Hao Zhou Learned to Read Starlight. Then He Put the World in a Spreadsheet.

On the hottest nights of his childhood, Hao Zhou climbed to the roof. The family did not have air conditioning, so the adults talked in the open air while he looked up. The practical question was when the house might finally become cooler. The extravagant question was what, exactly, all those stars were doing. One was a household budget problem. The other was the universe. Zhou would eventually approach both in roughly the same manner: collect the signals, find the variables, build a model.

He chose astronomy at Peking University. During a placement at an observatory in Hebei, he gathered light from distant objects and helped turn it into a database. Light was no longer only beautiful. Properly measured, it was evidence: a record that could reveal the history of a star or galaxy. Zhou later called the idea romantic. The universe had sent information across an impossible distance, and mathematics made it legible.

That may be the cleanest explanation of his career. The settings changed dramatically - observatory, American bank, Chinese startup, Hong Kong exchange - but the gesture remained. Look at something noisy. Ask what it is quietly telling you.

An astrophysicist goes off syllabus

Zhou graduated from Peking University in 2002 and completed a physics doctorate at Rice University in 2008. His graduate work moved easily among disciplines. He wrote simulations in C++ and Python, studied statistical mechanics, built a time-series forecast for natural-gas demand and price, and programmed in Java and Fortran. The list reads less like a narrow academic apprenticeship than a collection of ways to persuade unruly information to disclose a pattern.

He even modeled his own job search. While classmates hurried between interviews, Zhou stayed in his room and searched Facebook and LinkedIn for employees whose backgrounds resembled his. He sent 20 messages asking for referrals. More than ten people answered. Three days later he was interviewing; within a month, he had a job at Capital One.

This was resourceful, faintly mischievous and revealing. Zhou had taken a ritual governed by nerves and networking, then treated it as a matching system. He joked that it was an upgraded back door. More precisely, it was a front door whose hinges he had quantified.

At Capital One, he worked as a senior statistician on business strategy. At Barclays, where he rose to vice president, he developed mathematical models for strategy and risk control. At Morgan Stanley, he performed quantitative analysis of investment products. Finance suited a physicist: chance could be priced, populations could be segmented, and a decision that felt personal at the counter could look probabilistic from inside the model.

The expensive lunch

The hinge in the story arrived over a meal. An older Chinese boss at Morgan Stanley reflected that China had become an unusually promising place to start a company. If he were 20 years younger, he said, he would go back. Zhou heard not nostalgia but a deadline. He was young. He could go.

Within a month, by his telling, he resigned and gave up the path toward a US green card. He returned to China, spent a year in an e-commerce company and founded the business that became QuantGroup in 2014. The choice was emotional; the preparation was quantitative. Years in American finance had taught him how data could mediate between risk, demand and capital. China presented a larger, less settled version of the matching problem.

I thought, I’m still young. I have to give it a try.Hao Zhou, recalling the decision to return to China

There was an important refusal at the beginning. Investors urged Zhou to build a peer-to-peer lending platform, then a fashionable route through Chinese fintech. He declined. High promised returns without an adequate capacity to judge or bear risk looked unsustainable to him, and regulatory arbitrage did not look like an enduring advantage. QuantGroup instead built data-led matching and risk capabilities. A physicist may enjoy a gamble; a risk modeler insists on knowing the odds.

The harder problem was that a model needs history, and a new company has very little. Zhou’s team had to find business while building the data that would make the next piece of business more intelligent. It was an awkward loop: no useful model without observations, no observations without customers, no customers without a useful model. The answer was less glamorous than the phrase “artificial intelligence.” They ran operations, watched outcomes, adjusted variables and accumulated evidence. The early advantage was not omniscience. It was a willingness to learn faster from each completed cycle.

Zhou also imported people who understood the method. One employee who had worked at Capital One said several former colleagues from the bank’s risk operation later joined QuantGroup. They shared a vocabulary of probabilities, controls and testing. A founder’s network can resemble a vanity collection; this one functioned more like a reused research team. The relationships mattered because the company was trying to combine two temperaments that do not always enjoy the same meeting: the caution of finance and the appetite of a startup.

The young company was first understood as fintech. Zhou resisted the neatness of that label. Data systems, he argued, could travel into other industries. Over time QuantGroup did exactly that, moving toward online consumer marketplaces and digital operating services. Its Yangxiaomie platform became the center of the business, while AI supported recommendation, distribution and operations.

That evolution also exposed a distinction Zhou had been making from the start. Data was not the product in the way a shirt, a loan or a car could be a product. It was the connective tissue: a means of deciding which offer should meet which person, how risk should be controlled and where an operation was leaking time. The company could change categories while preserving that layer. In founder mythology, consistency is often mistaken for never changing direction. Zhou’s version was more supple. Keep the instrument; point it at a different sky.

Lightning, in a hoodie

Inside QuantGroup, Zhou acquired a more useful title than CEO: Lightning. Colleagues said he moved quickly, disliked low-efficiency work and believed repetitive tasks should be handed to numbers and machines. The nickname suggests impatience, but the fuller portrait is stranger and warmer. Employees described a leader who was strict without being severe, demanding about the work and easy in conversation.

He sometimes sat in the open office in a hoodie, where new hires failed to recognize the founder. Before the 2018 World Cup final, he told the staff to take the following morning off so they could watch the match and rest. One employee turned up early anyway and found Zhou already there. A holiday, like a model, apparently worked better on other people.

These stories matter because founders often turn operating preferences into company folklore. Zhou’s dislike of wasted motion became both nickname and product thesis. QuantGroup’s job was to make matches more precisely: consumer to product, business to customer, signal to decision. Speed was not simply velocity. It was the absence of unnecessary steps.

Data is my faith, my life.Hao Zhou, 2018

A ticker symbol, then another horizon

In November 2025, QuantGroup listed on the Hong Kong Stock Exchange. The moment converted a private, evolving wager into a public company with quarterly scrutiny and a four-digit code. Zhou, then chairman, executive director and chief executive, had taken the business from a Chinese fintech opening through regulatory change, commercial reinvention and several attempts at the public market.

The listing is the obvious climax, which is why it is probably not the most revealing one. Zhou’s career is less about arriving than about transferring a method. Astronomy taught him that remote things leave traces. Banking taught him that uncertainty can be managed without pretending it disappears. Entrepreneurship taught him that a sound model can still meet a disorderly world.

By 2026, his language had moved again. QuantGroup adopted the phrase “intelligent species” for AI-enabled products and services, announced collaborations around large models, cloud technology and embodied intelligence, and established a physical-intelligence laboratory with Peking University. Zhou spoke of intelligent systems stepping away from the screen and into production and daily life. The ambition was wider than another shopping algorithm. It was intelligence with a body and a place in the physical world.

Awards accumulated along the route: Beijing talent recognitions, entrepreneurship lists, AI honors and Forbes China citations. They map the changing labels applied to him - returned scholar, technical leader, disruptive founder, industry figure. None fits as neatly as the childhood scene on the roof. There is Zhou, looking at a distant system, impatient for it to yield an answer.

Numbers, of course, do not abolish uncertainty. Zhou has acknowledged the remaining gaps: places where data cannot reach and problems it cannot solve. That concession keeps the faith from becoming dogma. A model is useful because reality exceeds it. The job is to improve the correspondence, not confuse the spreadsheet with the sky.

The boy on the roof wanted air conditioning and an explanation of the stars. The executive in Beijing wants machines to take on friction and intelligence to enter the everyday world. Between those desires lies an unusual continuity. Hao Zhou did not leave physics for finance and finance for technology. He kept following the light, changing only the instrument.