webnovelvip Chapter 233 Chains
Chapter 233 Chains
Halfway through the midday interview on Hunan TV's "Financial Observation" program, Su Xiaoying was well aware that the atmosphere on set was cooling down.
After He Zhaoming uttered the phrase "180 billion RMB," the director behind the dual-camera setup in the studio would occasionally remind her through her earpiece: "Push the topic up a bit."
Su Xiaoying turned the page of the question outline in her hand.
"Professor He," she said, "let's put that legal line of work aside for now. I have another question for you."
He Zhaoming nodded.
"If Qiming abandons the SPICE system at its core," Su Xiaoying said, "from the Berkeley version in 1973 onwards, Qiming will completely ignore all the evolutions over the past 53 years. Qiming will start from scratch and develop a completely new core algorithm system. Couldn't it bypass those four companies?"
The moment this question was raised, all the portal websites downstream of the studio that received the live broadcast signal simultaneously received a push notification with a priority indicator.
He Zhaoming did not answer immediately.
He gently placed his hands on the table.
"Yes," he said.
Another person sitting in the audience area of the studio—Lin Dawei, 39, the technical director of the Shanghai branch of a certain EDA company—was in a "seat of industry observers" temporarily invited by the program team. Hearing the word "can," he breathed a sigh of relief.
He thought that what Zhaoming was going to say was "Qiming can bypass it".
But what He Zhaoming said next sent a chill down his spine.
"Qiming is capable of developing its own," He Zhaoming said. "With Qiming's material budget of up to 15.8 billion yuan over the next eight years, with the State Council special funds currently in Weilan's hands, with the group of people behind them who came from Berkeley, Stanford, MIT, Shanghai Jiao Tong University, and Fudan University, and with the group of overseas scholars they have been able to bring back in the past five years, Qiming is fully capable of developing a completely new core algorithm system on its own. This is a fact."
He paused for a moment.
"But," he said, "can Qiming's self-developed algorithm read Synaptics' Sentaurus node library? Can it read Kentech's Virtuoso design library? Can it read Siemens EDA's Calibre rule files? Can it run Ansys' HFSS electromagnetic field calibration files?"
The studio was silent for a few seconds.
“These four things,” He Zhaoming said, “Qiming can’t read them.”
Lin Dawei gently put down the glass of water in front of him from the audience.
He came from an EDA background. As soon as he heard He Zhaoming mention those four names, he understood what He Zhaoming was going to say.
Sentaurus is Synopsys's flagship TCAD product, running over 90% of the world's advanced process nodes. Virtuoso is Kintex's flagship simulation design product, running over 90% of the world's top simulation design libraries. Calibre is Siemens's flagship EDA physical verification product, running over 90% of the world's chip fabrication inspections. HFSS is Ansys's flagship electromagnetic field simulation product.
Qiming's self-developed simulator cannot read the input format of any of these four components, and Qiming's customers cannot provide historical data from the past twenty years. Without historical data, any generation node that Qiming runs must start from scratch to re-accumulate process parameters. Rather than saying it cannot run, it's more accurate to say it needs to run for another twenty years.
Lin Dawei's hands got cold.
He Zhaoming noticed that some people in the audience below were changing their expressions. He didn't stop.
"That's not the worst part," he said. "Have you heard of the FabFour-EDA chain?"
Su Xiaoying put down her pen.
This name hadn't appeared in the preliminary survey. She was just about to press for details when He Zhaoming began speaking.
"FabFour comprises the top four of the six most advanced global wafer foundries," he said. "TSMC, Intel, Samsung, and GlobalFoundries. These four companies are not just about manufacturing. Their most important asset is the PDK—Process Design Kit. Over 90% of their PDKs use the input formats of Synopsys Sentaurus, Kentech Virtuoso, Siemens Calibre, and Ansys HFSS as unique delivery codes."
"Conversely, over 90% of the underlying optimizations for these four products prioritize the PDKs of these four OEM manufacturers."
"This is not just an agreement on paper. It is a two-way lock-in that has actually been going on for twenty years."
He picked up the teacup.
"This two-way locking mechanism has a name in the industry," he said. "The FabFour-EDA chain. Three years ago, a professor at Harvard Business School first included this name in his textbook."
"What do you mean by chains?" Su Xiaoying asked.
"That's what I mean," He Zhaoming said. "Any company that wants to be included in TSMC's Tier 1 list for the 3nm node must submit its design documents using Synaptics Sentaurus, Virtuoso, or Siemens Calibre. It's not a written requirement from TSMC. It's just that TSMC's PDK can only be used with these three products. If you bring something else that doesn't match, TSMC won't accept it."
He paused for a moment.
"Conversely, if you are Synaptics, Cadence, or Siemens, whose PDK should you follow most closely? Vilan's current production capacity cannot keep up, and SMIC, Huahong, and JCET's high-end nodes cannot keep up. You can only follow these four companies. If you follow them, your product's underlying optimization can only prioritize adapting to their PDK."
He gently placed the teacup back.
"The door is in the hands of four contract manufacturers," he said. "The key is in the hands of four EDA companies. The door and the key are not in the same room. Anyone who wants to come in from the outside must have something that neither of them has. This isn't one door. It's two doors."
The studio was silent for a full five seconds.
Su Xiaoying raised the next question.
"Professor He," she said, "could Qiming take a different approach? For example, by acquiring a company that previously held a Sentaurus-compatible read/write license?"
He Zhaoming nodded.
