ChatGPT Just Crossed a Language Tipping Point. B2B SaaS Should Pay Attention.

The news, plainly stated

On 30 June 2026, OpenAI published its first major ChatGPT adoption analysis via its Signals dataset, covering accounts created between October 2025 and May 2026. The headline finding: users with a primary language other than English now represent more than half of all active ChatGPT users globally, with Spanish, Portuguese, and Arabic leading non-English usage.

That is not a rounding error. It is a structural shift in who is using the AI channel your prospects are increasingly turning to for vendor discovery, comparison, and shortlisting.

When this happened and over what timeline

OpenAI Signals measures aggregated activity across Free, Go, Plus, and Pro individual plans. The data tracks from July 2023 through to May 2026, giving us nearly three years of directional signal. The non-English majority tipping point appears to have been crossed during 2025, based on the language share chart OpenAI published, with the trend accelerating sharply through early 2026.

Regional growth compounds the picture. Africa and Asia have seen the fastest relative growth in weekly active users since July 2023, indexed against each region's own July 2023 baseline. Lower Human Development Index countries are growing fastest in relative terms, suggesting that English-language market dominance in AI adoption is already a legacy assumption, not a current reality.

How it works: depth and breadth, compounding together

The most strategically important finding in the Signals data is not the language shift in isolation. It is the behavioural deepening that happens as users get comfortable with the platform. Six months after signing up, individual ChatGPT users are sending 50% more messages per day than they did in their first 28 days. They have also doubled the number of distinct capability categories they use across that same window.

OpenAI classifies usage across 53 distinct capability categories using a message classifier. Users are not just doing more of the same thing. They are expanding into new use cases over time, which means the surface area for AI-mediated vendor discovery is growing, not staying static.

The table below summarises the key adoption metrics from the OpenAI Signals release:

Key ChatGPT adoption metrics from OpenAI Signals (data through May 2026)

Metric	Finding	Implication for B2B SaaS
Daily messages at month 6 vs. signup	+50% more messages per day	Users are leaning in, not drifting away. AI search frequency is rising.
Distinct capabilities tried at month 6 vs. signup	2x the number of capability categories	Discovery, research, and comparison tasks are all in scope, not just writing.
Non-English share of active users	Now over 50% of active users	English-only content strategies leave a majority of the AI-search audience unserved.
Fastest-growing regions (relative)	Africa, Asia	Emerging market expansion strategies need AI-search presence, not just traditional SEO.
Fastest-growing languages (share)	Uzbek, Kazakh, Burmese	Long-tail language markets are entering the AI-discovery funnel now.

What this actually changes for B2B SaaS growth teams

Most B2B SaaS content and SEO programmes were built around English-language Google search. The assumption was that global reach came via English content, with localisation as a nice-to-have. That assumption is under pressure from two directions at once.

First, AI Overviews, ChatGPT, and Gemini are now answering vendor discovery queries directly. We have written about this shift in our analysis of GEO versus traditional SEO for B2B SaaS. When a prospect in São Paulo or Riyadh asks ChatGPT which project management tools integrate with Salesforce, the answer is being generated from indexed content, structured data, and entity signals. If your content does not exist in Portuguese or Arabic, or if your structured data does not signal relevance in those languages, you are absent from that answer.

Second, the behavioural deepening data tells us that AI tool usage is becoming habitual, not occasional. Users who are six months in are twice as exploratory and 50% more active than they were at signup. The implication is that the AI-search audience is getting more sophisticated, faster. They are using ChatGPT for research tasks, comparison tasks, and vendor evaluation, not just summarisation.

The "but what about our ICP?" objection, answered

We hear this regularly from B2B SaaS clients: "Our buyers are enterprise, English-speaking, North American." That may be true for your current customer base. But pipeline starts with discovery, and discovery is increasingly AI-mediated. If a growth leader in Warsaw or a procurement lead in Jakarta is evaluating your category, they are likely doing that initial research in their native language, inside ChatGPT or Gemini.

The OpenAI Signals data does not break down by use case. We cannot confirm what share of ChatGPT queries are B2B vendor discovery versus personal tasks. But the breadth data, showing users exploring 53 capability categories over time, makes it clear that research and evaluation tasks are well within scope. Our advisory position is that if your category is searched in English on Google, it is being queried in multiple languages on AI platforms, right now.

The concrete action

Audit your AI-search presence in your top three non-English markets before your next planning cycle. Check whether your brand and product appear in ChatGPT and Gemini responses for category-level queries in Spanish, Portuguese, and French first, since these have the largest active user bases. Then assess your structured data and entity signals for those languages. If you are unsure how to run that audit, our GEO advisory work is designed to do exactly this: map where you are visible in AI-generated answers and close the gaps that matter for pipeline.

The shift has already happened. The question is whether your content infrastructure reflects it.

Cover photo by Levart_Photographer on Unsplash