Why OpenAI Hired Back a Top Scientist to Sell AI to Big Business

Barret Zoph has returned to OpenAI to lead the company's sales effort to large enterprises — big banks, hospitals, government agencies, and other large organizations. Luke Metz, his co-founder at a startup called Thinking Machines Lab, rejoined OpenAI alongside him. The Verge reported the move on June 19, 2026.

Zoph's original departure from OpenAI in September 2024 was itself a significant moment. He had been directly involved in building OpenAI's most important AI systems: GPT-4, GPT-4V (which could understand images), GPT-4o (a faster, more practical version), and o1 (a newer, more capable model). He was one of a small number of people who had hands-on responsibility across multiple generations of the company's flagship technology. When he left, he and Metz started their own company before the two ultimately returned to OpenAI.

The new job represents a significant shift: from building AI systems to selling them. Selling artificial intelligence to large companies is a different skill than training AI models. It requires technical knowledge, the ability to support customers after they buy, and the kind of negotiating skill that turns a working tool into a long-term business contract.

OpenAI has been aggressively pursuing large corporate customers because competition is intensifying. Anthropic, Google DeepMind, and numerous well-funded startups are now competing for the same deals, which puts pressure on prices and speeds up the timeframe in which customers make decisions.

Putting a scientist like Zoph in charge of enterprise sales is unusual. Most companies put experienced sales executives in these roles — people whose entire career has been about closing deals and managing customer relationships. But the reasoning makes sense: at the level of customers OpenAI is chasing — Fortune 500 banks, major hospital systems, federal agencies — the decision to buy is as much a technical evaluation as a business one. A senior technical expert who can talk credibly with a customer's chief technology officer about how the AI actually works, not just what the marketing pitch says, can help close deals faster.

It is worth noting that the return of Zoph and Metz from Thinking Machines Lab to OpenAI is an unusual career path, and it raises a practical question: what happened to the startup itself. The Verge's reporting does not answer this, and public statements have not clarified what became of the company's work or any technology it developed as of June 19, 2026.

OpenAI has been rapidly expanding its business operation over the past year and a half. The company hired a former revenue executive from a major cloud computing company, grew its public sector sales team, and started offering customized contracts with specific service guarantees to enterprises. Zoph's appointment adds a research scientist's credibility to this effort — someone who can speak knowledgeably to customers about what the technology can actually do beneath the surface.

Zoph's career path at OpenAI before his departure tells a broader story about the company's own evolution. OpenAI started as a research organization focused on building better AI systems. Over time, it has shifted toward building practical, deployed products that real organizations can use. Each model generation Zoph worked on — GPT-4, GPT-4V, GPT-4o, o1 — was not just an improvement in raw capability; each one was designed to be deployed at a much larger scale, in real-world situations. In that sense, moving into enterprise sales continues a trajectory he was already on, rather than representing a complete departure from his technical roots.