He glanced at the page of information in his hand. That page was the one he had drafted last night.
"Historically, there have been only six companies worldwide that have obtained Sentaurus-compatible read/write licenses," he said. "I'll list them one by one."
He raised his right hand.
"The first one is Mentor Graphics. They obtained their Sentaurus-compatible read/write license in 2002. However, Mentor was acquired by Siemens in 2017 and is now called Siemens EDA. They were already one of the four major players, and Qiming couldn't acquire them."
He raised his middle finger.
"The second one is Cadence itself. They hold the original manufacturer's license, the only one of its kind globally. They are also one of the four major players, and even Qiming couldn't buy them."
He raised his ring finger.
"The third company is Coventor. They started as a MEMS simulation company and obtained the interaction license for the MEMS part of Sentaurus in 2006. However, Coventor was acquired by Lam Research in 2017. Lam Research is the equipment supplier for FabFour. After Coventor was acquired, this interaction license was locked inside Lam. Qiming cannot obtain it."
He raised his little finger.
"The fourth company is IntelliSense. They are another company that does MEMS simulation and obtained some interaction licenses from Sentaurus. However, in 2015, IntelliSense was acquired by Coventor. Coventor was acquired by Lam in 2017. It was locked twice. Qiming couldn't get it either."
He raised his right hand again. He used all five fingers.
He lowered his right hand and raised his left.
"The fifth company is Comsol, a Swedish company. They only obtained a read license for the finite element portion of Sentaurus, not a write license. Furthermore, they are currently on the U.S. Department of Commerce's Entity List, preventing them from selling to mainland China."
He raised his left index finger.
"The sixth company is Magwel. It's a small Belgian company. They obtained a Sentaurus license in 1999, but in 2007, they were sued by Kenteng for patent infringement, which dragged on until 2010. In 2010, they won half the case, but the company's cash flow was dried up by the lawsuit, and it has not resumed operations since 2010. Now it's just a shell company with only three people."
He lowered his left hand.
"After studying six schools," he said, "only the last one, Magwel, was accepted by Qiming."
Su Xiaoying paused for half a second.
"After Qiming received Magwel," she asked, "could he read today's Sentaurus?"
He Zhaoming smiled.
That smile wasn't a sneer. This time, it was a smile of regret.
"The Sentaurus read license that Magwel has is the 1999 version," he said. "What node's PDK could the 1999 Sentaurus read? It could read 150 nanometers. What are the main nodes of FabFour today? They are 13 nanometers, 7 nanometers, and 5 nanometers the year after next."
He paused for a moment.
"From 150 nanometers to 7 nanometers, Sentaurus went through eight generations. None of the multi-quadrant elements, multi-physics couplings, and quantum correction terms added in these eight generations are mentioned in Magwel's reading permission."
"Qiming acquired Magwel," he said, "but we can only read the old nodes from fifteen years ago. Qiming is looking towards the next eight years, towards five nanometers and three nanometers. They can't take over the legacy Magwel had in 2008."
He picked up the teacup again.
"Host," he said, "what else can I say at this point?"
Su Xiaoying did not answer.
There was a silence in the studio for about three seconds. That was the longest pause in the entire program.
"Professor He," Su Xiaoying said, "I have one last question."
"Please speak."
"The Enlightenment Project," she said, "was it on the wrong track from the very beginning?"
He Zhaoming put the teacup back on the table.
He didn't reply immediately. Of all the times he'd ever been on a TV show, this was the most hurtful sentence he'd ever heard. He knew the host hadn't asked that question herself; it was coming through the earpiece.
If he had said his answer, "Qiming took the wrong path from the very beginning," he would have received a call from his university at eight o'clock the next morning. He understood.
However, the midday interview he presented today wasn't originally intended for Zhao Qiming. He wanted to make his voice heard by other people.
He looked up at the camera.
"Did Qiming take the wrong path at the beginning?" he said. "I won't answer that question. That's a question for the thing that will bear the name 'Qiming' eight years from now to answer itself."
He paused for a moment.
"What I'm giving you today isn't just this one answer," he said. "What I'm giving you today are six names, a chain, and a card. You can memorize the names, the chain, and the card. Once you've memorized them, those bosses who are still sitting in their offices watching the show after noon today can use these three things to get things done."
He gently placed his hand on the table.
"That phone call they made," he said, "wasn't to Qiming. It was to their own company's legal director."
After saying this, He Zhaoming put the microphone back on.
Su Xiaoying glanced at him.
She understood the true depth of He Zhaoming's interview that afternoon. She didn't ask any more questions.
The program started at midnight. Then, a 30-second time limit was reached, and the program ended in failure.
Just then, He Zhaoming heard the director say through his earpiece: "Professor He, we have urgently added a live connection for the last 56 minutes of this segment. It is with the technical director of the Shanghai branch of an EDA company."
He Zhaoming nodded.
He knew that this was only the beginning.
In the following fifty-six minutes, Lin Dawei, who had been brought in temporarily, mentioned a name fifty-one times.
FabFour-EDA chain.
That name, that afternoon, as that interview was cut into one short video after another—thirty seconds, three minutes, six minutes—was pushed to hundreds of millions of mobile phone screens.
By 8 p.m. that day, the phrase "FabFour-EDA Chains" had reached third place on Weibo's trending topics.
The first is Qiming.
The second question is, "Did Qiming take the wrong path at the beginning?"
The question He Zhaoming didn't answer was forwarded by millions of people that night, becoming the question everyone was asking.