The 1 Billion Number Is a Floor. The 640% Number Is the Story.

ChatGPT crossed 1 billion monthly active users in May 2026, per Sensor Tower data confirmed by Sam Altman on June 3. That figure is real and it is large. It is also, structurally, the less interesting number in this dataset.

What was actually disclosed. Sensor Tower, a mobile analytics firm that measures app-store installs and in-app engagement, reported ChatGPT at 1 billion MAU (monthly active users — people who open the app at least once per month). Altman confirmed the figure publicly. Sensor Tower also characterised this as the fastest any app has ever reached 1 billion MAU — roughly three years post-launch, compared with approximately six years for Instagram and five for TikTok.¹

So far, so milestone. Here is the part that matters for capital allocation.

Claude's year-over-year MAU growth rate. Claude, Anthropic's consumer AI product, sits at approximately 56 million MAU as of May 2026, growing at roughly 640% year on year.² ChatGPT, by contrast, grew approximately 62% over the same period. The absolute gap is 18

. The growth-rate gap is approximately 10 in the other direction.

For anyone pricing Anthropic's confidential S-1 (IPO registration, filed under seal), that growth-rate gap is the load-bearing number. A 640% growth rate on a 56M base is the argument for a high forward multiple. Investors do not buy the current snapshot; they buy the trajectory. The question is how far Claude can compound before the rate normalises, which is what every expansion-phase growth story turns on.

The Sensor Tower number is a floor, not a ceiling. This point is structural and it matters for how you read the competitive picture. Sensor Tower measures mobile app installs and in-app engagement. It does not capture ChatGPT web sessions, ChatGPT Enterprise (sold per-seat to organisations), ChatGPT Teams, or API usage. OpenAI's enterprise and API channels are material revenue contributors; their users do not show up in this 1 billion figure.

For a capital-flows analyst, the implication is straightforward: the 1 billion MAU is a floor on OpenAI's actual platform footprint, probably a significant undercount. But it also cannot be directly compared with, say, Google Search's daily active user base, because the measurement methodologies are entirely different. The headline is not wrong; it is simply incomplete in a specific direction.

The 5% time-displacement figure. Sensor Tower's panel data shows that US users who installed Claude in Q1 2026 spent approximately 5% less time on ChatGPT one month later.¹ This is the first publicly quantified consumer-side substitution rate between two frontier AI labs, and the number is small.

Five percent is not a haemorrhage. The SaaS apocalypse frame (per-seat compression as competing products eat into installed bases) would predict more aggressive substitution if the products were genuinely interchangeable at the consumer level. The 5% figure suggests one of three things: product differentiation is working; switching costs are real; or Claude's 56M user base is simply too small relative to ChatGPT's 1B to move the aggregate needle yet. I'd expect the third explanation is doing most of the work. Worth watching whether this figure compounds as Claude's absolute base grows.

A methodological note on the displacement figure: it is a correlational panel estimate, not a controlled experiment. Users who install Claude may already be higher-engagement AI users who would have intensified ChatGPT usage regardless. Causal inference requires caution.

Why the MAU metric is structurally different here than it was for social platforms. For Instagram or TikTok, MAU proxied engagement and therefore advertising inventory. A monthly active user was a potential ad impression, and the revenue model was roughly proportional to the base. ChatGPT's monetisation is fragmented: free tier (ad-free), Plus subscriptions at $20 per month, Teams, Enterprise, and API. A user who opens the app once a month to ask one question counts as an MAU under standard methodology. The revenue per MAU dispersion across these tiers is likely enormous, and without cohort data on subscription conversion rates or session depth, 1 billion MAU tells you about distribution reach, not about the economic value of that distribution.

OpenAI's revenue is increasingly enterprise and API-driven. The consumer MAU milestone, while real, may be a weaker predictor of OpenAI's financial trajectory than enterprise ARR (annual recurring revenue — the run-rate of subscription and contract revenue) or API consumption metrics, neither of which were disclosed in this announcement.

The IPO-season context. Both OpenAI and Anthropic are in or approaching public-market processes. Anthropic has filed a confidential S-1; OpenAI's IPO timeline remains publicly unconfirmed. If both clear into public markets within the same window, investors will be pricing two frontier-lab consumer platforms with no directly comparable public peers. Standard comparable-company analysis (finding publicly traded companies with similar business models and applying valuation multiples) breaks down when the comparables don't exist.

Valuation in that context gets driven by three arguments: growth-rate trajectories, inference-economics margin projections (whether the labs can close the gap between what it costs to run a query and what they charge for it), and platform-lock-in narratives. The 1 billion MAU confirmation is partly OpenAI managing the narrative entering that window. A 10-figure user base is a real competitive moat argument, even if the revenue-per-user economics are unclear.

What to watch. Three things in the next two quarters.

First, whether Anthropic's S-1 discloses revenue alongside MAU. The growth rate is the valuation argument, but revenue per user is the reality check. If Claude's ARR growth is tracking the MAU growth, the multiple is defensible. If it is lagging, the 640% figure is a distribution story, not a monetisation story.

Second, whether the 5% ChatGPT displacement figure compounds. As Claude approaches 100 million MAU, a stable 5% displacement rate starts to become visible in aggregate. If the rate grows, that is evidence for real substitution. If it stays flat or shrinks, product differentiation or switching costs are the better explanation.

Third, whether OpenAI discloses enterprise seat counts or API revenue alongside the consumer MAU milestone. The 1 billion consumer figure is impressive; what I actually want to know is what share of OpenAI's reported ~$10 billion annualised revenue is coming from consumer Plus subscriptions versus enterprise contracts versus API. That breakdown tells you whether the 1 billion MAU translates into durable ARR or whether it is a large but economically thin distribution base sitting on top of a B2B revenue engine.

The milestone is real. The interesting analysis starts where the press release ends.

---

Glossary

MAU Monthly active users; people who open a product at least once in a calendar month.

S-1 The SEC registration document a company files before an IPO; a confidential S-1 is filed under seal before public disclosure.

ARR Annual recurring revenue; the annualised run-rate of subscription and contract revenue.

Forward multiple A valuation expressed as a ratio of current price to projected future earnings or revenue.

S-curve The characteristic growth pattern for technology adoption: slow start, steep acceleration, then plateau as market saturation approaches.

SaaS apocalypse The compression of per-seat software revenue as AI agents substitute for human users of productivity tools.

Inference economics The cost of running AI models in production, as distinct from the one-time cost of training them; the binding constraint on AI platform margins.

Comparable-company analysis A valuation method that applies revenue or earnings multiples from similar publicly traded companies.

---

Footnotes

Footnotes

1. Reuters, "ChatGPT app hits 1 billion monthly active users in record time," Reuters Technology, https://www.reuters.com/technology/chatgpt-app-hits-1-billion-monthly-active-users-record-time-data-shows-2026-06-02, June 2, 2026. Source for the Sensor Tower methodology note, the fastest-app characterisation, and the 5% time-displacement figure. ↩ ↩²

2. AI

   , "ChatGPT Hits 1 Billion Monthly Active Users, Fastest App Ever," https://aiproductivity.ai/news/date/2026-06-03, June 3, 2026. Aggregates Altman's public confirmation and the Sensor Tower growth-rate figures for both ChatGPT (62% YoY) and Claude (640% YoY, 56M MAU). ↩

¶