How Will OpenAI Compete?

Key Insights

1 — Four Fundamental Strategic Questions

Evans identifies four core problems for OpenAI: (1) No clear competitive advantage — models are interchangeable. (2) Experience, product, and value creation will change massively. (3) OpenAI must survive the “messy middle” without existing products and cash flows. (4) The product team doesn’t control its own roadmap — it reacts to whatever the labs deliver.

2 — 800 Million Users, a Mile Wide, an Inch Deep

ChatGPT has 800–900 million users — but only “Weekly Active.” 80% of users sent fewer than 1,000 messages in 2025 (averaging less than 3 prompts per day). Only 5% pay. US teens use it a few times a week or less. Evans’ diagnosis: if people only use AI a few times a week and can’t find a use for it most days, it hasn’t changed their lives.

3 — Chatbot = Browser Problem

The most provocative analogy: chatbots share the same differentiation problem as web browsers. Both are “just an input box and an output box.” For browsers, the last successful innovations were tabs and merging search with the URL bar. For chatbots: how many more buttons can you add? The apps all look the same — including ChatGPT.

4 — Distribution Beats Innovation

Google (Gemini) and Meta (Meta AI) are gaining market share through existing distribution, not better models. Anthropic’s Claude consistently ranks at the top of benchmarks but has no consumer strategy and virtually zero consumer awareness. When the product is undifferentiated, competition shifts to brand and distribution — and there, incumbents have structural advantages.

5 — The Netscape Lesson

Microsoft won the browser market — and it didn’t matter, because the actual experiences and value creation emerged elsewhere. Evans’ implication for AI: even if OpenAI dominates the chatbot market, the real value will lie in new experiences built on top of the models. The question isn’t who has the best model, but who invents the next products.

6 — Altman’s Strategy: Trading Paper for Position

Evans reads OpenAI’s activities over the past 12 months as an attempt to convert paper valuations into more durable strategic positions before the music stops. This includes fundraising, deals, partnerships — all aimed at becoming more than just a model provider.

Critical Assessment

What Holds Up

• The engagement data is sobering and empirically grounded — fewer than 3 prompts per day for 80% of users
• The browser comparison is the clearest articulation of the differentiation problem I’ve read to date
• The distribution thesis is confirmed by Gemini’s and Meta AI’s market share gains
• The analysis is product-strategic, not technical — exactly the perspective often missing from the AI debate

What Needs Context

• Evans writes from a consumer perspective. In the enterprise market (API, coding tools, agents), different dynamics apply — OpenAI is much better positioned there
• The browser analogy has limits: AI models are still improving rapidly, while browser rendering engines converged quickly. The window for differentiation through capability is still open
• The article ends at the most interesting point: “What do you do?” — the answer is missing (likely a follow-up post)
• Platform plays (Operator, Agents, GPT Store) are not addressed, even though they could be OpenAI’s best response to the distribution problem
• No mention of the API/developer side, which may be economically more significant for OpenAI than consumer

Discussion Questions for the Next Lab

01 Consumer vs. Enterprise: Evans analyzes the consumer market. Do we see the same interchangeability in our enterprise projects — or are there lock-in effects (integration, custom models, workflows)?

02 Design as Differentiation: If all chatbots look the same — is that a design problem? Can fundamentally different interaction paradigms (beyond chat) create differentiation? What would “tabs for AI” look like?

03 Lab Positioning: If value creation lies in new experiences built on models — isn’t that exactly our opportunity? Product and service design for AI-native experiences?

04 Engagement Gap as a Research Question: Why do 80% of users only use AI sporadically? Is it a capability problem, a UX problem, or an imagination problem? How would we investigate this as a Lab?

Sources

• Original: Benedict Evans — How Will OpenAI Compete?

Glossary

Moat A strategic competitive advantage that durably protects a company from competition — like a castle moat. Controversial for AI models: if models are interchangeable, the moat is absent.

Distribution Access to users through existing channels and products. Google reaches billions via search, Meta via social media — OpenAI has no comparable channel and must acquire users directly.

Engagement Curve A graphical representation of how intensively and frequently users engage with a product. A “flat” curve means: many users, but shallow usage depth — most use the product rarely and superficially.

Platform Play A strategic approach where a company opens its technology as a platform for others to build on. Examples: App Store (Apple), GPT Store (OpenAI). Goal: value creation through an ecosystem rather than a single product.