Rebirth: Starting from Lighting Up the Tech Tree

Chapter 192 Reunion with the Stars

On the third day after returning from Shenzhen, Zuo Cheng received a message from Zhao Wenbo that the chairman of Xingchen Technology wanted to meet with him.

The message arrived via encrypted email, with the meeting agenda attached: a demonstration of the neural interface chip, details of technical cooperation, and a long-term strategic framework. Zuo Cheng forwarded the email to Yu Ying and Shen Yiming, instructing them to prepare.

Shen Yiming worked through the night to compile a technical integration plan, the core of which was the combination of federated learning and brain-computer interface signal decoding. Traditional decoding models require a large amount of personal data for training, posing extremely high privacy risks. Federated learning allows models to be trained collaboratively on distributed data, with the original data remaining locally.

After reviewing the plan, Zuo Cheng only changed one thing: he changed "suggest trying" to "core direction".

A week later, Zuo Cheng took Yu Ying, Shen Yiming, and Tang Ning to Shenzhen again. Xingchen Technology sent a car to pick them up; it was a black van with privacy film on the windows.

The car didn't go to the office building from last time, but instead drove to another industrial park in Nanshan District, to a plain-looking gray building. Zhao Wenbo was waiting in the lobby, with two security guards standing behind him.

Mr. Zuo, Dr. Yu, Dr. Shen, Researcher Tang, welcome. The Chairman is already waiting.

The elevator went straight to the third basement level, and the moment the doors opened, Tang Ning gasped. The space before her was at least three times larger than the floor above, with rows of glass-partitioned laboratories lining the white corridor.

Zhao Wenbo led them to the innermost conference room. The room was small, with a monitor on the wall displaying the words: Star Technology Neural Engineering Center.

Before the chairman arrived, Zhao Wenbo asked everyone to sit down. He said that today the chairman would be showcasing the third-generation neural interface chip, codenamed NX-30, which is currently the world's most advanced semi-invasive neural interface solution.

Zuo Cheng asked, "What does this have to do with the NX-07?"

Zhao Wenbo glanced at him and said, "Does President Zuo remember NX-07?"

Zuo Cheng said, "You guys used it for me two years ago. It's a patch-type device that you stick on your temple to collect signals."

Zhao Wenbo nodded and said, "The NX-07 is the first-generation product, and the NX-30 is the third-generation. The technical approaches are completely different, but the core materials are the same." He paused for a moment and said, "I'll explain in more detail when the chairman arrives."

Five minutes later, the conference room door opened, and a man in a gray Zhongshan suit walked in. He looked to be in his fifties, with gray hair, a thin face, and unusually sharp eyes.

"Zuo Cheng, I've heard so much about you." The man extended his hand. "I am Li Mingyang, founder of Xingchen Technology."

Zuo Cheng stood up, shook hands, and said, "Chairman Li, it's a pleasure to meet you."

Li Mingyang sat down and got straight to the point, saying, "I've reviewed the 402 documentation. Your capabilities in AI algorithms, especially your federated learning framework, are exactly what we need. The bottleneck for neural interfaces isn't in data acquisition, but in decoding. We've already achieved global leadership in data acquisition, but the accuracy of our decoding algorithms is stuck at 85% and can't go any higher."

Zuo Cheng said, "We've done some analysis. Traditional deep learning models require a lot of labeled data, but the individual differences in neural signals are huge. For the same action, different people have completely different brainwave patterns. The advantage of federated learning is that it allows the model to be trained collaboratively on data from multiple users, learning more general feature representations."

Li Mingyang's eyes lit up, and he said, "Continue."

Zuo Cheng gestured to Shen Yiming. Shen Yiming opened his laptop, connected it to the projector, and displayed the architecture diagram.

Each user's neural signal data is stored locally and not uploaded to the cloud. The cloud only receives model parameter updates, aggregates them using a federated averaging algorithm to generate a global model, and then distributes it to each device for the next round of training. The raw data never leaves the user's device.

