Tech Giants: Starting from Bankrupt Companies

111: Chapter 111 A Spark from a Spark Can Start a WildfireNovember 5th, 3:00 AM, Shenzhen Pingshan District ’flint one Factory’.

The lights in the cleanroom were as bright as day, but the atmosphere was as oppressive as the silence before a storm.

Li Weimin stood before the control console, hands braced against the edge, his knuckles white from the pressure.

He hadn’t closed his eyes for thirty-six consecutive hours; his sclera were bloodshot, and his eyeballs looked like two charcoal fires burning in deep wells.

The large screen opposite the console displayed real-time data streams, with green characters cascading down like a waterfall.

Zhang Wei sat in the operator’s chair, his fingers hovering above the keyboard, trembling slightly.

This former senior director who had survived the most rigorous technical tests at Intel now had palms slick with sweat.

There were over twenty other engineers in the workshop, all wearing white anti-static cleanroom suits, masks, and hoods, with only their eyes visible.

In those eyes lay the same tension, anticipation, and a hint of irrepressible fear.

Fear of failure, and even more, fear of letting others down.

The electronic clock on the wall jumped: 03:01.

’Begin,’ Li Weimin’s voice was as hoarse as sandpaper rubbing together.

Zhang Wei pressed the Enter key, and the test program started.

This was the second tape-out.

When the first one failed, the yield was 0.7%, almost a total wipeout.

In the ten days that followed, the entire team worked like mad—analyzing contamination sources, adjusting photoresist formulas, optimizing process flows, and retraining operators.

Li Weimin had lost weight, and Zhang Wei’s hair had turned gray in patches.

Now, it was time to verify the results.

Step one: Wafer visual inspection.

A robotic arm picked up a fresh wafer and fed it into a high-power microscope.

The image was magnified onto the screen—on the dark gray silicon wafer, dense circuit patterns were neatly arranged, like a miniature urban planning map.

’Surface is clean, no abnormal spots,’ the young female engineer in charge of visual inspection said, her voice trembling but her report clear. ’The first wafer is qualified.’

A slight sigh of relief echoed in the workshop.

Step two: Electrical testing.

The probe station lowered, and tiny probes made contact with the test points on the wafer.

Data began to jump: threshold voltage, leakage current, transconductance... all parameters were within the green zone.

’Transistor threshold voltage fluctuation is less than 5%,’ the test engineer’s voice rose a notch. ’Qualified!’

Step three: Functional testing.

This was the real test.

The test machine loaded the preset program into the chip, checking if the hundreds of millions of transistors could work together to execute the correct logic operations.

The progress bar on the screen began to move:

chip 1-100 testing...

Yield: 47%

Someone in the workshop clenched their fists.

chip 101-500 testing...

Yield: 52%

Someone started praying softly.

chip 501-1000 testing...

The progress bar crept slowly upward: 53%... 54%... 55%...

Finally, it settled at: 56.3%.

Time froze at that moment.

Three seconds of absolute silence swallowed all sound like a vacuum.

Then, the first sob broke the silence.

It was the young female engineer from visual inspection; she crouched on the floor, her shoulders shaking violently.

Then came the second, the third... The young engineers threw off their dust caps; some wept loudly, some looked up and roared, some embraced tightly.

Zhang Wei stood up from the operator’s chair, his body swaying slightly.

He walked toward Li Weimin, his steps unsteady, like a drunkard.

Reaching the old man, he opened his arms and hugged Li Weimin tightly.

’Old Li... we did it... we did it...’ Zhang Wei’s voice choked, tears streaming down his cheeks and dripping onto Li Weimin’s lab coat.

Seventy-two-year-old Li Weimin stood there, letting Zhang Wei hold him.

The old man’s body trembled slightly; after a long while, he finally raised his hand and gently patted Zhang Wei’s back.

’Good...’ he said only one word, his voice very soft, as if afraid to wake from a beautiful dream, ’Good...’

Then he cried too.

Turbid tears welled up from his deep-set eyes, flowing down his wrinkled cheeks.

