Tech giants start by disassembling systems

Chapter 85 Subtle Sensations

Wuxi. New District. A five-story building in an inconspicuous science park.

Microsensor Technology (Wuxi) Co., Ltd.

This company was founded in 2011 by Zhao Jiancheng. At 39, he held a PhD in Microelectronics from Nanjing University and worked at Bosch's Shanghai R&D center for over eight years, rising to the position of project manager. Ultimately, due to the glass ceiling effect for non-mainland Chinese citizens, and because he wanted to work in the MEMS industry in China rather than spend his entire career in a foreign company, he resigned to start his own business.

MicroSense primarily focuses on three things: Inertial Measurement Units (IMUs), magnetometers, and barometers. While its products have some customers in consumer electronics and automotive sectors, it receives very few orders from the drone industry. The reason is simple—the drone industry has extremely high requirements for the temperature drift stability and batch consistency of IMUs. Bosch can achieve this. ST can achieve this. TDK can achieve this. MicroSense—currently cannot.

Zhao Jiancheng was aware of this gap. He publicly admitted at an industry summit at the end of last year: "Our temperature drift stability is two orders of magnitude worse than Bosch's. Our batch consistency is also one order of magnitude worse. It will take at least three years to catch up."

Three years. Su Chen remembered this number very clearly. Because when he wrote "sensor" and "wall" on that blank sheet of paper in July, he had already included MicroSensor as one of the four domestic alternative suppliers.

Among the four companies, MicroSense ranks third. The advantages of the first two lie in their scale and number of customers. But MicroSense has something the other three don't—Zhao Jiancheng himself. He's someone who worked at Bosch for eight years. He knows the details of Bosch's processes. He also knows exactly where the gap between him and Bosch lies in the specific steps.

Su Chen examined Microsensor's products in the virtual teardown lab. The teardown results showed that the design approach for the MEMS microelectromechanical structure was excellent. However, the computational accuracy was not as high as Bosch's. The manufacturing process had significant weaknesses in three areas: temperature control accuracy in the deep hardening process, stability of wafer stress relief treatment, and humidity control during packaging.

There are three shortcomings. None of them are algorithmic problems; they are all process-related issues. Solving process-related problems relies on production experience, equipment investment, and time-consuming parameter adjustments.

Su Chen knew he couldn't help MicroSense solve these three shortcomings. The virtual disassembly lab could pinpoint problems, but it couldn't replace real-world debugging on the production line. However, if MicroSense could improve these three shortcomings to above the magnetic strength standard in specific scenarios—then it could become a "backup plan" in Hongyuan's sensor supply chain.

It's not a complete replacement for Bosch. Rather, it's a system that can be switched to immediately should Bosch cut off supplies or implement differential pricing.

At 9:00 a.m. on October 28, Su Chen walked into the front desk of Microsensor Semiconductor.

Zhao Jiancheng was waiting for him in the conference room on the third floor. The conference room was small, with only six chairs, a whiteboard, and a projector. The air conditioner was a bit noisy.

Zhao Jiancheng's first impression was that he was thin. He looked several years older than his peers. He had deep-set eyes and prominent forehead wrinkles. But his hands were dry and strong—the kind of hands that someone who spends most of their time working in cleanrooms handling jewelry and work clothes.

"President Su, I'm very surprised that you flew here in person." He spoke directly, without any formalities.

"I don't like discussing technical matters over the phone." Su Chen sat down.

What technology?

"Your IMU."

Zhao Jiancheng's eyes flickered slightly. Few people in the industry pay attention to micro-sensing IMUs. This was the first time someone from the drone industry had come to him.

"How much does Mr. Su know about our IMU?"

"I know quite a bit." Su Chen took a document out of his bag and placed it on the table.

Zhao Jiancheng picked it up and started reading. After quickly flipping through three pages, his expression changed.

"President Su, some of the data in this report—we haven't even completed the third round of testing ourselves. How did you obtain this data?"

"We have our own testing methods," Su Chen said without further explanation.

