"Chat is dead": OpenAI preps overhaul of ChatGPT

On June 8, 2026, a blockbuster report from the Financial Times shook the tech world: OpenAI is about to perform a radical "surgery" on ChatGPT. CEO Sam Altman stated bluntly in an internal meeting: "Chat is dead." — the traditional open-ended chat model has reached its end, to be replaced by a more structured, high-margin product system. This decision is widely interpreted as OpenAI's need to prove its sustained profitability to capital markets before its IPO sprint.

From "Universal Conversationalist" to "Precision Profit Engine"

According to sources, the new version of ChatGPT will no longer offer unlimited free conversations, but instead adopt a combination of "free trial + tiered subscription + pay-as-you-go" pricing. The basic tier will limit the number of conversation rounds and disable advanced features such as image generation and code execution; the Pro tier (approximately $40/month) retains the current experience; while the Enterprise tier is specifically designed for B-side customers, integrating CRM, ERP and other enterprise software interfaces, charged by API call volume. In addition, OpenAI plans to launch a "task-based Chatbot" module that allows users to customize dedicated assistants (e.g., customer service bots, data analysis assistants), billed by usage time. This is clearly a reference to the pricing strategies of Salesforce and Microsoft Copilot.

"OpenAI is mimicking the monetization logic of the SaaS industry, packaging conversational capabilities into quantifiable resource bundles." — Tech analyst Ben Thompson commented in his blog.

The Anxiety and Ambition Behind "Chat is Dead"

Altman's "Chat is dead" declaration is not unfounded. Although ChatGPT still had over 400 million monthly active users at the end of 2025, its revenue growth has slowed significantly — most users remain on the free tier, with a paid conversion rate of only about 8%. Meanwhile, Google's Gemini, Anthropic's Claude, and a range of open-source models are rapidly eating away market share. More critically, maintaining massive free dialogue services incurs high computing costs — according to estimates, each deep-reasoning conversation costs about $0.03, causing OpenAI's gross margin to remain below 60% for a long time, far from the 80%+ level of mature SaaS companies.

Therefore, this overhaul is internally dubbed "Project Phoenix." The core objectives are threefold: First, increase ARPU by limiting the free tier to push users toward paid subscriptions; Second, expand the enterprise market to support revenue scale with high-value contracts; Third, build a moat by leveraging the massive conversational data generated by users to train next-generation models, forming a data flywheel. According to projections, the new strategy could enable OpenAI to break even in fiscal year 2027, supporting its IPO valuation of up to $300 billion.

Editor’s Note: Farewell to the "Universal Wonder Tool" — The Inevitable Commercialization Path of AI

Looking back at ChatGPT’s three-year evolution — from a "phenomenal product" in 2023 to today’s "enterprise toolkit" — its trajectory mirrors the fate of most disruptive technologies: early stages rely on free access and surprise to ignite user growth, but sooner or later they face the challenge of commercialization and efficiency optimization. While OpenAI's transformation may hurt the feelings of some ordinary users, it is a rational choice for sustainable operations. However, risks also exist: overly aggressive charging may accelerate user churn, especially in the face of free competition from open-source communities (such as Meta’s Llama series). Moreover, "commoditizing" AI capabilities could spark a new round of ethical debates about AI fairness and the digital divide — when the smartest brains are packaged as subscription products, can the original intent of technology accessibility still be upheld?

At the same time, this move reveals the "second awakening" of the entire AI industry: large models are no longer content to be "chat toys"; they are embedding into enterprise processes, decision-making systems, and business loops. In the future, we may see more "ChatGPT-like" products splintering into free trial versions and professional editions, just as Microsoft Office moved from a perpetual license model to a subscription model. This is driven both by capital market pressures and the inevitable maturity of the technology.

Final Thoughts

In an internal letter to employees, Altman wrote: "We are killing a monster, and then creating a god." This sentence is dramatic, yet it speaks to the cruelty and rebirth of innovation. If "Project Phoenix" succeeds, ChatGPT will no longer be a chatbot but an AI-driven enterprise operating system. If it fails, it could become the first domino in the bursting of the AI bubble. Either way, this overhaul will be one of the most noteworthy tech business experiments of 2026.

This article is compiled from Ars Technica