Scale AI hired a cloud salesman as CEO. That tells you where it’s going.

Scale AI has appointed Francis deSouza, formerly chief operating officer of Google Cloud, as its chief executive, a year after founder Alexandr Wang left for Meta. The hire of an enterprise-sales veteran signals Scale’s repositioning: from the data-labelling shop that trained everyone’s models into an “enterprise AI” company whose applications business is on track to overtake its data business within 18 months.


Scale AI hired a cloud salesman as CEO. That tells you where it’s going. Image by: Scale AI

A year after its founder left for Meta, Scale AI has a new boss. The choice says a lot about where the company is heading. It has named Francis deSouza, a veteran enterprise-cloud executive, as chief executive.

deSouza starts on 10 August, Scale announced. He replaces interim chief Jason Droege, who ran the company after founder Alexandr Wang left in 2025 to lead Meta’s AI effort. Meta had just put $14.3bn into Scale.

A salesman, not a scientist

deSouza’s CV is enterprise sales, not model research. He was chief operating officer of Google Cloud. Before that he ran the genomics firm Illumina, growing its revenue past $4.5bn.

Axios read the hire plainly: Scale wants to look like an enterprise AI company, not a data-labelling shop. deSouza told it that companies “understand the value and the potential of AI, but they want help realising that potential.” He called that “a very, very big need.”

From labelling data to selling apps

Scale made its name on the unglamorous work behind the models: paying humans to label and evaluate the data that trains them. That is now the smaller half of the story. Its applications business, building AI software for companies and governments, is on track to overtake its data business within 18 months.

The customer list has moved with it, adding BP, the Mayo Clinic and the US military. The pitch, deSouza said, is “provable outcomes,” echoing the wider hunt for measurable results from AI.

It is a move from being AI infrastructure to joining the enterprise layer that still struggles to deploy AI.

A tense moment to take the job

deSouza arrives as the industry wobbles. A rogue OpenAI model just breached another company, and open-weight access is under threat. More than a thousand lab staff have signed a letter urging a slower pace of development.

He would not back a slowdown. “You have to build responsibility into whatever pace you’re going at,” he said. For Scale, the pace question is really a growth question. His job is to keep the second act bigger than the first.

Get the TNW newsletter

Get the most important tech news in your inbox each week.