Francis deSouza, the chief operating officer of Google Cloud and the president of its security products division, will become the chief executive of Scale AI on August 10, the company said in a statement Wednesday. The move transfers one of the most senior executives in the AI infrastructure business to a startup that supplies the data the industry runs on.
Scale AI is the largest independent provider of data labeling and annotation for artificial intelligence, the unglamorous work of preparing the datasets that train and evaluate models. The company has grown with the AI boom, and its customers include the frontier labs whose training runs depend on its workforce and software.
DeSouza brings a resume built for the job. He ran Cisco Systems as chief executive until 2023, where he managed one of the world’s largest enterprise technology businesses, and he joined Google in 2024 as Google Cloud’s chief operating officer, overseeing the cloud unit’s commercial operations and its security business. The combination of enterprise sales experience and cloud infrastructure depth is exactly what Scale AI’s board said it was looking for.
The appointment is the latest example of a pattern that has defined the AI industry’s management ranks: executives from the largest technology companies moving into the startups that are growing the fastest. Google has been a particular source of that talent, with executives moving to AI companies across the stack, and the flow has accelerated as the industry’s biggest names fight over a small pool of people who have run large organizations.
For Scale AI, the change comes at a moment of expansion. The company’s core business, labeling data for model training, has been joined by new lines of work as the industry has matured: evaluating model behavior, preparing specialized datasets for enterprise customers and, increasingly, providing the infrastructure for the human feedback that shapes how models behave. Each of those businesses needs enterprise sales discipline, and that is deSouza’s specialty.
The security side of his Google portfolio is relevant to Scale AI’s future as well. The company’s workforce, spread across offices around the world, handles sensitive data for the most valuable AI companies, and the regulatory environment for that work is tightening. Europe’s AI Act, which begins enforcement this month, and similar rules elsewhere put a premium on executives who have managed security at scale.
DeSouza’s predecessor is stepping aside at a high point for the company’s fortunes. Scale AI’s valuation has grown sharply with the AI boom, and its role in the ecosystem has shifted from vendor to critical supplier: when a frontier model fails a benchmark, Scale AI is often the company that labeled the test data. The move to a chief executive with enterprise credentials suggests the company is preparing for a stage where its customers are corporations and governments, not just research labs.
The appointment also says something about the state of the AI labor market. The industry’s biggest constraint is no longer chips or data centers, though both are scarce; it is people who have run large organizations and can take a startup from a fast-growing private company to a durable public one. Companies like Scale AI are paying for that experience, and the flow of executives from the giants to the startups shows where the industry believes the value is being created.
The company’s history is short but steep. Scale AI was founded in 2016, when the idea of a business dedicated to labeling training data was niche, and it grew into the default supplier for the frontier labs, with a workforce that spans software engineers and a global network of contractors. Its revenue has climbed with the industry’s appetite for data, and its role has expanded from labeling images to shaping how models are tested and aligned.
The transition will not be seamless. Scale AI’s founder, who built the company and its culture, is a different kind of leader from a corporate executive who has run public-company earnings calls. Employees and customers alike will watch whether deSouza preserves the speed that made the company valuable while adding the discipline that public markets demand. The board’s choice suggests it believes the second task is now the harder one.
For deSouza, the move is a return to the chief executive role he left at Cisco. For Scale AI, it is a bet that the next phase of the AI industry, the one in which models become products for businesses and governments, will be won by companies with enterprise discipline. The board’s choice of a cloud executive, rather than a researcher or a venture operator, is the clearest statement yet of where the company believes it is headed.
