Chapter 212 Two Legs of the Semiconductor Industry
"How to build more transistors within a limited area—this isn't just the design direction for CPUs; it's also the future path for memory devices. No need to explain that. What we really need to talk about is the structure of the MOSFET," Su Yuanshan said.
As soon as the discussion turned technical, a few more people gathered in the conference room. Su Yuanshan grabbed a sheet of draft paper and casually sketched the structure of a MOSFET.
"Wait, no, you do need to explain," Li Mingliu interrupted. "President Shan, don't gloss over it. Everyone knows that you can fit more rooms in a high-rise than in a single-story house. But chip design isn't like architecture. For buildings, you only worry about structural and material mechanics. As long as you're not building some weird shape, construction's usually no problem."
"But chip design? The difficulty is in the manufacturing process. I can draw a multi-layer structure right now—but getting it fabricated, that's the real challenge."
Su Yuanshan chuckled. "Well, leave process issues to the fab."
Li Mingliu clearly didn't believe it. "You make it sound so simple? If it were that easy, fab engineers would've beaten you up already."
"Fine, it's not that simple," Su Yuanshan laughed, setting his pen down and glancing around the room.
Everyone here was among the sharpest minds of their era, veterans in the industry. Of course they had thought of building upwards, "from houses to high-rises." But as Li Mingliu pointed out, going multi-layer wasn't just a design problem—it was overwhelmingly a manufacturing problem.
Building a single-story house? Grab three masons and you're good. Building a skyscraper? You need specialized management teams, hundreds of technical solutions. Just gathering a thousand workers won't get the job done.
To create a new architecture, you have to create new processes to realize it.
From trenching, stacking, P/N doping, etching, cleaning, ion implantation, oxidation, evaporation, vapor deposition... the full manufacturing chain could involve dozens of steps.
You might not need radical new technology every time, but you definitely needed redesigned processes—and that's without mentioning persistent issues like leakage currents in memory cells, or poor adhesion in polysilicon gates.
Engineering isn't like material science where you could get lucky—here, it was always about money, time, and enduring loneliness.
...
After a moment of silence, Su Yuanshan spoke slowly.
"Design and manufacturing are like two legs," he said. "No matter whether it's a breakthrough in new architecture or new processes, both legs are needed to move the whole industry forward."
Su Yuanshan pressed his lips together, carefully choosing his words. "But ever since large-scale integrated circuits first appeared, semiconductor progress has mostly been riding the wave of process breakthroughs. Just look at the shrinking process nodes and the exploding transistor counts."
"But process scaling will eventually hit a limit. In the foreseeable future, manufacturing advances will become much harder. When that happens, architectural innovation will be the only way to keep Moore's Law alive."
Images flashed through Su Yuanshan's mind: how, when lithography light sources stagnated at DUV 193nm, ideas like FINFETs and GAA structures emerged, and EUV technology was developed in parallel.
"Right now, we're not just behind in manufacturing. If we also fall behind in architecture... how are we supposed to compete in the future?" Su Yuanshan asked, looking around the room.
"You don't want to live forever selling VCDs and business units, do you?"
A heavy, indescribable emotion welled up among the crowd.
Previously, Yuanxin's main sources of revenue had been twofold: EDA and VCDs.
Later, Su Yuanshan sold off 20% of Xinghai shares to acquire Cyrix, using the proceeds to build the Yuanxin Technology Park. Then he sold most of the EDA business, pulling in another $2 billion-plus, which funded the Shanghai Tech Park, the Xinde International joint venture, and the Silicon Valley R&D center.
As for income from microcontrollers, YX CPUs, licensing fees, and commercial chips—those were mere pocket change.
Especially with the 91-series microcontroller: to force its way into the market, Yuanxin had set licensing fees absurdly low. Even though it broke into Europe, the revenue barely covered operating expenses.
Thus, in just two years, Yuanxin had almost single-handedly driven microcontroller prices to rock bottom—making it dramatically cheaper to produce everything from TV remote controls to radios and cassette players.
Science Daily had even published an article lavishly praising Yuanxin for lowering the cost structure of China's electronics industry—a political victory, at least.
But financially?
At present, nearly all of Yuanxin's salaries—over 5 billion yuan for more than 10,000 employees across factories and R&D centers—came from the VCD division.
Even Li Mingliu, hailed externally as the "father of VCDs," had to admit: VCD technology had no real technical depth.
For a supposed high-tech company to rely on a low-end consumer product for survival... was deeply shameful.
...
After the meeting, Su Yuanshan fed the pile of draft papers—covered in his brainstorming sketches—into the shredder.
"I'll stay here for the next few days," he said. "We'll organize our ideas carefully, then lock down the final direction. Once we decide, we'll bury our heads and grind."
Turning to Jiang Tao, he added, "Senior Brother Jiang, controller chips involve a lot of complex technologies. But once you master them, you'll be able to design almost any chip."
Jiang Tao nodded. "If we follow your direction, it sounds like the flash controller will eventually need a dedicated processor architecture like YX CPUs."
"Exactly," Su Yuanshan said. "Large-scale data throughput demands a high-speed processor. But for now, a basic microcontroller will do."
That evening, Su Yuanshan returned to the hotel.
Originally, Jiang Tao had wanted to lend him a laptop. But even at a major international hotel like the Hilton, there was no internet access yet—so it was pointless.
"Any important updates today?" Su Yuanshan asked as Zhou Xiaohui pulled his phone from her briefcase—she had taken it during the afternoon's technical meeting to prevent interruptions.
"Not much," Zhou Xiaohui replied. "Just that the Special Zone says the electronic post office system has passed internal testing."
"Oh?" Su Yuanshan straightened up, interested. "What's Senior Brother Ding saying?"
"He's planning to deploy a test server at the tech park tomorrow using Yuanxin's servers. If internal testing goes smoothly, they'll launch it to the market," Zhou Xiaohui said. "I already told him to coordinate with President Xi."
"Good," Su Yuanshan nodded. Suddenly he remembered something funny and asked, "Did he give the system a name yet?"
"He's calling it '163,'" Zhou Xiaohui said, blinking. "Old Ding said he found out that the Telecommunications Bureau's public dial-up network account number is 163. He figured he'd ride their coattails."
Zhou Xiaohui shrugged. "His exact words."
"Hahahahaha!"
Su Yuanshan laughed uproariously as his phone buzzed noisily in his hand.
Thank you for the support, friends. If you want to read more chapters in advance, go to my Patreon.
Read 20 Chapters In Advance: patreon.com/Albino1