UGB Chapter 211 Delivery
Chapter 211 Delivery
Chapter 245 Delivery
May 21st.
Two o'clock in the afternoon.
The computing center on the second floor of Vilan's Shanghai headquarters.
Su Chen sat in front of a dual-screen workstation. The left screen displayed the production line configuration parameter table sent by Bosch, while the right screen showed the input interface of the production line simulation system.
Bosch sent over the parameter list last night. It contains thirty-six items, just as Su Chen listed in the meeting. However, Bosch did some extra processing—two of the parameters were marked as "extra-domain," meaning "outside the verified parameter space."
This is a test.
Bosch wanted to see how Su Chen's simulation system would handle unverified parameters. Would it provide a seemingly definitive answer, or honestly label it "extrapolation"?
Upon seeing these two parameters, Su Chen didn't hesitate at all. He also labeled these two as "extra-domain" in the system's input interface.
This is a very important step in the project.
If a simulation system still outputs an "interpolated" confidence level when faced with unverified parameters, it is an dishonest system. A dishonest engineering system has no credibility in the industry.
However, if the simulation system actively marks the confidence level as "extrapolated", it means that the system knows the boundaries of its capabilities.
Su Chen pressed the start button.
A progress bar appeared on the screen. Estimated completion time: 5 hours and 42 minutes.
He leaned back in his chair and took a sip of water.
On a nearby machine, Gao Chaoyang was preparing a template for a delivery report. The report consisted of four parts: an overview of the input parameters, prediction results, confidence level assessment, and comparison with existing data.
"Old Su, how much space did they give us on the other side?" Gao Chaoyang asked.
Five percent.
What does 5% mean?
"Their internal model predicts a yield range of 68% to 73%. If our prediction deviates from their internal model by more than five percentage points, then our simulation system is considered 'not yet mature' in their eyes. If the deviation is within five percentage points, it is considered 'preliminarily reliable'. If it is within three percentage points, then it is 'worth further cooperation'."
How many do you estimate you can accomplish?
"One o'clock."
Gao Chaoyang paused. "That specific?"
"I have reviewed Bosch's 36 parameters. 34 of them fall within the validated domain, and two are marked as extra-domain by them. The prediction accuracy of the 34 parameters is between 1.2 and 2.0. I will provide the interval predictions for the two extra-domain parameters separately, so they will not drag down the overall accuracy. The overall accuracy is around 1.8."
Gao Chaoyang remained silent for a few seconds.
He wanted to ask, "How can you be so sure?" but he changed his mind.
He recalled a statement Su Chen made at the Bosch conference yesterday: "In principle, the system can predict yield ranges for any MEMS production line configuration."
That wasn't a declaration. That was a statement of facts regarding engineering.
"Okay. I'll go prepare the report template," Gao Chaoyang said.
……
8 PM.
Calculation complete.
The prediction results are displayed on the screen.
Median yield forecast: 70.2.
Yield prediction range: 68.4% to 71.9%.
The impact of the two extra-domain parameters: The expanded confidence interval is widened by an additional 0.8 percentage points. The system actively declares that the confidence level of this part of the prediction is lower than that of the interpolated part.
Su Chen exported the results and sent them to Gao Chaoyang.
After reading it, Gao Chaoyang swallowed hard.
Bosch's own internal forecast range is 68 to 73. The median is approximately 70.5.
Su Chen's prediction differed from Bosch's internal prediction by 0.3 percentage points.
It's not 1.8. It's 0.3.
"Old Su, you underestimated it," Gao Chaoyang said. "It's more accurate than you predicted."
Su Chen glanced at the data.
"1.8 is the upper limit of deviation for Bosch's internal model. 0.3 is the median deviation. Both figures are within a reasonable range."
He wasn't smug. He was simply stating a fact.
Gao Chaoyang understood. Su Chen knew the approximate accuracy range from the beginning—this wasn't the accuracy of the prediction, but the accuracy of the model. The accuracy of a third-order model in the validated domain was naturally around this order of magnitude.
He generated the delivery report. It was thirty-two pages long. It included an overview of the input parameters, visualization of the prediction results, the confidence assessment method, and a comparison with Bosch's internal model.
The report was sent to Bosch promptly at 8:00 a.m. on the 22nd.
It was 20 minutes ahead of the promised 48 hours.
……
May 22nd.
Stuttgart, 2 p.m.
Graf received Su Chen's report.
He spent three hours carefully reviewing the report. Then he did something that Bosch had never done before—he printed out the report and compared it page by page with Bosch's internal forecast report.
Page 32 to page 46.
Wei Lan's report was even shorter.
However, each prediction has a clearly defined confidence interval. Every extra-domain parameter is explicitly labeled. Every boundary of interpolation and extrapolation is clearly defined.
This is not an ordinary "we did it" engineering report. This is an engineering report that shows "we know what we can and cannot do".
In industry, the second type of report is far more valuable than the first.
Graf closed the report and went to find Stein.
When he knocked and went in, Stein was looking at a budget assessment for a 400mm production line.
"Have you read the report?" Stein looked up.
Graf nodded. "The median deviation is 0.3 percentage points."
Stein put down his pen.
"0.3," he repeated.
"Our internal forecast range is 68 to 73. Their median forecast is 70.2. Our median is 70.5."
"How do they handle the extra-domain tags?"
"Actively declare extended confidence intervals. Instead of giving definitive predictions for the two parameters that are outside the validation domain, a widened confidence interval is given, along with an explanation."
"In other words—" Stein paused for a moment, "their system knows its own boundaries."
"Yes. That's the key."
