109: Chapter 109 The Darkest Hour of Chips

Time returned to October 25th, Pingshan District, Shenzhen, the "flint one" chip manufacturing base.

This might be the cleanest place in China.

Entering the core area required passing through three air showers, wearing specially made anti-static cleanroom suits, and even breathing had to go through a specialized filtration system.

The number of dust particles in the air was controlled to under ten per cubic meter—a hundred times cleaner than an operating room.

Li Weimin stood before the lithography machine, looking up at the massive apparatus.

The 28nm lithography machine from Shanghai Micro Electronics Equipment was painted a deep blue, resembling a silent, colossal beast.

Its lenses were from Germany, its light source from Japan, and the control system was developed in-house by Spark Technology.

It was a "hybrid" device, but the core algorithms and integration were Chinese.

"Old Li, the wafer is ready," said young process engineer Xiao Liu softly.

Li Weimin nodded without speaking.

He was seventy-two this year; logically, he should be enjoying his retirement at home, but he chose to stand here.

Why?

Because twenty years ago, his teacher had held his hand on his deathbed and said: "Weimin, our country's semiconductor industry depends on your generation."

Twenty years had passed.

China had its own mobile phones, computers, and satellites, but the most critical chips still relied on imports.

Today, they were going to change this history.

"Begin," Li Weimin said.

The command was issued through the control system.

The robotic arm moved smoothly, placing a silicon wafer with a diameter of 300mm into the lithography machine.

The laser light source lit up, passing through the photomask to engrave tiny circuit patterns onto the wafer.

This process had to be repeated dozens of times, layer by layer, eventually integrating billions of transistors onto an area the size of a fingernail.

The entire workshop was so quiet one could hear one's own heartbeat.

Over fifty engineers were each performing their duties, staring at the data on the screens.

Zhang Wei stood beside Li Weimin, fists clenched, nails digging into his palms.

Su Chen had also arrived.

He was wearing a cleanroom suit, standing in front of the observation window, watching everything inside through the glass.

He didn't enter the core area—it required absolute concentration, and one extra person would increase the risk of contamination.

The tape-out process lasted six hours.

At four in the afternoon, the first wafer completed all processes and was sent out from the end of the production line.

Xiao Liu picked up the wafer with tweezers and placed it under the microscope.

His hands were shaking.

"How is it?" Zhang Wei asked.

Xiao Liu looked for a long time, his face gradually turning pale.

"The surface... has abnormal stains," his voice trembled. "It might be contamination."

Li Weimin pushed him aside and leaned into the microscope himself.

Sure enough, there were areas on the wafer surface with uneven colors, like the marks left after water stains dried.

"Continue testing," the old man's voice was steady.

The electrical test results were even worse.

The transistor threshold voltage drift exceeded 30%, meaning the chip's performance would drop significantly and power consumption would rise sharply.

Functional testing was the final hurdle.

The test machine loaded the preset program into the chip to check if the logic units could function normally.

The progress bar on the screen moved slowly: 10%... 20%... 50%...

Then it stuck.

An error code popped up: Large-scale failure of logic units.

Final yield calculation result: 0.7%.

The industry mass production standard was over 90%.

0.7% meant this batch of wafers was almost entirely scrapped.

The laboratory was deathly silent.

Someone began to sob softly; it was a twenty-five-year-old female engineer in charge of photoresist ratios. She was squatting on the ground, burying her head in her arms.

Zhang Wei punched the wall, producing a dull thud.

His hand was bleeding, but he couldn't feel the pain.

Li Weimin took off his reading glasses and rubbed his eyes.

He flipped through the scrapped wafers one by one, as if inspecting the remains of fallen soldiers.

"Where is the contamination source..." he muttered to himself, "Is it the air? Although our cleanliness is already top-tier... Is it the ultrapure water? Although we used seven-level filtration... Is it the photoresist? The formula was adjusted for three months..."

Su Chen walked in.

He first helped the weeping female engineer up and patted her shoulder: "Don't cry, failure is normal."

Then he walked to Li Weimin's side: "Old Li, you've worked hard. Rest first, we'll have an analysis meeting tomorrow."

"Su Chen, I..." Li Weimin wanted to say something, but was interrupted by Su Chen.

"It's fine. There's a saying in the semiconductor industry: tape-out failure is the norm, success is the exception." Su Chen's tone was terrifyingly calm. "Today is just the first time. Our budget already includes room for three tape-out failures."

He turned to everyone: "Everyone has worked hard, go back and rest. We'll have a failure analysis meeting tomorrow at nine in the morning. Remember—today's failure is not the end, it's the starting point. Because we now know that this path doesn't work, and we need to take a different one."

