statistics.
Studies measure how many jobs AI affects in different ways. The chart separates two studies: in January 2024 the IMF estimated the share of jobs exposed to AI, and in May 2025 the ILO estimated the share of workers in occupations exposed to generative AI:
Exposure means AI can take over or support tasks in these jobs. The IMF writes that in advanced economies about half of exposed jobs could benefit from AI. The other half could face lower demand, lower wages or disappear.
The ILO considers the transformation of jobs the most likely outcome. The World Economic Forum expects 170 million new and 92 million displaced jobs worldwide by 2030, driven by several trends rather than AI alone. More, including individual occupations, is in the AI job market statistics.
8. Data centers: Electricity use and projections
The International Energy Agency publishes a projection for electricity use across all data centers.
In its updated central scenario, the IEA projects global data center electricity use to rise from 485 TWh in 2025 to 950 TWh in 2030. That would be around 3% of global electricity demand in 2030. Electricity use by AI-focused data centers is projected to triple over that period.
The series of IEA figures shows how fast electricity use is growing:
How much electricity and water a single query uses is covered in the AI energy statistics.
9. The U.S., China and Europe
The Stanford AI Index compares countries by AI models and by private AI investment.
9.1 Notable AI Models by Country
For 2025, the Stanford AI Index counts 59 notable models affiliated with the U.S., 35 with China and 8 with South Korea. The selected countries below had one model each:
Models whose authors are affiliated with several countries count more than once.
In March 2026, four companies were within 25 points of one another on the human-rated Arena Elo leaderboard, according to Stanford.
9.2 Private AI Investment by Region
U.S. private AI investment is high, but the worldwide total also includes Europe and other regions:
The remaining amount is $25.45 billion. U.S. investment was about 23 times China's, although Stanford notes that private investment may not fully capture China's government guidance funds.
10. Which AI chatbot leads in web traffic?
A Similarweb web-traffic analysis published on September 7, 2026, reported the following shares for the chatbot websites it tracked:
The Decoder reported that Claude's share rose from 1.9% to 9.3% year over year, about 4.9 times as high. Similarweb estimated about 5.6 billion monthly visits to ChatGPT's website in May 2026.
A newer Comscore report published September 22 compared each platform's share of measured AI prompt volume from January to June 2026. ChatGPT fell from 70% to 50%, Gemini rose from 17% to 30% and Claude from 2% to 11%. See our comparison of the best AI chatbots for more.
11. How people view AI
Selected survey results on sentiment toward AI, each with sample and survey period:
In Pew's spring 2025 survey across 25 countries, the median share of adults who felt more concerned than excited about AI was 34%. Another 42% felt equally concerned and excited, and 16% were more excited. In none of the countries did more than 30% say they were mainly excited.
The Stanford AI Index also reports that 59% of respondents in 2025 saw more benefits than drawbacks from AI products, while 52% said those products made them nervous.
On trust in AI regulation, a separate 2026 Pew survey across 12 countries found these medians:
A 2025 survey across 25 countries reported 53% for the EU, 37% for the U.S. and 27% for China.
Among U.S. Gen Z respondents ages 14 to 29, the share excited about AI fell from 36% in 2025 to 22% in 2026. One source of concern, AI-generated fake content, is covered in my deepfake statistics.
How often chatbots invent content when summarizing is shown by the hallucination rate per model in the LLM statistics.
12. AI forecasts: What do older studies project?
Earlier studies estimated substantial economic potential. PwC estimated in 2017 that AI could contribute an additional $15.7 trillion to the global economy by 2030, in 2016 prices. McKinsey estimated $13 trillion in additional global output by 2030 in 2018. A 2023 McKinsey analysis estimated $2.6 trillion to $4.4 trillion in annual generative AI value across 63 use cases.
13. Conclusion
The available evidence points to widespread reported use, large private investments and widely varying company valuation markers.
For technical measures, see our LLM statistics; for tools, browse our guide to the best AI tools.
