Scale AI appoints Francis deSouza CEO for enterprise AI push

Scale AI appointed Francis deSouza as CEO, effective August 10th, replacing interim CEO Jason Droege, as the company pushes beyond training-data services into enterprise AI applications. DeSouza, former COO of Google Cloud and president of Google Cloud Security Products, takes over after co-founder Alexandr Wang stepped down in June 2025 to lead Meta's AI efforts, following Meta's $14.3 billion investment for a 49% stake that valued Scale AI at over $29 billion.

Scale AI https://scale.com/about 's board appointed Francis deSouza @fdesouza https://x.com/fdesouza as CEO, effective August 10th, putting a veteran enterprise technology operator in charge as Scale AI expands beyond the training-data services that built its business. DeSouza will replace Jason Droege, who has served as interim CEO since June 2025. Scale AI said in its July 30th announcement https://scale.com/blog/scale-appoints-new-ceo that Droege will work with deSouza during the transition. Alexandr Wang @alexandr wang https://x.com/alexandr wang , Scale AI's co-founder and board chairman, welcomed deSouza in a post on X https://x.com/alexandr wang/status/2082869348121591816?s=46 , calling him the "right steward." Wang co-founded Scale AI in 2016 at age 19 while studying artificial intelligence at MIT, after encountering machine-learning infrastructure problems during his work at Quora. He also previously worked at Hudson River Trading and Addepar, according to Scale AI's founder biography https://scale.com/blog/author/alexandr-wang . Wang left the CEO role in June 2025 to lead Meta's AI efforts, while remaining on Scale AI's board. The move accompanied Meta's $14.3 billion investment for a 49% stake, according to TIME's reporting https://time.com/7294699/meta-scale-ai-data-industry/ . Scale AI described Meta's holding as a minority stake and said the transaction valued Scale AI at more than $29 billion. An operator for the applications business DeSouza joins Scale AI after serving as chief operating officer of Google Cloud and president of Google Cloud Security Products. He joined Google in January 2025, following executive roles as CEO of DNA-sequencing supplier Illumina and president of Symantec. He also brings a founder's record. DeSouza co-founded AI alignment venture SynthLabs, security software provider IMlogic and messaging startup Flash Communications. IMlogic was acquired by Symantec, while Microsoft acquired Flash Communications. He holds bachelor's and master's degrees in electrical engineering and computer science from MIT, according to his Google Cloud leadership biography https://www.googlecloudpresscorner.com/Francis-deSouza?pagetemplate=popup . That background matches the business Scale AI is trying to build. Scale AI began by supplying annotated data and human feedback for model training, then added model evaluations, red-teaming and full-stack systems for enterprises and governments. Scale AI now needs to sell implementation, security and measurable results to large organizations whose deployment problems extend beyond access to a capable model. "Organizations understand the value and the potential of AI, but they want help realizing that potential," deSouza told Axios https://www.axios.com/2026/07/30/scale-ai-google-cloud-coo-francis-desouza . He said Scale AI's data and applications operations can grow together as customers deploy AI across additional workflows. Scale AI told Axios that its applications business is on track to overtake its data business within 18 months. In a January review of its 2025 performance https://scale.com/blog/scales-next-era-building-for-2026 , Scale AI said applications revenue more than doubled during the second half of 2025 and projected that the operation would roughly double again in 2026. Scale AI also reported more than $1 billion in new business during 2025, nearly half booked in the fourth quarter. That figure represents new business rather than disclosed revenue or annual recurring revenue. Scale AI named Mayo Clinic, BP and Allianz among recent enterprise customers and said its applications operation serves public-sector customers in the United States and overseas. Scale AI's Meta problem remains DeSouza inherits a strategic complication created by Scale AI's proximity to Meta. Frontier laboratories share sensitive information about training methods, datasets and model weaknesses with data suppliers. Meta's large stake raised questions over whether rival laboratories would continue trusting Scale AI with that work. TIME reported in June 2025 that OpenAI and Google had begun winding down work with Scale AI after the Meta transaction. Scale AI rejected suggestions that the investment weakened its customer protections and said its commitments to data security remained unchanged. Scale AI later said its data operation grew during 2025, performed best in the second half of the year and became profitable. Those claims were not accompanied by revenue figures. Scale AI's current site lists more than 1,000 employees, 15 billion human decisions used for model training and $1 billion paid to contributors globally, all self-reported figures. The new CEO appointment makes Scale AI's response to that pressure explicit. Wang built Scale AI around the idea that data was a core pillar of AI development alongside computing power and algorithms, a thesis he outlined when Scale AI raised $1 billion at a $13.8 billion valuation https://scale.com/blog/scale-ai-series-f in May 2024. The round was led by Accel and included Y Combinator, Index Ventures, Founders Fund, Coatue, Thrive Capital, Nvidia, Amazon, Meta and other investors. DeSouza's mandate reaches further up the stack. Scale AI is asking enterprises and governments to hire it to build and operate AI systems, while continuing to supply the data and evaluations used by model developers. His tenure will be measured by whether Scale AI can turn that broader pitch into recurring deployment revenue without losing the neutrality that made Wang's original data business valuable.