Before leaving, Su Chen said to the shift supervisor: "Keep a few of the scrapped wafers and encapsulate them. These are our 'Medals of Failure.' When we succeed, we will put them in a museum to tell future generations—success grows out of these failures."

But the news still leaked.

Early on October 26th, a certain tech self-media published an article: "Xinghuo Chip's First Tape-out Fails Miserably, Yield Less Than 1%! Is the Domestic chip Dream Shattered?"

The article was highly inflammatory, disclosing detailed failure data and including pictures—though it was later proven that the pictures were Photoshopped, no one doubted it at the time.

The article concluded: "Spark Technology spent 20 billion to build a factory, and the resulting chip yield is less than 1%. This is no longer a technical issue; it's a strategic misjudgment. Su Chen should wake up; chips are not like building cars, there are no shortcuts."

This caused a huge stir.

The comment section was quickly overrun:

"I said it before, chips are the jewel in the crown of industry; they aren't that easy to make."

"What's the use of Red Flag selling well? chips are the real moat."

"Suggest Spark focus on car manufacturing and leave chips to professionals."

"I bet China won't master 16nm within five years."

The stock market reacted quickly.

At the 10 AM opening, Spark Technology's stock price plummeted 8%, falling from 2200 yuan to 2024 yuan.

Analyst reports flew out like snowflakes, with the most pessimistic one stating: "Downgrading Spark's target price to 1500 yuan, chip business valuation to zero."

But there were also rational voices.

A researcher from the Institute of Microelectronics of the Chinese Academy of Sciences posted a long article on the Chinese microblogging platform: "As a researcher with thirty years in the field, I must say—first tape-out failure is the industry norm. TSMC's first-generation 16nm process had a yield of only 3% on its first tape-out. When Intel was doing 22nm back in the day, they failed seven times. If Spark had succeeded on the first try, that would be a miracle. Semiconductor is the art of trial and error; you must allow room for it."

There was a highly-upvoted answer on Zhihu, written by a former Intel process engineer: "Looking at the tape-out data, the problem is obvious—contamination led to threshold voltage drift. This is the most common problem in the early stages of factory construction, and the solution is clear: strengthen environmental control, adjust material formulas. Spark's team is very professional; I believe they can solve it. On the contrary, if there were no problems at all, that would be worth worrying about—it would mean they might have hidden the real data."

The most moving support came from ordinary users.

Old user Wang Jianguo, the taxi driver who spoke at the Red Flag delivery ceremony, posted on a stock forum: "I won't sell the Spark stock I bought. How many times did the Two Bombs and One Satellite project fail back then? We Chinese people fear failure the least. Su Chen, stay strong!"

This post was forwarded tens of thousands of times, and many people left comments below: "Support domestic products, support Spark," "Believe in Chinese engineers," "Failure is not scary, giving up is."

But negative voices still dominated.

Especially when some compared the success of Red Flag with the failure of the chip, the irony was even sharper:

"Spark is currently walking on one leg—the car business is booming, but the chip business is a miserable failure. Shouldn't Su Chen consider cutting the chip business and focus on building cars?"

"Facts have proven that Chinese companies are still suited for assembly and integration, not for underlying innovation."

"Suggest the state stop subsidizing Spark and give the money to more reliable companies."

October 27th, 9 AM, Spark chip base conference room.

The atmosphere was heavier than the day of the failed tape-out.

This time, it wasn't just the technical team, but also the PR, marketing, and investor relations departments.

Pressure was written on everyone's face.

Li Weimin spoke first.

The old man hadn't slept all night, his eyes bloodshot, but his voice was still strong.

"After forty-eight hours of analysis, we have preliminarily identified three reasons for the failure." He pulled up the PPT. "First, the ratio of photoresist to developer. Our formula was based on laboratory data, but there is a difference between the mass production environment and the laboratory, which needs adjustment."

"Second, fluctuations in workshop temperature and humidity. Although our environmental control system is the most advanced in the country, in the critical lithography steps, the temperature fluctuation exceeded the allowed range of plus or minus 0.1 degrees. This fluctuation leads to linewidth deviation."

"Third, and most importantly—" Li Weimin paused, "insufficient operational experience of the personnel. Our engineers are excellent, but most are operating the full tape-out process for the first time. Some details were not handled properly, leading to contamination."

He provided the solutions: "Adjust the formula; we already have three new plans and expect to complete verification within three days. Strengthen environmental control; we are already ordering more precise temperature control systems, to be installed within a week. Personnel training; starting today, all key personnel will be retrained and must pass an assessment before going on post."

"Time for the second tape-out?" Su Chen asked.

"Ten days later," Li Weimin said. "If everything goes smoothly, the yield is expected to increase to 30%."

