Two-Sided Markets

ChatGPT isn't just a product — it's a platform that connects two distinct groups: end users who chat with AI, and developers who build on the API. Understanding two-sided markets explains why AI platforms make the strategic decisions they do.

What Is a Two-Sided Market?

A two-sided market (also called a two-sided platform or multi-sided market) is a business that serves two interdependent groups of customers. The value for each group increases when the other group grows.

Classic examples:

Platform	Side A	Side B
Credit cards	Cardholders	Merchants
Uber	Riders	Drivers
App Store	Users	Developers
ChatGPT	End users	API developers

For ChatGPT, the two sides create a reinforcing cycle:

1. More end users → more data → better models → more developers want to build on the API
2. More developers → more apps powered by ChatGPT → more end users discover the platform

The Chicken-and-Egg Problem

Every two-sided market faces the same fundamental challenge: which side do you attract first?

• Without users, developers won't build on your platform
• Without developer apps, users have fewer reasons to choose your platform

OpenAI solved this by launching ChatGPT as a consumer product first, rapidly growing to 100 million users. Only then did they aggressively push the API to developers. This is the demand-side first strategy.

Google took the opposite approach with Gemini — they already had massive developer distribution through Google Cloud, Android, and Search. They went supply-side first, integrating AI into existing developer tools.

Simulating Market Dynamics

Adjust the sliders below to see how API pricing, subscription costs, and free tier availability affect supply and demand in the AI platform market.

Cross-Side Network Effects

The most powerful force in two-sided markets is cross-side network effects: growth on one side benefits the other side.

For ChatGPT's ecosystem:

• When OpenAI adds more users → developers get a larger potential customer base
• When developers build more ChatGPT plugins and GPTs → users get more reasons to stay on the platform

This is different from same-side network effects, where growth on one side benefits that same side (e.g., more users on a social network makes it more valuable for existing users).

Platform Pricing Strategy

Two-sided markets use asymmetric pricing — they often subsidize the more price-sensitive side to attract volume:

• Credit cards: Subsidize cardholders (rewards, cashback), charge merchants (interchange fees)
• App Store: Free for users to browse, charge developers 30% commission
• ChatGPT: Free tier for end users, charge developers per API token

The economic logic: the side that is harder to attract or more price-sensitive gets subsidized. The side that captures more value per transaction pays more.

Platform Envelopment

Once a platform establishes itself in one market, it can envelop adjacent markets by leveraging its existing user base. OpenAI has done this by expanding from:

1. Text generation → Image generation (DALL-E)
2. Text generation → Code execution (Code Interpreter)
3. Text generation → Web browsing
4. Text generation → App platform (GPT Store)

Each expansion increases the value of the Plus subscription and makes the platform stickier. In economic terms, this is economies of scope — the cost of offering multiple products together is lower than offering them separately.

The Multihoming Problem

Multihoming occurs when users participate in multiple competing platforms simultaneously. Most AI users today multihome — they use ChatGPT AND Claude AND Gemini.

Multihoming is bad for platforms because it:

• Reduces switching costs (users can easily leave)
• Splits data across platforms (less training signal)
• Forces price competition (users compare value constantly)

AI platforms fight multihoming by creating exclusive features: custom GPTs, memory features, and workflow integrations that make each platform uniquely valuable.

Key Takeaways

• ChatGPT operates as a two-sided market connecting end users and API developers
• Cross-side network effects create reinforcing growth cycles
• Platforms subsidize the more price-sensitive side to bootstrap the market
• Platform envelopment lets AI companies expand into adjacent markets
• Reducing multihoming through exclusive features is key to platform stickiness