“This is a Once-in-a-Civilization Opportunity”

Wang describes growing up in Los Alamos, working at Quora after high school, then attending MIT, where he trained early models. He says working at a company taught him how teams build and make decisions, while MIT gave him room to explore AI and develop the idea for Scale.

Wang’s first YC idea was an AI agent to help people get medical care, but the team changed direction after YC partner Jared Friedman questioned its prospects. While training models at MIT, Wang saw that compute and code were easy to access, but training data was not. He says that conviction led to Scale, even when investors doubted whether data would be a durable business.

Wang says founders need to identify important truths before they become consensus and build their own view of the future instead of following the crowd. He also stresses that founders start out inexperienced and improve by learning quickly. He describes AI as a major opportunity for startups, whose use of agents could help them compete with established companies.

Wang describes Meta’s goal of giving people personal AI that understands their context and helps them achieve more, alongside a larger ecosystem of businesses and agents. He says Meta rebuilt its frontier lab and launched Muse Spark within nine months, with talent density and a research-focused operating model as key priorities. He also says Meta plans more models, developer tools, and open-source work, and wants models to remain affordable and available across different harnesses.

Wang expects intelligence and agency to become abundant, making vision and ambition more important constraints on progress. He says builders must help businesses and governments adapt and address security risks, while also pursuing opportunities in science, health, biology, and creative work. For practical AI development, he points to agentic feedback loops optimized against clear metrics, and says rigorous systems thinking will remain useful as work shifts toward coordinating agents.