This old Academician who had witnessed the birth of the Huaxia semiconductor industry from nothing had seen too many failures, too many disappointments, and too much helplessness of ’wait another ten years’.

And today, he had finally waited for it.

Su Chen walked in from the observation room.

He wasn’t wearing a cleanroom suit, but he had undergone strict air shower decontamination.

Watching the 56.3% figure on the screen, he closed his eyes and took a deep breath.

When he opened his eyes again, his gaze had returned to calmness.

’Immediately initiate three procedures,’ Su Chen’s voice was exceptionally clear in the excited workshop, like an order, and also like a sedative.

Everyone quieted down and looked at him.

’First, package one thousand sample chips. Before eight o’clock this morning, send them to the MIIT and the Huaxia Electronics Standardization Institute for authoritative certification.’

’Second, prepare for mass production—the 56.3% yield has reached the break-even point. Start the small-batch production process; the first batch will produce five thousand chips.’

’Third, lift the confidentiality, and prepare to hold a press conference.’

He paused and added: ’Put the scrapped wafers from the first tape-out and the successful wafers from this time in the same display case. Put a sign next to it that says—’

Su Chen looked at the big screen frozen at 56.3%:

’From 0.7% to 56.3%, it took us ten days.’

Applause erupted in the workshop.

At first sparse, then like a tide.

’For all participating teams, this month’s bonus will be tripled,’ Su Chen continued, ’Academician Li, Zhang Wei, and all core members... name your own reward requirements, as long as they are reasonable, I will approve them all.’

Li Weimin wiped his tears and shook his head: ’I don’t want a reward. I just want... to keep working. 16 nanometers is done, next there’s 14 nanometers, 7 nanometers... This old bag of bones of mine can still fight for another ten years.’

Zhang Wei laughed, his smile laced with tears: ’Old Li, you need to rest. Leave the rest to us young people.’

’Rest? What rest?’ Li Weimin glared, ’In this chip business, if you don’t advance, you retreat. We just caught up a little today, and tomorrow they’ll run ahead again. We can’t stop, not for a moment.’

Su Chen looked at the old man and the young man, complex emotions surging in his heart.

There was pride, emotion, and also heaviness—because he knew, this was truly just the beginning.

At four in the morning, the packaging workshop began working.

The qualified wafers were cut into thumbnail-sized chips and packaged in black plastic shells.

On every chip, a small flame logo was laser-etched—that was the Flint logo.

At six in the morning, the first batch of one hundred packaged chips was sent to the capital by a special vehicle.

At eight in the morning, the expert team from the MIIT was already waiting in the laboratory.

At ten in the morning, the preliminary certification report was released: Spark 16-nanometer ’flint no. 2’ chip, process maturity reached mass production standards, and performance parameters met design specifications.

November 6th, exactly 10:00 AM.

Spark Technology’s official website, Chinese microblogging platform, and WeChat official account updated simultaneously.

The title had only one line of text: ’Spark Technology 16nm chip ’flint no. 2’ tape-out successful, yield 56.3%’.

The main text was simple, with three attached images: a scan of the MIIT certification report, a comparison photo of the first and second tape-out wafers, and a performance parameter table.

The performance parameters were restrained, but those in the know could see the weight at a glance:

CPU core: Eight-core Flint architecture, main frequency 2.8 GHz

GPU core: Self-developed ’Yanhuang’ graphics processor, performance equivalent to ARM Mali-G72

Process technology: 16nm FD-SOI

Transistor density: approx 33 million/square millimeter

Power consumption: 15% lower than ARM chips of similar performance in typical scenarios

Features: Integrated self-developed NPU (Neural Processing Unit), AI computing power 4TOPS

The last sentence was: ’Accepting sample applications from corporate clients starting today.’

The first minute after posting, the number of reposts exceeded ten thousand.

The fifth minute, it reached number one on the Chinese microblogging platform trending list.

The tenth minute, the server started to lag.

At twelve noon, CCTV News Channel broadcast a flash: ’Our country has achieved a major breakthrough in the field of high-end chip manufacturing. Spark Technology’s self-developed 16nm chip tape-out was successful, with a yield reaching international mainstream levels...’

