For most doctoral candidates, the dissertation is a stack of research data — defended, bound, and shelved. For Kao Yung-chung (高永中), chairman of IntelliEPI Technology, it was something you could plug in and run.
Kao's PhD project at UCLA was not a paper. It was a fully operational molecular beam epitaxy (MBE) machine — the kind of precision instrument that grows semiconductor crystals one atomic layer at a time inside a space-grade vacuum chamber. Building it from scratch, rather than operating an existing system, was a choice that defined everything that followed.
"My PhD thesis was building an MBE machine," he said, simply. "That was it."
Engineering from the Ground Up, Not the Manual Down
MBE systems in the 1980s were expensive, specialized, and treated by most research teams as tools to be operated rather than understood. Kao and his doctoral supervisor — Professor Wang Kang-lung (王康隆) of UCLA's electrical engineering department, now a member of Taiwan's Academia Sinica — took a different approach entirely.
Rather than focusing on what the machine could produce, they focused on why it worked: designing the vacuum chamber, material evaporation sources, heating elements, temperature control systems, and crystal growth parameters from scratch. Two fellow graduate students joined the effort. None of them had a finished machine to reference. Any miscalibration — in temperature, vacuum level, or material flux — meant a failed crystal and another iteration.
The experience trained a specific kind of problem-solving instinct. What Kao developed was not fluency with a particular machine, but a structural understanding of the relationship between equipment design, material behavior, and process outcome. When existing tools later failed to meet a new technical requirement, his first move was never to wait for the market to produce a better product. It was to figure out what needed to change and change it himself.
That instinct remains central to how IntelliEPI operates today. The company has invested in modifying and redesigning its own MBE equipment, including completing a 4-inch gallium nitride MBE system delivered to Taiwan's Semiconductor Research Institute (TSRI). In Kao's view, knowing what a material needs is inseparable from knowing what the tool that grows it must do.
Texas Instruments and the Lesson of Daily Success
Doctoral training teaches engineers how to make something work. Kao's eleven years at Texas Instruments, where he worked after completing his PhD, taught him that making it work once is almost beside the point.
"Succeeding once is easy," he said. "The real challenge is succeeding every day."
In an academic lab, a single clean result can establish proof of concept and carry a paper. In a production environment, that same logic is irrelevant. Equipment must run reliably across every batch and every wafer. Yield, delivery timelines, and customer specifications are not goals to be achieved on a good day — they are the baseline, maintained without interruption.
This manufacturing discipline reshaped how Kao evaluates new technology. Whether discussing quantum dot lasers, high-speed photodetectors, or IntelliEPI's own equipment development, he tends to address material uniformity, yield consistency, and customer qualification before turning to market potential. In his framework, a technology's real commercialization timeline is determined by engineering validation — not by investor sentiment.
Three Decades in One Domain, Compounding Across Every Cycle
Compound semiconductors have long sustained themselves on a phrase: "Tomorrow's Material." Every technology wave — fiber optics, smartphones, 4G, 5G, electric vehicles — seemed to promise the sector's moment, only for large-scale commercialization to arrive more slowly than expected.
Kao has watched every one of those cycles without changing course. While the dominant market applications for compound semiconductors shifted repeatedly, IntelliEPI continued accumulating capability in the same MBE domain: materials knowledge, equipment expertise, and process depth that persisted through each downturn. Each cycle, in his view, left the company with a stronger technical foundation than it had going in.
This consistency also shapes how Kao thinks about investment. Near the end of the interview, the conversation turned to a less expected topic: his personal portfolio.
More than twenty years ago, he bought shares of TSMC's American depositary receipts (TSM) on the US market and held them. The reasoning was not a short-term price call. It was that he understood the semiconductor industry well enough to believe that genuinely valuable technology would eventually be reflected in competitive position.
"There's no way I can watch those stocks every day," he said. "I only invest in industries I actually understand."
The same logic, he suggested, is how he has run IntelliEPI — not chasing cycles, but staying in the same domain long enough to accumulate something no single market trend can replicate. （Related： Art, Drones, and Non-Red Supply Chains: Inside Taiwan's Quiet Push Into European Markets ｜ Latest ）
