Alexandr Wang: From MIT Dropout to Meta's Chief AI Officer

Scale's early years had recurring labour and platform disputes documented in coverage by The Verge, The Washington Post, and Time. The company relied heavily on contract annotators in the US and globally; questions about compensation, working conditions, and the labour rights of gig-economy annotators recurred throughout Scale's history. The company has publicly addressed these concerns in various ways over the years.

Commercially, Scale weathered the periodic AV-market downturns that eroded some of its earliest customer base. It also navigated a shift from being primarily a labelling vendor to being a data-and-evaluation partner for frontier LLM labs — a business-model migration that many observers thought would be difficult and that the company executed on.

The Meta transaction itself is a form of struggle-conclusion: Scale remains independent in structure but with Meta as a 49% owner and Wang effectively transferred to Meta leadership. Whether Scale continues to serve its non-Meta customers with the same neutrality is a question the industry continues to watch.

Two breakthroughs stand out. The first was demonstrating that a data-infrastructure business could be venture-scale — Scale disproved the industry belief that data labelling was too low-margin to build a fast-growth company around. The second was becoming an indispensable partner to the frontier LLM labs. When OpenAI, Anthropic, Google, and Meta each needed massive volumes of high-quality preference-labelled data for RLHF, Scale was among the small number of vendors capable of producing it at the required scale and quality.

That partner status is what made Scale strategically valuable enough for Meta's acquisition and Wang's appointment.

Two lessons from Wang's arc. One: picking a boring bottleneck can beat picking a glamorous problem. Scale's labelling wedge felt unglamorous in 2016 relative to the model-training work being written about; it became one of the most valuable businesses in AI because every model-training pipeline depended on it. Two: a founder-turned-executive-at-acquirer is an increasingly common exit pattern — the Wang transition to Meta CAO is one of several 2025 deals structured to retain the founder's operating capability rather than let it walk after a check clears.

## Sources

- TechCrunch coverage of Scale AI funding rounds — techcrunch.com - Forbes profile of Alexandr Wang — forbes.com - Bloomberg reporting on the Meta-Scale transaction, 2025 — bloomberg.com - The Information reporting on Meta's AI reorganisation — theinformation.com - The Verge, The Washington Post, Time reporting on annotator labour disputes - Scale AI company blog — scale.com