Zhipu CEO Zhang Peng: Inside the IPO of China’s First Publicly Listed LLM Company

In the interview 'Zhang Peng: Zhipu, a Trailblazer in AGI History,' Zhipu AI founder Zhang Peng recounts the journey from a university lab to the frontier of China's AI industry. The office building he sits in once housed fellow startups like Baichuan and Guangnian, but only Zhipu has endured. This seven-year saga encapsulates the bumpy road from the bottlenecks of last-generation AI to the pursuit of artificial general intelligence (AGI).

Around 2016, while the world celebrated computer vision surpassing humans, Zhang’s team at Tsinghua’s Knowledge Engineering Lab sensed the ceiling. They coined the term 'cognitive intelligence' – not just perceiving but understanding, reasoning, and creating. The idea germinated from an age-old intelligence studies question: could a machine predict tomorrow’s tech hotspots from millions of research papers? Their system, A-minor, did just that, counting Google and IBM as clients. It wasn’t mere search; it modeled the way human knowledge evolves, a genuine engineering of cognition.

A pivotal moment came in 2018 when China issued a policy allowing researchers to commercialize their findings. But details were absent. Zhipu became the first at Tsinghua’s computer science department to navigate this uncharted territory. For nearly two years, Zhang and his co-founders negotiated with the university over equity splits and IP valuations, inch by inch finding a path. 'It only opened a window,' Zhang recalls, 'not a wide-open door.' That painstaking process, however, set a precedent for many academic spin-offs that followed.

Outside the lab, the world was less accommodating. Early investors couldn’t grasp 'cognitive intelligence' and asked bluntly, 'Can this make money?' One even suggested halving the valuation during a downturn. A pundit likened Zhipu’s tech to concrete – solid and functional, but devoid of emotional appeal. The team survived on contract projects, barely breaking even, yet clung to the conviction that cognitive intelligence was the missing step toward AGI, even if the business model would come later.

Looking back, Zhang marvels at how six years compressed fifteen years’ worth of change. From perception to cognition, from paper to product, Zhipu’s story mirrors the struggles of Chinese deep-tech startups: the chasm between vision and venture capital, policy promise and real execution, academic purity and market survival. When asked how history books should remember Zhipu, Zhang hopes for a simple epithet: 'the one who first blazed the trail to AGI' – not because it has arrived, but because it had the courage to set out when the road was barely visible.