At 2:00 PM, the Huaxia Daily app pushed a headline: ”From Hardship to the Stars and Sea — Written on the Occasion of the Success of the Domestic 16nm chip.”

The article was long, reviewing the sixty-year arduous journey of Huaxia’s semiconductor industry, from the birth of the first transistor in 1956 to today’s 16nm chip.

The final paragraph read: ”We once thought that chips were unreachable stars. Today, Spark Technology tells us — though the stars are far, we can reach them by raising a torch. This torch is called independent innovation; this torch is called serving the country through industry; this torch is called never giving up.”

At 7:00 PM, the News Broadcast covered the event for three minutes and twenty seconds.

In the footage, Li Weimin explained the technical principles in the laboratory, Zhang Wei displayed the chip sample, and Su Chen gave an interview.

Su Chen’s interview segment was played repeatedly: ”The success of the chip is not the end, but a new beginning.”

”Next, we need to solve the ecosystem problem — letting more software adapt to the Flint architecture, and letting more devices use the spark core.”

”We will continue to invest in R&D; 14nm is already on the way, and 7nm is also in the planning stages.”

”This road is difficult, but it must be taken. Because core technology cannot be bought, begged for, or obtained through pleas.”

”Finally, I want to say — thank you to everyone who supported Spark Technology, especially those who still believed in us when our first tape-out failed. This trust is more precious than the chip itself.”

The self-media was in a complete frenzy.

The tech blogger ”Hardcore Gamer,” who had mocked Spark Technology the hardest, posted a long apology: ”I was wrong.”

”Ten days ago, I said the Xinghuo Chip was a bubble, a hype. Today, Spark Technology slapped my face hard with a 56.3% yield rate.”

”I welcome this face-slapping, I look forward to it, and I hope to be slapped in the face every day from now on. Because this means our country is making progress.”

This article received over a million likes.

The digital blogger ”Geekerwan” conducted a deep teardown and comparison.

They obtained a sample through their channels, opened the packaging, photographed the chip’s microstructure under a microscope, and compared it with TSMC’s 16nm and Intel’s 14nm chips.

The video’s conclusion: ”In terms of process, Spark Technology’s 16nm FD-SOI has a transistor density of about 85% of TSMC’s FinFET, but it has a clear power consumption advantage.”

”In terms of performance, the CPU single-core performance is about 65% of Intel’s i5-7600K, but the multi-core performance is comparable.”

”The GPU performance is average, but the NPU is a highlight — the AI computing power has already reached an industry-leading level.”

”The most important thing is — this is a completely independent architecture and process.”

”No ARM licensing fees, no TSMC foundry fees, no reliance on America technology. This is the true ’domestic chip’.”

The number one question on the Zhihu hot list: ”How to evaluate the historical significance of the success of the Spark Technology 16nm chip?”

The top-rated answer came from an anonymous user, later confirmed to be a chip expert at a military research institute: ”There are three layers of significance.”

”First, technical significance: it proves that Huaxia has the ability to research and mass-produce advanced process chips.”

”A 56.3% yield rate, while still trailing TSMC’s 95%, already possesses commercial value.”

”More importantly, the FD-SOI route may become a differentiated advantage for Huaxia’s semiconductor industry — in areas sensitive to power consumption, such as the Internet of Things and automotive electronics, FD-SOI has more advantages than FinFET.”

”Second, industrial significance: Spark Technology has built a complete industrial chain from EDA tools and chip design to manufacturing, packaging, and testing.”

”This means that even in the most extreme scenarios (such as a total blockade), Huaxia still has the ability to produce advanced chips. This is the cornerstone of national security.”

”Third, symbolic significance: it tells the world, and tells ourselves — the people of Huaxia can do more than just assembly and integration; we can also do underlying innovation.”

”The value of this confidence cannot be measured by money.”

This answer had over 100 million views and over 3 million likes.

The most moving scene appeared in an ordinary person’s social media feed.

