Worldwide AI spending reached about $1.76 trillion in 2025 and is forecast to hit $2.59 trillion in 2026, up 47% (Gartner). ChatGPT reached 1.2 billion weekly users. One in six humans used generative AI last year. And many artificial intelligence statistics in circulation are mislabeled projections, category errors, or 2023 numbers that primary sources have since cut by 10x. This page collects the numbers that survive sourcing, each labeled measured, forecast or reported, plus the figures journalists most often get wrong.
How big is the AI market in 2026?
Gartner's estimates put the AI market at about $1.76 trillion for 2025, heading for $2.59 trillion in 2026, up 47%. Gartner's September 2025 release first estimated 2025 at $1.48 trillion; the January 2026 forecast restated 2025 to $1.76 trillion and put 2026 at $2.52 trillion, raised in May to $2.59 trillion. Growth from the $1.48 trillion first estimate would read as 75%.
Inside it: generative AI spending was forecast at $644 billion for 2025, up 76.4%, and AI infrastructure is the largest segment at over 45% of 2026 spending, per the May 2026 release. The "trillions added to the economy by 2030" headlines mix market revenue, GDP impact and spending; quote them only as vendor projections.
What percentage of companies actually use AI?
Between 10% and 88%, depending on who is asked. The honest AI adoption rate is a range whose endpoints measure different worlds:
- 88% of organizations regularly use AI in at least one business function, per McKinsey's State of AI (November 2025 edition, 1,993 respondents; 78% a year earlier), and 79% regularly use generative AI. The 2026 edition (1,719 respondents) again finds almost nine in ten using AI, and only 37% reporting a positive EBIT contribution.
- Roughly 10-20% of US firms use AI, per the US Census Bureau's Business Trends and Outlook Survey: about 10% under the original "producing goods or services" wording in late 2025 (8.8% for firms under 250 employees in mid-2025, SBA), and 17-20% after the question widened to "any business function" from December 2025 to May 2026.
The gap is methodology: McKinsey surveys self-selected, digitally engaged participants at all levels; the Census samples all US businesses. Value capture is thinner still: only about 6% qualify as McKinsey's high performers, attributing 5% or more of EBIT to AI. We unpacked that gap in AI ROI reality.
How many people use AI?
About 1 in 6 people on Earth (16.3%) used a generative AI tool in the second half of 2025, up from 15.1% in the first half (Microsoft's Global AI Adoption in 2025 report, January 2026). The per-product numbers:
- ChatGPT: 1.2 billion weekly users as of September 2026 (OpenAI DevDay 2026 recap), up from 800 million in October 2025.
- Google Gemini: 1 billion monthly users as of August 2026 (Google), up from 650 million in Q3 2025 and 750 million in Q4.
- Meta AI: more than 1 billion monthly actives across Meta's apps as of October 2025 (Q3 2025 earnings call), up from "almost 1 billion" in Q1 2025 and a billion in May 2025.
- 44% of US adults have used ChatGPT (February 2026 survey), including 61% of under-30s, up from 34% a year earlier, and 38% of employed adults use AI chatbots for tasks at work (Pew Research, June 2026). In 2025, 28% of employed adults used ChatGPT for work; the work question changed wording (any chatbot in 2026, ChatGPT in 2025), so that pair is not a clean trend.
Unit hygiene: OpenAI reports weekly actives; Google and Meta report monthly. Comparing 1.2B WAU to a 1B MAU figure as like-for-like is a category error.
How much money is going into AI?
AI took about half of all global venture funding in 2025: $211 billion of $425 billion, up 85%. In 2026 it went higher: roughly 80% in Q1 and 64% in Q3.
- Crunchbase counts $211B and roughly 50% for 2025; CB Insights has $226B and 48% of a $469B total; PitchBook has more than half of $512B. In 2026, Crunchbase counted $242B, or 80%, in Q1, more than 70% in Q2 and $102B of $159B, or 64%, in Q3.
- US private AI investment was $285.9 billion in 2025 vs China's $12.4 billion, a 23x gap (12x a year earlier), per the Stanford HAI AI Index 2026. Global private AI investment reached $344.7 billion, up 127.5%.
