Alexandr Wang

After building Scale AI into a central supplier of training data and model evaluation, Alexandr Wang moved in 2025 to lead Meta’s consolidated AI effort. The appointment placed a founder known for infrastructure and execution at the centre of an unusually consequential frontier-model rebuild.

Alexandr Wang co-founded Scale AI at 19, initially addressing the unglamorous but essential work of preparing data for machine-learning systems. The company later expanded into model evaluation, red-teaming and public-sector deployments. It says it supported early reinforcement-learning work, autonomous-vehicle development and national-security programmes; those claims are broad and should be treated as company-attributed rather than solely Wang’s personal output. His role changed decisively in June 2025, when Meta invested in Scale at a valuation above US$29 billion and Wang joined Meta. Scale said he would remain a board director while a new interim chief executive took operational control. In Meta’s July 2025 earnings remarks, its chief executive stated that Wang was leading the company’s overall AI team, subsequently identified with Meta Superintelligence Labs. The first major public model from the rebuilt organisation arrived in April 2026. Muse Spark was presented as a natively multimodal reasoning system with tool use and multi-agent orchestration. Meta attributed the model to the broader laboratory, not to Wang individually, and disclosed both benchmark results and evaluation-awareness concerns. Wang’s selection rests on the infrastructure institution he built and the scale of the new mandate he accepted, with the 2026 model treated as an organisational outcome rather than a personal invention.