Zhao Jiancheng put the document down. He was silent for ten seconds. Then he looked up and stared directly at Su Chen.

What brings you here?

"I'm here because your design is excellent. Really excellent. The computational precision of your MEMS structures is comparable to Bosch's. However, three shortcomings in your manufacturing process are holding your product back—temperature control in deep hardening, wafer stress release, and encapsulation humidity. You yourself said at last year's summit that you would catch up with Bosch in three years."

Zhao Jiancheng's lips twitched. It's not a pleasant feeling to have someone remember so clearly what you said in public.

"President Su wants to help us shorten these three years?"

"I can't shorten the timeline. The technological issues can't be solved simply by contributing my technology as equity. But I can do something else."

"What?"

"We're giving you an order. We're giving you a chance to test your technology in a real drone flight environment. Your IMU has only been tested in the lab. It hasn't run for 10,000 hours on a real drone. Bosch's IMU has run for hundreds of millions of hours on tens of thousands of drones worldwide. Do you know why their temperature drift stability is better than yours? It's not because their design is more sophisticated. It's because they've fine-tuned it 10,000 times on the production line. How many times has your production line been fine-tuned?"

Zhao Jiancheng didn't speak. But Su Chen saw his eyes move.

"My proposal is simple. Starting from Q1 of next year, Hongyuan will allocate 5% of its production capacity from each batch of F4 and F5 aircraft to conduct real-world flight tests using MicroSense's IMU. All test data will be transmitted back to MicroSense. This is data that Bosch will never give you—because Bosch will not share its customers' flight data with a small factory on the mainland."

Zhao Jiancheng's eyes finally lit up.

In return, what you gave me was—

"Two things. First, over the next three years, Microsensing aims to achieve a performance score of at least 75 on the Bosch BMI088 for its drone-grade IMUs. We don't require you to completely match Bosch; 75 is enough. Second, when Hongyuan needs to switch suppliers, Microsensing will be able to ramp up production capacity within four weeks."

Zhao Jiancheng looked at him.

"What you need is a backup provider."

"It's not a backup. It's insurance. You've worked at Bosch for eight years. You should know better than me—sometimes Bosch isn't just a supplier. It can also be a wall."

Zhao Jiancheng slowly traced a line on the table. Su Chen had seen this gesture on many technicians before—it was a mental calculation.

"All test data sent back?"

"All. No withholding whatsoever."

"I need an on-site engineer to work on your production line."

"Welcome. We can give him a workstation on SMT line 2 at Zheng's OEM factory."

Zhao Jiancheng nodded slowly.

"Mr. Su, MicroSense has been established for eight years. Since leaving Bosch, I've visited various drone manufacturers every year to promote our products. Not a single drone company owner has ever personally flown to Wuxi to discuss technology with me. Nor has any drone company been willing to provide production capacity and data to help us with testing."

He stretched out his hand.

"You give us your orders and your application scenarios—and I'll give you a promise. Within three years, MicroSense's drone-grade IMU will achieve 75% of Bosch's performance."

Su Chen grasped his hand.

On the flight back to Shenzhen, Su Chen leaned back in his seat with his eyes closed. Outside the window was the night sky, thousands of meters above the ground.

He mentally analyzed the current situation.

The price war surrounding the S2 Pro will soon siphon off some entry-level F4 customers. This is unavoidable. However, the S2 Pro's flight controller is open-source, meaning its performance ceiling is already locked. Hongyuan's F5 will significantly outperform it in the second half of next year.

On the sensor side—Haotai's supply chain pressure won't take effect immediately. Bosch and TDK are factional companies, and they will weigh their options. The joint procurement volume after Lianchuang's entry will make them cautious. Meanwhile, the testing collaboration with Weigan will begin generating real data next year.

Three years. Zhao Jiancheng said three years. Su Chen felt that if the engineers at Weigan could obtain Hongyuan's actual flight data, this timeframe might be reduced to two years.

The wall is still there. But the ladder now has a second rung.

He closed his eyes and drifted off to sleep. There was still much to do after returning to Shenzhen tomorrow. But the direction was clear.