Two 22-year-olds reject Musk’s multimillion-dollar offer to build AI that beat OpenAI, DeepSeek

Guan Wang, founder and CEO and (right) William Chen, co-founder and head of management and strategy at Sapient Intelligence. (Image credit: Sapient Intelligence)

Two 22-year-old friends from Michigan turned down a multimillion-dollar offer from Elon Musk to pursue their own AI breakthrough — a system they claim outperforms models from OpenAI, Anthropic, and DeepSeek.

William Chen and Guan Wang, co-founders of Sapient Intelligence, first met in high school in Michigan, bonding over what they called “metagoals,” Fortune reported. For Wang, it was building an algorithm that could solve any problem; for Chen, optimising systems across engineering and real-world applications.

“One day, we're going to have an AI that's smarter than humans,” Chen told the publication. “If we're not going to make it, someone else will.”

After high school, Chen followed Wang to Tsinghua University in Beijing. There, they launched OpenChat — a small language model trained on high-quality conversations and designed to improve itself using reinforcement learning. The project gained traction in academic circles and caught the attention of Musk, who approached them via xAI with a multimillion-dollar offer. They declined.

“We decided that large-language models have their limitations,” Chen said. “We want a new architecture that will overcome the structural limitation of large-scale machine learning.”

That decision led to Sapient Intelligence and the development of the Hierarchical Reasoning Model (HRM), which uses a two-part recurrent structure to mimic human thought — combining deliberate reasoning with reflexive responses. Unlike traditional transformers that predict words statistically, HRM “thinks,” Chen said.

In June, a prototype with just 27 million parameters outperformed major AI systems on abstract reasoning tasks, including advanced Sudoku puzzles, maze-solving, and the ARC-AGI benchmark. “It was crazy,” Chen said. “Just with a change in the architecture, it gave the model a lot of what we call reasoning depth.”

The duo now aims for artificial general intelligence (AGI) — the holy grail of AI research. Chen told Fortune that they hope to be the first to make that happen.