2,500 dots = 8.1 billion humans. Color = most advanced AI interaction, .
The Feb 2026 snapshot is the anchor — it reproduces Damian Player’s original viral chart, which was built from the Microsoft AI Economy Institute diffusion report. That report found 16.3% of the global population had used generative AI in H2 2025, giving a baseline of roughly 1.3 billion people.
All other months are extrapolated from public disclosures: ChatGPT weekly active users (OpenAI blog + Sam Altman posts), Gemini monthly active users (Google I/O), Meta AI (embedded in WhatsApp/Instagram), and paid subscriber counts from OpenAI, Anthropic, and Google. Coding-tool numbers draw on GitHub Copilot and OpenAI Codex usage announcements.
Because companies report different metrics (DAU vs MAU vs WAU), all figures carry real uncertainty — that’s why ranges are shown instead of false precision.
The raw number of new users is still growing — roughly 50–60 million more people try AI each quarter. But the growth rate (percentage change) has been flattening since late 2025, which is typical S-curve diffusion: early adopters are saturated, and growth now depends on converting the skeptical majority.
What is accelerating: paid subscriptions and coding tool usage. Those segments are growing faster than free usage, suggesting the engaged core is deepening its habit even as broad casual use slows.
The big unknown is the next step-change — a truly ambient AI (wearables, OS-level integration) could re-accelerate the curve. Without that, expect gradual S-curve flattening through 2027.
Because most of humanity hasn’t used AI yet. Even at Jul 2026’s estimate of ~1.56 billion ever-users, that’s only about 19% of 8.1 billion people. The gray dots aren’t empty placeholders — each one represents ~3.2 million people who have never interacted with a generative AI tool.
This is partly geography (limited internet access in parts of Africa, South Asia, Southeast Asia) and partly choice. Adoption is still highly concentrated in North America, Europe, East Asia, and urban India.
Gold represents people paying roughly $20/month for an AI subscription — ChatGPT Plus, Claude Pro, Gemini Advanced, and similar. As of Jul 2026 estimates: around 34 million people globally, or about 0.4% of the world.
That’s fewer than the number of people who subscribe to a single major streaming service in the US alone. The price point ($20/mo) is affordable in wealthy countries but prohibitive for much of the world. It’s also a chicken-and-egg problem: people won’t pay until they see value, and they often don’t see value until they’ve used it enough to build a habit.
Getting from ~20% to 50% adoption would require reaching ~4 billion people — roughly 2.5 billion more than today. The Microsoft report’s diffusion model suggests that at current rates, 50% global adoption is a 2030–2033 target, depending on infrastructure buildout and whether a killer use case (like AI-assisted health or translation) breaks through in lower-income markets.
The fastest historical analog is smartphone adoption, which took about 10 years to go from 10% to 50% global penetration. AI could move faster — or slower, if regulatory friction or public trust issues intervene.