Statistics pages in this category are usually a wall of numbers with no sources, and the numbers rot fast — figures that were right in early 2025 are commonly still being quoted in 2026 having grown several times over.
Everything below is attributed and dated. Where a widely-repeated figure could not be traced to a credible source, it is not here.
📋 Key Takeaways
- Gartner forecasts $2.59 trillion in worldwide AI spending for 2026, a 47% year-on-year increase
- Amazon, Google, Meta and Microsoft have guided to roughly $725 billion of combined 2026 capex
- ChatGPT passed 900 million weekly active users in March 2026, up from 400 million in February 2025
- Microsoft disclosed 4.7 million paid GitHub Copilot subscribers, up about 75% year on year
- 84% of developers use or plan to use AI tools — but only 33% trust the accuracy of the output
Market Size and Spending
Gartner’s 2026 forecast breaks down as follows:
| Segment | 2025 | 2026 | Change |
|---|---|---|---|
| Worldwide AI spending | — | $2.59T | +47% |
| AI software | $282.8B | $453.2B | +60% |
| AI cybersecurity | — | $51.3B | ~2x |
| AI models | $15.5B | ~$32.6B | +110% |
AI infrastructure — optimised IaaS, servers, network fabric, processing semiconductors and devices — is the largest single segment at over 45% of total spending.
Source: Gartner AI spending forecast. Last checked 22 July 2026.
Long-run value estimate. McKinsey Global Institute puts the potential contribution of generative AI at $2.6–4.4 trillion annually in global corporate profits, from 63 use cases across 16 business functions. About three-quarters of that concentrates in customer operations, marketing and sales, software engineering, and R&D. The analysis examined 850 occupations and 2,100 work activities across 47 countries.
Source: McKinsey Global Institute.
Infrastructure Capex
The four largest hyperscalers have guided to roughly $725 billion of combined capital expenditure in 2026, up from about $410 billion in 2025.
| Company | 2026 capex guidance |
|---|---|
| Amazon | ~$200B |
| $175–185B | |
| Meta | $115–135B |
| Microsoft | $110–120B |
These are company guidance figures, not outturn, and they are revised at each earnings call. Roughly three-quarters of the total is directed at AI infrastructure specifically.
Source: 2026 capex guidance as reported by CNBC. See our AI data centers breakdown.
Users and Revenue
ChatGPT: more than 900 million weekly active users as of March 2026 — up from 400 million in February 2025 and roughly 700 million by July 2025. It reached 100 million users within two months of launch, the fastest consumer product to that milestone.
OpenAI revenue: annualised run-rate above $25 billion as of March 2026, roughly $2 billion a month. Reported split is about 65% ChatGPT subscriptions, 25% API, 10% partnerships, with over 50 million consumer subscribers and more than 9 million paying business users.
GitHub Copilot: 4.7 million paid subscribers, disclosed by Microsoft at FY26 Q2 earnings on 28 January 2026 — the first time the figure was officially released — up about 75% year on year.
Perplexity: over 1 billion search queries a month by mid-2026, up from 780 million in May 2025. Valued at $23 billion in its Series E-6 round, closed January 2026.
Developer Adoption
The largest published dataset on how developers actually use AI is Stack Overflow’s annual survey. The 2025 edition drew 33,662 responses on the AI usage question.
The interesting movement is not adoption — that is close to saturated, up from 76% the prior year. It is trust, which fell: 46% now actively distrust accuracy against 33% who trust it, and positive sentiment dropped from over 70% in 2023–24 to 60% in 2025.
Agent adoption lags well behind general AI use. 52% of developers either do not use agents or stick to simpler tools, 38% have no plans to adopt them, and 14.1% use agents daily.
Tool usage among AI IDEs: Cursor at 18%, Claude Code at 10%, Windsurf at 5%.
Source: 2025 Stack Overflow Developer Survey — AI.
Model Pricing
Frontier API pricing per million tokens, input / output, from the vendors’ own pricing pages:
| Model | Input | Output |
|---|---|---|
| Claude Fable 5 | $10 | $50 |
| GPT-5.6 Sol | $5 | $30 |
| Claude Opus 4.8 | $5 | $25 |
| Claude Sonnet 5 | $3 | $15 |
| GPT-5.6 Terra | $2.50 | $15 |
| Gemini 3.1 Pro | $2 (≤200k) | $12 (≤200k) |
| GPT-5.6 Luna | $1 | $6 |
| Claude Haiku 4.5 | $1 | $5 |
Sources: Claude models · OpenAI pricing · Gemini API pricing. Last checked 22 July 2026.
The long-run direction is down. Cost per unit of capability has fallen sharply since 2023 as hardware improved, distillation matured and competition intensified — DeepSeek’s efficiency work accelerated that. Note that Gemini 3.1 Pro doubles above 200k tokens, and Sonnet 5’s introductory rate expires 31 August 2026.
What We Are Not Publishing
Several figures that circulate widely in this category could not be traced to a credible primary source, so they are not on this page: precise enterprise AI adoption rates by industry, “percentage of new code written by AI”, global AI job-posting multiples, and share-of-workforce-displaced estimates.
Those numbers exist in many articles. They generally trace back to vendor marketing or to a survey whose methodology is not published. If you need them for a business case, find the underlying study first — in several cases there isn’t one.
Also see: AI data centers 2026 · OpenAI vs Anthropic vs Google · AI agents in 2026 · AI Tool Finder
FAQ
How big is the AI market in 2026?
Gartner forecasts $2.59 trillion in worldwide AI spending for 2026, up 47%. Lower figures you may see typically count only AI software or only generative AI, not infrastructure.
How many people use ChatGPT?
More than 900 million weekly active users as of March 2026. The frequently-quoted 400 million figure is from February 2025.
How much are big tech companies spending on AI?
Amazon, Google, Meta and Microsoft have guided to roughly $725 billion of combined capital expenditure in 2026, up from about $410 billion in 2025.
Do developers actually trust AI tools?
Adoption and trust have diverged. 84% use or plan to use AI tools and 51% of professionals use them daily, but only 33% trust the accuracy of output and just 3.1% highly trust it.
Is AI getting cheaper?
Per unit of capability, yes — frontier API prices have fallen substantially since 2023. Total spending is rising fast regardless, because usage is growing faster than unit costs are falling.