- Revenue is catching up, with a label: leaked, prospectus or source-reported figures, not audited filings. OpenAI: $13.07 billion in 2025, up from $3.7 billion (Fortune), a $21.4 billion year-end run rate (The Information via Reuters) approaching $70 billion by late September 2026 (Axios via Reuters). Anthropic: nearly $4.6 billion in 2025, 12-fold growth (IPO prospectus via Reuters), with the run rate up from about $9 billion at end-2025 to $65 billion by end-July 2026 (Bloomberg).
How much energy does AI use?
System-wide, data centers used 485 TWh in 2025, and the IEA expects that to roughly double to about 950 TWh by 2030, around 3% of global electricity by then; per query, a Gemini text prompt uses a measured median of 0.24 Wh. Sources:
- The system numbers are the IEA's Key Questions on Energy and AI (April 2026), updating its 2025 Energy and AI report: 415 TWh in 2024 (1.5% of global electricity), 945-950 TWh by 2030, slightly more than Japan's consumption today. The 2025 report called AI the most important driver, with the US and China leading the increase.
- The per-query numbers are Google's measured 0.24 Wh, 0.03 gCO2e and 0.26 mL of water per median Gemini Apps text prompt (August 2025) and Epoch AI's ~0.3 Wh ChatGPT estimate; Altman has cited 0.34 Wh. Google's comparison: under nine seconds of TV. Both are about 10x below the 2023-era "3 Wh per query" estimate. Caveats: medians, text-only, training and video excluded.
Will AI take jobs?
The best employer-side data: the WEF Future of Jobs Report 2025 (1,000+ employers, 14 million workers, 55 economies) expects 170 million jobs created and 92 million displaced by 2030, a net gain of 78 million, with 39% of existing skill sets transformed or outdated. Editions come every two to three years; these are the current figures.
The most misquoted number: "AI will eliminate 300 million jobs." Goldman Sachs' March 2023 analysis said work equivalent to 300M full-time jobs is exposed to automation, and projected generative AI could raise global GDP by 7% over ten years. Exposure is not elimination; the same report is an argument for growth.
What share of code is AI-written at big tech companies?
- Google says 75% of its new code is AI-generated and approved by engineers as of April 2026, up from more than 25% in late 2024. The metric is AI-generated code that engineers approved: approval, not autonomy.
- A peer-reviewed Science study (Daniotti et al., February 2026) estimated that about 29% of Python functions committed to GitHub by US developers were AI-written by the end of 2024, against 23-24% in Germany and France and 20% in India.
- Frontier labs go further: an Anthropic spokesperson put the company-wide share at 70-90%; Claude Code's head, Boris Cherny, says 100% of his own code is.
Adoption among working developers is near saturation: 84% use or plan to use AI tools (Stack Overflow 2025) and 85% use them regularly (JetBrains 2025). Trust lags (Stack Overflow): 46% distrust the accuracy of AI output against 33% who trust it. We covered that split in the AI code trust paradox; the people numbers are in our developer headcount statistics.
Which AI numbers are most often quoted wrong?
Check articles against this list:
- "300M jobs eliminated." Goldman said exposed, and projected +7% global GDP (2023).
- "MIT: 95% of AI pilots fail." A preliminary, self-published 2025 report from MIT's Project NANDA about measurable P&L impact of enterprise pilots (Fortune); it also found workers at over 90% of surveyed companies using personal AI tools.
- "47% of jobs will be automated." Frey & Osborne (2013) estimated occupations at risk over a decade or two; still cited as current 13 years later.
- "An AI query uses 10x a Google search / 3 Wh." A 2023 estimate; current estimates are ~0.24-0.34 Wh, roughly 10x lower.
- "ChatGPT: fastest app to 100M users." A UBS estimate built on Similarweb data of 100 million monthly users two months after launch (January 2023, reported that February); Threads reported 100 million sign-ups in five days that July.
- Training costs. GPT-4 ~$78M and Gemini Ultra ~$191M are the 2024 AI Index's cloud-rental-price estimates; Epoch AI's amortized hardware-plus-energy method gives about $40M and $30M. Name the methodology.
Every number here carries its source and year; the next refresh lands with full-year 2026 funding data in January 2027. If your question is less "how big is the AI wave" and more "what would it take to ship AI inside our product," our AI development team answers that one for a living.