30%, still far from the 90% mass production standard, but it was already a huge improvement.

Su Chen looked around the conference room: "Add a 500 million budget for equipment debugging and material replacement. Establish a 'Failure Analysis Committee' with Old Li as the chairman. Every failure must be traced to its source, a report formed, and studied by everyone."

He paused and looked at the head of the PR department: "Public response strategy: do not avoid, do not argue, let actions speak. Issue a short statement acknowledging that the first tape-out did not meet expectations, but the team is making full efforts to improve, and a second tape-out will happen in ten days. Do not explain technical details, do not promise specific results."

"Mr. Su, wouldn't that be too simple? The outside world expects a more detailed explanation..." the PR director hesitated.

"Explain what?" Su Chen retorted. "Explain why we failed? That would be providing ammunition for our opponents to attack us. Tell them what we improved? That would be leaking technical secrets. The rule of the semiconductor industry is—let the product speak, everything else is noise."

He stood up and walked to the window.

Outside the window was the sky of Shenzhen, cloudy today, very much like the mood at this moment.

"I know what people are saying outside. Let them talk. When we built Red Flag, people also said it wouldn't work, and now? The road of chips is ten times harder than cars, and failing a hundred times wouldn't be strange."

He turned around, looking at these engineers in the conference room with an average age of less than thirty-five:

"But I just need you to remember one thing: every failure brings us one step closer to success. Ten days later, we'll try again."

After the meeting, Su Chen kept Li Weimin behind.

"Old Li, tell me the truth—ten days later, are you really confident?"

Li Weimin was silent for a long time.

Finally, he said: "Su Chen, science has no hundred-percent certainty. But I can tell you—if the problem we found this time is indeed the root cause, then increasing the yield to 30% is possible. But if there are deeper problems we haven't discovered..."

"Then keep looking, keep trying," Su Chen said. "Old Li, I'm not trying to pressure you. I want to say—regardless of success or failure, Spark will support chip R&D to the end. This is not a business decision; it is a strategic choice. China can exist without Spark cars, but it cannot exist without autonomous chips."

Li Weimin's eyes moistened.

He held Su Chen's hand: "With those words from you, this old body of mine is worth it."

On the afternoon of October 27th, the official Spark statement was released:

[Statement Regarding the First Tape-out Situation]

Spark Technology's first 16nm chip tape-out did not meet the expected goals, and the team is comprehensively analyzing the reasons and actively improving.

Semiconductor R&D is a long and arduous process, and we have sufficient preparation and patience for this.

The second tape-out is expected to take place in ten days, and specific results will be announced in due course.

Thank you for your attention and support.

The statement was short, but the attitude was clear: no buck-passing, no playing the victim, no promises.

Public opinion split again.

Some praised Spark for being pragmatic, while others mocked them for being "stubborn."

But the capital market gave a positive reaction—the stock price rose 3% after the statement was released.

Late at night, Su Chen received an encrypted email from Lin Wan in Germany:

"Mr. Su, the Infineon negotiations have made a breakthrough. The other party agreed to sell the 'de-Americanized' 16nm production line core equipment and form a joint venture to build a factory. But they require us to provide material process support in exchange. An agreement has been reached in principle, and we will sign in three days. Additionally, I encountered an unidentified vehicle following me, but I have shaken them off. The security situation is under control."

Su Chen replied: "Stay safe, make the tracking incident public if necessary. Send the contract details back for review."

He closed his computer and walked to the window.

The night view of Shenzhen was always brilliant, but tonight, those lights looked like countless eyes watching him—expectant eyes, suspicious eyes, kind eyes, malicious eyes.

Red Flag succeeded, the chip failed, and Europe was playing a game.

This October, interwoven with ice and fire, was just like the current state of the country: some fields were rising, some were struggling, and some were breaking through.

But regardless, the road must continue.

Su Chen recalled a sentence his university teacher once said: "Living life is like walking on a dark road; when you can't see the distance, just watch every step under your feet."

Then walk step by step.

If the chip fails, analyze the cause and re-tape out.

If blocked in Europe, find a new partner.

If public opinion questions, respond with products.

He picked up his phone and sent a message to Zhou Qingyu: "On the Red Flag side, continue to do a good job with delivery and after-sales. Don't be distracted by the chip matter."

Soon, the reply came: "Understood. Mr. Su, believe in Old Li, believe in the team. What Red Flag can do, the chip can too."

Su Chen smiled.

Yes, what Red Flag could do, the chip could too.

It just needed more time, more patience, and more failures.

Outside the window, Shenzhen was a sea of lights.

And in this sea of lights, there was a building, a group of people, making the most difficult attempt for this country's technological destiny.

The night was still long.

But dawn would always come.