# ChatGPT Ads Achieve $1 Billion Revenue by Prioritizing User Trust > Discover how OpenAI's novel ad stack transformed ChatGPT, achieving $1 billion in revenue in under 200 days by prioritizing user intent and trust through advanced conversational understanding. **Source**: openai.com | **Published**: 2026-10-21 | **Type**: case_study ## Key Facts - ChatGPT Ads reached $1B in revenue within 200 days, showcasing rapid market acceptance. - Over 100,000 advertisers engaged, indicating strong competitive positioning in digital advertising. - Stateful ad sessions enhance user experience, revealing vulnerabilities in traditional ad systems. - Ads system's independence ensures ChatGPT's performance remains unaffected, reducing operational risk. - LLM-based orchestration allows for nuanced ad delivery, signaling a strategic shift in ad personalization. ## Summary \## Summary OpenAI developed a novel advertising stack for ChatGPT to enhance user trust and ad relevance. The challenge was to create an ad system that could understand evolving user intent across conversations, rather than relying on discrete search queries. The solution was a bespoke ads delivery system powered by a custom, post-trained large language model (LLM), achieving over $1 billion in annualized revenue run rate in less than 200 days after launch. \## Background OpenAI, a leader in artificial intelligence, operates in the technology sector and has developed ChatGPT, a conversational AI platform. Before the deployment of the new ad stack, OpenAI lacked an advertising system capable of integrating seamlessly with the conversational nature of ChatGPT, which posed challenges in delivering relevant ads based on user intent. \## Challenge The primary challenge was to create an advertising system that could interpret user intent that evolves over multiple turns in a conversation, rather than responding to isolated search queries. Traditional ad systems were not equipped to handle the nuanced and dynamic nature of conversations, which could lead to irrelevant ad placements and a degraded user experience. \## Solution OpenAI implemented a stateful ads delivery system that remembers user context across conversations. This system utilizes a custom, post-trained LLM to interpret conversation context, select, rank, and deliver ads. The ads selection process operates separately from ChatGPT's response generation, ensuring that ads do not influence the AI's answers. The system was designed to adapt to user intent dynamically, making tool calls for retrieval, ranking, and auction processes. \## Results The new ads system reached a remarkable milestone, achieving over $1 billion in annualized revenue run rate in less than 200 days following its launch. Since then, more than 100,000 advertisers have utilized the ad platform, demonstrating its rapid adoption and success in the market. \## Key Insights The deployment of the ads stack illustrates the importance of understanding user intent in a conversational context. By separating ad delivery from response generation, OpenAI maintained the integrity of the user experience while ensuring that ads were relevant and beneficial. This case highlights the value of iterative development and feedback loops in creating effective AI systems. \## Customer Testimonial No direct quote is available in the source material. ## Entities - **Companies**: OpenAI - **Products**: ChatGPT, Codex - **Technologies**: post-trained LLM, semantic embeddings, approximate-nearest-neighbor (ANN) search - **People**: Ritendra Datta, Karthik Gopalratnam ## Key Concepts user intent, stateful ads sessions, ads delivery system, orchestrator LLM, advertising principles, self-improving architecture, user trust, dynamic ad formats ## Definitions - **stateful ads sessions**: A system design that remembers user information across multiple turns of conversation to enhance ad relevance. - **orchestrator LLM**: A custom, post-trained language model that manages the ads flow by interpreting user intent and selecting appropriate ads. - **semantic embeddings**: Representations of stored meanings that facilitate the retrieval of ads based on user intent. - **approximate-nearest-neighbor (ANN) search**: An algorithm used for fast retrieval of ads by approximating the best matches from large semantic indices. - **self-improving architecture**: A system design that continuously learns and improves its performance based on user feedback and interactions. ## Use Cases - personalized ad delivery during conversations - dynamic ad formats allowing user interaction with businesses - improving ad relevance through user feedback - enhancing advertiser strategies with reporting data - maintaining user experience without ad interference - iterating on ad system performance based on detailed logs ## Frequently Asked Questions **How does the ads system ensure user trust?** The ads system is designed to prioritize user trust by maintaining a separation between ad delivery and ChatGPT's response generation. This ensures that ads do not influence the organic responses provided to users. **What is the role of the orchestrator LLM?** The orchestrator LLM interprets user conversations to identify intent and manages the entire ads flow, ensuring that the most relevant ads are selected and delivered based on the evolving context of the conversation. **How does the ads system handle user intent across multiple turns?** The system maintains stateful sessions that remember user information and context across multiple turns, allowing it to adapt and deliver ads that align with the user's evolving goals. **What happens if an ad is not delivered?** An ad may not be delivered if it is deemed inappropriate for the conversation context or if it does not arrive in time before ChatGPT's response is generated. The system is designed to prioritize user experience. **How does the system improve its ad delivery over time?** The ads system employs a self-improving architecture that gathers feedback from users and advertisers, allowing it to iterate and enhance ad relevance and personalization continuously. ## Links - [Read on Welcome.AI](https://welcome.ai/content/chatgpt-ads-achieve-1-billion-revenue-by-prioritizing-user-trust) - [Original source](https://openai.com/index/building-chatgpt-ad-stack) - [ChatGPT](https://welcome.ai/company/chatgpt): Featured company --- Source: Welcome.AI | https://welcome.ai/content/chatgpt-ads-achieve-1-billion-revenue-by-prioritizing-user-trust