Li Mingyang stared at the architecture diagram for a long time and said, "What about latency? Brain-computer interfaces require real-time decoding. Won't the communication overhead of federated learning be too high?"

Shen Yiming said that after model compression, only a few hundred KB of parameters need to be transmitted in a single communication, and the latency can be controlled within fifty milliseconds under a 5G network. This latency is acceptable for motion simulation applications.

Li Mingyang nodded, turned to Zuo Cheng, and asked, "How long do you need to make a prototype?"

Zuo Cheng said, "Three months. The prerequisite is that you open up the NX-30's interface protocol and data format."

Li Mingyang and Zhao Wenbo exchanged a glance and said, "Okay. But there's one condition."

Zuo Cheng said, "Please speak."

Li Mingyang stated that the patents for the results of this joint project are jointly owned by both parties, but Xingchen Technology has priority in using them. Furthermore, 402 cannot license this technology to any other neural interface company.

Zuo Cheng asked, "What about the deadline?"

Li Mingyang said, five years.

Zuo Cheng said, "Three years. After that, it can be licensed to other fields, but it cannot be used for competing neural interface products."

Li Mingyang laughed and said, "Zuo Cheng, you're more troublesome than I imagined. Okay, three years."

He stood up and said, "Next, I'll show you the actual NX-30."

They went to the adjacent laboratory, where a tray the size of a petri dish sat on the central workbench. Inside lay a black chip with fine metal contacts on its surface.

Li Mingyang put on gloves, picked up the chip with tweezers, and held it up to the light. The chip was as thin as a cicada's wing, with 1024 electrode contacts on its edge, each less than fifty micrometers in diameter.

This is the NX-30, with 1024 channels, a sampling rate of 30 kHz, and a power consumption of 50 milliwatts. It is implanted semi-invasively, placed outside the dura mater, without the need for craniotomy, and the surgery only takes 20 minutes.

Tang Ning asked, "What about rejection reaction?"

Li Mingyang said that the chip surface is coated with a biocompatible material called a neural affinity coating. Animal experiments showed that three months after implantation, there was no significant inflammatory response in the surrounding tissues. Human clinical trials are underway, and the current data are also promising.

Zuo Cheng asked, "What are the recruitment criteria for volunteers in clinical trials?"

Li Mingyang said that the first batch of patients were quadriplegia caused by spinal cord injury, with a disease course of more than one year and conventional rehabilitation proving ineffective. The safety and effectiveness need to be verified. Safety will be assessed by detecting rejection reactions and the stability of the equipment, while effectiveness will be assessed by determining whether patients can control the external equipment with their thoughts.

Zuo Cheng nodded and said, "If the validity verification is successful, what's the next step?"

Li Mingyang said the next step is exoskeleton control to help patients stand and walk again. The ultimate goal is to restore paralyzed patients to near-normal living abilities.

Zuo Cheng was silent for a moment, then said, "402 is willing to participate in this vision."

Li Mingyang put the chip back on the tray and said, "It's a pleasure working with you."

On the plane back to Hangzhou, Yu Ying leaned on Zuo Cheng's shoulder and said, "What do you think of Li Mingyang?"

Zuo Cheng said he's a tech fanatic, but he's got a plan. His willingness to open up the NX-30's interface shows that he really needs us.

Yu Ying said that he seemed surprised that you still remembered the NX-07 incident.

Zuo Cheng said, "Yes. The NX-07 was the first thing I encountered on the day I was reborn; it was the beginning of everything for me."

The plane pierced through the clouds, sunlight streaming through the porthole. Zuo Cheng gazed out, his mind already sketching out a roadmap. Three months for prototype development, six months for animal testing, twelve months for clinical trials. If all went well, the first paralyzed patient would be able to stand again using only their willpower in a year.

That will be the starting point of the eighth branch.