2015, Cambridge, Massachusetts
Alexandr Wang stared into his dorm refrigerator, frustrated. The milk carton was empty—again. As an MIT freshman obsessed with machine learning, he’d rigged a camera to track his fridge contents but hit an invisible wall: the AI couldn’t recognize objects without massive amounts of labeled data. Meanwhile, his phone buzzed with another internship offer from Silicon Valley. He declined it. At 19, Wang had a more audacious plan: solving AI’s fundamental bottleneck—the lack of high-quality training data. Within a year, he’d drop out, recruit a fellow Quora engineer, and launch Scale AI from a garage. Their first office? A former nuclear weapons lab where his parents had worked. The symbolism was perfect: they were building a different kind of weapon—data for the AI revolution. This moment became the spark of the Scale AI story.

continue reading