Li Zhu

The short version of his resume is a recruiter's dream and a storyteller's trap. Tsinghua undergraduate. UC Berkeley PhD. Apple. Forbes 30 Under 30. Each line is true and each line hides the interesting part. So start instead with a single fact: before he was 30, the depth camera module he helped design had shipped inside roughly 100 million iPhones. That is the little black sensor that unlocks the phone when it recognizes your face. He led the team that made it manufacturable at a scale almost nobody on earth has matched.

Then he quit.

Not because the work was bad - because the work was finished, and because he had seen something. Apple had wrapped 3D structured-light sensing in a thicket of patents. Anyone who wanted to build a depth camera the same way would either pay Apple or lose in court. Zhu's read was contrarian and a little stubborn: if the front door is locked, the answer is not to knock louder. Build a different door. At Deptrum he set out to make a structured-light system that does not depend on the VCSEL-based recipe Apple had fenced off - and his team produced what is described as the world's first non-VCSEL structured light depth solution.

That phrase - non-VCSEL structured light - is the kind of thing that sounds like jargon until you realize it is the whole company. A VCSEL is a tiny laser that most depth cameras use to paint an invisible dot pattern on your face. Apple owned the best way to do it. Zhu's lab went back to the physics, leaned on nanophotonics - manipulating light at scales smaller than its own wavelength with a custom chip - and found another route to the same destination. Cheaper. Domestically sourced. And, crucially, his own.

It helps to know where the instinct came from. Zhu's doctoral work at Berkeley, under VCSEL pioneer Connie Chang-Hasnain, was not about cameras at all. He built an artificial chameleon skin - a flexible material that changes color when you bend it - and the optics community named it one of the breakthroughs of 2015. He also built the first flexible 1550nm-wavelength VCSEL chip. The throughline is light: how to bend it, shape it, and put it to work on surfaces and at scales where it usually refuses to cooperate. The chameleon skin was a magic trick. Deptrum is the same trick pointed at a market.

He has a line he comes back to, and it doubles as the company's operating system. Between innovation and cost, we always choose innovation. It sounds like a poster until you remember it is being said by a hardware founder in China, in a category where the usual move is to undercut on price. Zhu's bet is the opposite: that the cheap version of someone else's idea is a race to the bottom, and the durable version is a better idea that also happens to cost less because you own the chip underneath it.

The products carry the romance of the lab into the catalog. The depth cameras are named like an astronomy syllabus - Stellar for time-of-flight, Aurora for structured light, Nebula for a hybrid sToF approach - and then there is HandPass, a dual-mode palm-recognition camera for paying with your hand. They show up in places you would not think to look: under a phone screen, where Deptrum and ZTE co-developed a 3D camera that disappears beneath the display; inside a car's cabin, watching the driver with an automotive-grade ToF sensor; on a robot, giving it the spatial sense to not bump into the furniture; and at a checkout counter certified by China's Bank Card Test Center, reading the unique map of your palm.

There is a geography to the story too. Zhu grew up in Qidong, a city on the Jiangsu coast, and got to Tsinghua the hard way and the impressive way at once - a gold medal at the International Physics Olympiad that earned him direct admission. From there: Berkeley, Silicon Valley, Oracle, Apple. And then the reverse migration that defines his generation of Chinese hardware founders - from the valley back to Shenzhen, with the company later anchoring in Shanghai. He has called the move from Silicon Valley to Shenzhen the central arc of his thirties, and he has spent five-plus years arguing, mostly through products, that the frontier of 3D vision does not have to be drawn in Cupertino.

What makes him worth watching is not the credentials - it is the refusal to treat them as the point. He is a scientist who learned to ship, an academic who learned to manufacture at the scale of nine zeros, and a founder who keeps choosing the harder engineering road on the theory that it is the only one with a moat at the end. He frames it simply: in a field you genuinely care about, depth and persistence eventually manufacture their own luck. He has been depth-and-persistent about one narrow, strange, beautiful problem - how machines perceive the third dimension - for the better part of fifteen years.

The cameras are getting smaller. The chips are getting cheaper. And the man who could have spent his career as a brilliant footnote inside a trillion-dollar company instead decided the more interesting job was building the thing that competes with it.

The funding history tracks the conviction. Deptrum has raised through a Series B - a roughly 200 million RMB round led by CICC Capital, with strategic backers joining along the way - and total funding sits in the tens of millions of dollars. For a hardware company, that is not a number you spend on marketing. It is a number you spend on tape-outs, on clean rooms, on the unglamorous grind of getting a custom nanophotonic chip from a simulation into a part that yields. Zhu's pitch to investors is less a vision deck than a thesis about ownership: control the chip, control the cost curve, and you control whether the company is a feature supplier or a platform.

There is a quieter discipline underneath the bravado, too. For all the talk of out-engineering Apple, Zhu is careful about where 3D vision actually earns its keep. He has warned, more than once, against deploying the technology because it is impressive rather than because it is needed - find the real problem first, he says, then let the depth camera solve it. It is an unusual stance for someone whose whole company is a hammer looking for nails. It is also why Deptrum's deployments tend to be specific and load-bearing: a payment terminal that must not be fooled by a photograph, a car cabin that needs to know whether the driver's eyes are open, a robot that has to judge a doorway in real time.

Put it all together and you get a founder who is hard to file. Too academic to be a pure operator, too operational to be a pure scientist, too patient to chase the quarter and too restless to sit still inside someone else's roadmap. He has bet his thirties on the idea that the third dimension is worth obsessing over - and that the company which owns the physics of seeing it will matter long after the current cameras are forgotten.