OpenAI Bets on an 8-Dollar ChatGPT Go Plan to Grow Its Paying Base 37-Fold

OpenAI is facing a monetization crisis: internal forecasts predict that its flagship ChatGPT Plus plan will lose 80% of its subscribers by the end of 2026. The company plans to offset the churn with a new low-cost plan, ChatGPT Go, at $8 a month, and to grow its audience from 3 million to 112 million users. That is a 37-fold increase in a single year.

Why users are leaving Plus

ChatGPT Plus ($20 a month, or $240 a year) has proved too expensive for most users. AI has become part of the daily workflow, but for a broad audience its practical value rarely reaches the level of $20 a month, especially when competitors offer free alternatives.

Competitive pressure comes from several directions at once. Anthropic's Claude Free offers Sonnet at no charge. Gemini Pro is built into Google Workspace. Perplexity provides powerful AI search for free. Llama and other open models let users run AI locally with no subscription at all. In this environment, holding on to users at $20 is getting harder.

According to leaks from internal documents, OpenAI is seeing slower revenue growth and lower retention on the Plus plan. The forecast is bleak: 80% of the current base may leave the plan this year. This is not a temporary outflow but a systemic problem with price positioning.

ChatGPT Go: a bet on volume

OpenAI's answer is a sharp cut to the entry barrier. The new ChatGPT Go plan at $8 a month is aimed at users who want more than the free version but are not willing to pay $20:

• Current paying base: ~3 million people
• Target by the end of 2026: 112 million
• A 37-fold increase in one year
• Price comparable to a basic Spotify or Disney+ subscription
• Positioning: an AI subscription for everyday use

The logic is simple: even with a lower average bill, mass reach brings OpenAI more total revenue than a narrow premium audience. Plus, it yields usage data needed to improve the models.

The typical Go user is a freelancer, student or manager who uses ChatGPT a few times a week for mid-scale tasks: writing texts, analyzing data, translation, preparing for meetings. Not a power user, but not a one-off visitor either.

Risks of the strategy

A cheap, mass-market plan is a classic SaaS move, but in AI it is harder to pull off. OpenAI's infrastructure costs are enormous: GPT-5 inference is significantly more expensive than earlier versions. With heavy usage at $8, unit economics may turn negative, and the company will have to either impose strict limits or subsidize Go with revenue from pricier plans.

The second risk is cannibalization. If Go offers enough features, why upgrade to Plus ($20) or Pro ($200)? OpenAI has to draw the lines carefully: access to new models, context length, queue priority, API.

What this means

OpenAI is betting that an AI subscription should become as commonplace as streaming or cloud storage. If the numbers work out, by the end of 2026 ChatGPT will be the largest paid AI service in the world. If not, the company risks being left with an unprofitable base and without the premium buffer that covered the losses.