Stein stood up from his chair and walked to the window.
It was a late May afternoon in Stuttgart, and the sunlight wasn't too strong.
He stood by the window for a few seconds.
"Hans," he called Graf by name. "How do you think the current cooperation framework should be changed?"
Graf thought for a moment.
"The original plan was to begin strategic cooperation discussions after June. I suggest bringing it forward. If possible, we can start drafting the framework text next week."
"reason?"
"This system can do more than I expected. If we don't get involved in defining the collaboration framework in the early stages, the cost of joining later will be much higher."
Stein nodded.
"I agree. But moving it to next week won't work—Weilan won't keep up with the pace. Lin Wei will maintain the original schedule until June 1st."
Why?
"Because June 1st is the official online publication date for Nature Materials. She's waiting for that date. If she initiates strategic discussions before that date, she'll lose some bargaining power."
Graf pondered for a moment.
"So what can we do?"
"We can make internal preparations before June 1st. We can have the legal, technical, and procurement departments start assessments simultaneously, and strive to start framework negotiations within a week after June 1st."
"clear."
Graf turned to leave.
Stein called him back.
"Hans, there's one more thing."
"?"
"We need to be mentally prepared: this collaboration will not be led by Bosch. It will be a 'co-initiated' collaboration, with Villan likely leading the agenda."
Graf nodded. "I'm ready."
He left the office.
Stein stood alone by the window.
He was thinking about something.
He recalled the debate between Bosch and STMicroelectronics over a certain MEMS standard ten years ago. At that time, both Bosch and STMicroelectronics felt they were in control of the agenda because they both had production capacity and market share. But in the end, the standard was led by a professor from a German university—because he possessed the theoretical framework.
Whoever owns the theory owns the agenda.
Ten years later, this pattern remained unchanged.
……
May 22nd afternoon.
Shanghai.
Lin Wei received a reply from Bosch.
The reply consisted of only two lines:
"The simulation result has been reviewed internally. The deviation is within Bosch's acceptable margin. We propose to initiate the framework discussion immediately after June 1."
After reading the email, Lin Wei forwarded it to Su Chen, adding, "Bosch accepted it directly."
Su Chen replied with two words: "Okay."
Lin Wei then forwarded it to Jiang Mingyuan. Three minutes later, Jiang Mingyuan replied with a document—a list of key points for negotiating the strategic cooperation framework that he had prepared in advance.
The first line of the list reads: "Licensing fee pricing: one-time fee on a project basis. Suggested starting price: €28 million."
The second line reads: "Service fee: Annual subscription. Tiered based on the number of production lines of the partner. The recommended minimum subscription amount for the first year is 12 million euros."
The third line reads: "Verification Data Contribution Terms: For each production line verified, the partner can offset 12% of the annual service fee, up to a maximum of 40%."
The three lines of numbers combined far exceed the level of typical technology licensing.
Lin Wei read Jiang Mingyuan's list three times without making any changes.
She added a line at the end of the document:
"This is the initial offer. The bottom price will be adjusted after Bosch demonstrates the actual demand. However, it will not be lower than 70% of these three lines."
After sending it, she picked up her phone.
A new message notification appeared on the screen. It was from He Wentao.
"Mr. Lin, I have some bad news. Bertoli has made a move."
Lin Wei frowned slightly.
"What's that noise?"
"He published two editorials in two European industrial journals. One was titled 'The Distance Between Mathematical Models and Industrial Robustness,' and the other was titled 'Laboratory Precision Does Not Equal Production Line Precision.'"
Lin Wei remained silent for three seconds.
"He changed tracks."
"Change tracks?"
"He abandoned the 'slowdown strategy' and the 'alliance concept.' Now he's attacking robustness."
He Wentao paused for a moment on the other end of the phone. "You figured it out."
"He has no other cards to play. The academic criticism was crushed by NM. The supply chain pressure was thwarted by the packaging line and Qirui. The alliance concept has been left unresolved by us. The only thing he can do now is the 'real production environment' line."
"But his two articles weren't bad. He mentioned some specific industry issues—such as the impact of water quality in different regions on etching processes, and the impact of summer and winter temperature differences on equipment debugging. Can simulation systems directly handle these things?"
Lin Wei smiled.
"Yes, it can. But he doesn't care whether the simulation system can handle it. What he cares about is 'establishing doubts before anyone disproves them'."
How do you plan to respond?
"Don't respond yet. Let his article circulate for a few more days. Then let Su Chen respond in a specific way."
"In what way?"
"No response in magazines. No response in the media. Let the announcement of the cooperation framework for the Bosch 400mm production line be the response itself."
He Wentao paused for a moment on the other end of the phone.
"Mr. Lin, do you mean that Bosch should specifically mention 'real-world production environment verification' in their cooperation framework announcement?"
"Yes. Let Bosch say it himself: 'The simulation system has been validated under our real production line parameters.' Saying it from Bosch is more effective than Su Chen saying it a hundred times."
He Wentao let out a long sigh of relief on the other end of the phone.
"I understand. I'll stop following Bertolli's article for now and wait for Bosch's announcement."
"Don't stop tracking. Keep tracking. But don't rush to refute."
"it is good."
Lin Wei hung up the phone.
She picked up a pen from the table and wrote down three time points on a piece of white paper.
June 1st – NM was officially published online, and Kosuke Yamamoto's complete paper was released.
Early June – Negotiations on the Bosch cooperation framework were launched.
Mid-June – Initial communication with Infineon Weber.
These three points in time form an irreversible chain. Each point elevates Wei Lan's position.
Bertoli's article is simply a race against time.
He can't win.