Long-time user Wang Jianguo, the ride-hailing driver who used to be a taxi driver, posted a photo: his left hand holding a phone, with the screen showing his stock account — Spark Technology, holding cost 1.03 yuan, current price 2024 yuan.

His right hand was holding the key to a car model.

The caption had only one sentence: ”Two years ago, everyone laughed at me for being stupid, buying Spark Technology stock at one yuan.”

”Today, the stock is worth two thousand, and I’m driving the car. This is the power of believing in Spark Technology manufacturing.”

This photo was shared crazily and was called the ”2017 Huaxia Confidence Photo of the Year.”

The international reaction was swift and complex.

The America Semiconductor Industry Association issued a statement: ”The technical parameters announced by Spark Technology remain to be verified by an independent third party... but if the data is true, it will significantly change the global semiconductor industry landscape.”

”America companies need to re-evaluate their competitive strategies.”

The statement was cautiously worded, but the internal tension was already overflowing from the text.

A TSMC internal memo leaked online: ”Preliminary conclusion of technical assessment: The FD-SOI route adopted by Spark Technology shows unique power consumption advantages at the 16nm node.”

”In emerging markets such as the Internet of Things and automotive electronics, it may pose a threat to our FinFET route.”

”Recommendation: Accelerate FD-SOI process R&D, or consider technical cross-licensing with Spark Technology.”

South Korean media had sensational headlines: ”The Speed of Huaxia’s chip Rise Exceeds Expectations; Samsung Needs to Re-evaluate Competitive Strategy.”

The article pointed out that although Samsung has a more advanced 7nm process, it may face low-price competition from Spark Technology in the mature process market.

The commentary from Nikkei was even more profound: ”The success of Spark Technology validates the feasibility of non-America technology routes.”

”Under the trend of ’de-Americanization’ of the global supply chain, Japanese companies might be able to learn from the Spark Technology model — developing an autonomous and controllable semiconductor ecosystem while avoiding America patents and technical restrictions.”

”This holds important implications for the revitalization of Japan’s semiconductor industry.”

Meanwhile, in Shenzhen, at the Spark Technology headquarters, Su Chen stood in front of the office window, watching the crowd gathered downstairs — there were reporters, investors, ordinary citizens, and even some people holding banners saying ”Spark Technology is Awesome.”

Lin Wan walked in: ”Mr. Su, the leaders from the MIIT called to congratulate us on the success of the chip and stated that they will fully support the ecosystem construction of the Flint architecture.”

”Hmm.”

”In addition, Huawei, Xiaomi, OP, and VO have all sent letters of intent for cooperation, hoping to obtain samples of the flint no. 2 for testing.”

”Send it to them. But remind them — the software ecosystem of the Flint architecture is still under construction and needs time.”

”Also...” Lin Wan hesitated, ”That ’Nightingale’ from the CIA has appeared again. This time, she is contacting our materials team.”

Su Chen turned around, his eyes turning cold: ”Details.”

”She disguised herself as a business representative from a European materials company and wanted to tour our carbon fiber production line.”

”After being identified by the security department, she tried to contact Wu Yi through a third party.”

”What was Wu Yi’s reaction?”

”According to the plan, he showed hesitation and wavering, and gave her some peripheral technical data — it was all real, but outdated,” Lin Wan said.

”The security department is monitoring and preparing to close the net during her next contact.”

Su Chen thought for a moment: ”Don’t rush. Let her continue to ’succeed’ a few more times to buy us more time.”

”The chip is a success, and next, materials will become a key target. We want to use her to convey some information we want to pass on.”

”Understood.”

After Lin Wan left, Su Chen looked out the window again.

The cheers from downstairs could be heard faintly.

It was the joy of victory, the release of long-suppressed emotions.

But Su Chen knew that the real battle was only just beginning.

The success of the chip meant that Spark Technology had officially entered the most brutal battlefield — head-to-head competition with the world’s top technology giants.

There is no warmth there, no retreat, only winning or dying.

Outside the window, in the sky over Shenzhen, the dawn light was dispelling the night.

A new day had begun.