OpenAI has redrawn its organization chart for the third time in six months. The change, reported by WIRED and 36Kr, consolidates Codex, ChatGPT, and the API into a single product team now overseen by co-founder Greg Brockman, who keeps his role running the company’s AI infrastructure work.
The move concentrates power in the co-founder who has emerged as the steady hand inside a leadership structure that has churned repeatedly since 2024. OpenAI did not respond to a request for comment.
What changed
Under the new structure, the three biggest customer-facing businesses report into a single product organization. Codex, the coding agent that has become one of OpenAI’s fastest-growing products, previously ran with substantial independence, as did ChatGPT, the consumer assistant, and the API, the developer platform that powers a large share of the industry’s third-party AI applications. That independence created friction: separate roadmaps, separate pricing decisions, and, at times, competing priorities for the same pool of GPUs.
The consolidation is designed to end those fights. A unified product team means a single roadmap, a single pricing strategy, and one set of decisions about which features ship first. For developers who build on the API, the change matters because it ties the platform’s evolution to the consumer business – features that win inside ChatGPT are likely to flow to the API, and vice versa. For enterprise customers, it means one sales motion and one account team rather than three.
Brockman’s dual mandate is the notable part. Product strategy is a full-time job at a company of OpenAI’s size, and infrastructure is another. By giving both to one person, OpenAI is signaling that it sees the binding constraint on its business as the connection between what the models can do and whether the company can actually ship it to customers. The person who controls infrastructure also controls how much product promise can be delivered.
Brockman’s return to the center
Brockman’s path back to the center of the company has been winding. He co-founded OpenAI with Sam Altman and Ilya Sutskever in 2015 and served as president for years, running the research-to-product pipeline that produced GPT-3 and, later, ChatGPT. In August 2024, he stepped away from the board and took a sabbatical, a period that coincided with a wave of leadership departures that included several researchers who later founded or joined rival labs. He returned in early 2025 as senior vice president of AI infrastructure, taking over the job of securing chips, data centers, and the internal systems that keep ChatGPT running.
The new role extends that mandate upward. People who have worked with him describe a builder who cares less about titles than about systems, and the consolidation under his name appears designed to remove friction between the teams that make models and the teams that package them for customers. It also gives the company a clear public face for two of its most important functions at a time when investors and customers are asking who, exactly, is in charge. Internally, the move was announced with little fanfare, in keeping with a company that has learned to treat reorgs as routine maintenance rather than news.
Why OpenAI keeps redrawing the org chart
Three major restructurings in six months is a lot of movement for any company, and it reflects genuine pressure. OpenAI faces the most competitive field in its history: Anthropic is winning enterprise accounts with Claude, Google is bundling Gemini across its own distribution, and Meta and xAI are spending aggressively. Meanwhile, OpenAI’s own revenue is heavily concentrated – a small number of enterprise deals and a subscription base that must keep growing to justify the company’s valuation and its compute bill.
Organizational reshuffles at AI companies are rarely just about reporting lines. They are usually about who controls compute allocation, which is the true currency of the industry. Every team wants more GPUs, and the allocation decision determines which products move fastest. By uniting product and infrastructure under Brockman, OpenAI is effectively saying that its next stage of growth will be won by execution – shipping reliably, deploying efficiently, and turning research advantages into products that customers will actually pay for. People close to the company said the change also reflects a simpler problem: headcount has grown faster than management ranks, and the quickest way to get models out of research and into revenue is to put one executive over the whole pipeline.
The reorganization is a bet on Brockman and on operational focus. Consolidating Codex, ChatGPT, and the API into one product organization removes internal competition for resources, but it also concentrates enormous responsibility in one executive. If the unified team ships faster, OpenAI’s third restructuring of the half-year will look like the one that worked. If it does not, the next headline will be about a fourth.
