The past of China's core manufacturing: hot money is the least important

Mirror Studio

2024-05-17 11:49Posted on the official account of Beijing Jiupai News Mirror Studio

The past of China's core manufacturing: hot money is the least important

Text丨Li Jinjin

Editor丨Lu Zhen

A year ago, also in May, a sudden dissolution meeting put OPPO's chip research and development company "Zheku Technology" on the focus of public opinion. The failure of Zheku not only made more than 3,000 practitioners lose their jobs overnight, and their careers and even life trajectories were changed, but also cast a haze on China's high-end chip research and development.

The dissolution of Zheku is a microcosm of the turbulence in the domestic high-end chip industry in recent years. Since 2018, driven by various factors, the wave of high-end chip entrepreneurship has been surging, and the rolling hot money, and a large number of people with wealth dreams and family and country feelings, have plunged into this traditional industry that has been the same for decades.

It wasn't until the tide receded that the people caught in it realized that money wasn't always everything. The logic of modern business, such as tuyere, financing, and dividends, has failed in the ancient industry of chips. Perhaps, this has never been an industry that needs miracles, and after the excitement, China's high-end chips need more patience and time.

An ending that no one expected

When he was thrown into the vast sea of recruiters again, the young Wang Wei'an still had the illusion at first, "If there is no job, I will find another job, and I am not the only one who has been laid off anyway." On May 12, 2023, OPPO announced the closure of all businesses of Zheku, and the team of more than 3,000 people, including Wang Wei'an, was disbanded overnight.

This young man born in 1997 is good at learning from his lessons, and he has a very clear goal for his next job, "the company should be stable", the source of money is the main measure, "it depends on whether it is financing or state control".

The cold water of reality has been poured again and again. Dozens of resumes were submitted, and less than half of them received a response, and more than 1/3 of them were rejections. The hard-won interview opportunity also made him feel frustrated again and again, "You can't answer professional questions, you know it in your heart."

Talking about the reasons for the setback in the job market, Wang Wei'an was straightforward: "Because I am not competitive. ”

Just a year ago, Wang Wei'an was still a chip engineer admired by his peers: a C9 college with a master's degree in integrated circuits, and when he graduated, he had an annual salary of 400,000 yuan and stepped into Zheku, a chip "new star" enterprise.

In just one year, the job that he was proud of has become an obstacle to finding a new job. It deprived Mr. Wang of the advantages of being a fresh graduate, but it didn't give him enough growth cycle — most companies require three years of work experience, and Mr. Wang has only one year of work experience on his resume, as a marginal member of a high-profile project that is not even successful.

"For novices like us, who have just joined the company for a year or two, no headhunter is willing to look for us." Wang Wei'an said.

He was not discouraged, and in the cold job market, he lowered his standards and expectations step by step. The first thing that was crossed out was the size and size of the company, "no matter how big or small the company is, there is definitely something to learn". Followed by salary, he said that he was quite peaceful, "At the beginning, I said that it would be good to have an equal salary." But obviously, his optimism is still beyond reality, and there is hardly a company on the market that is willing to take such a new engineer at the same salary.

Time was getting tighter, and in the end, Wang Weian had to compromise and "only improve his competitiveness by cutting salaries".

Such a situation would have been unimaginable a year ago. When recruiting in the fall of 2022, Wang Wei'an, who graduated from the undergraduate class, had almost no difficulty in finding a job, "as long as it is a professional class, there are projects that can be handled, and the offer is really soft." At that time, more than a dozen offers were placed in front of Wang Wei'an, and the average salary could be 25% higher than in previous years.

Wang Wei'an is not a person who likes to catch up with the tide, he was keenly aware of the 35-year-old crisis and the traps brought by the times and industry dividends, and began to lay out his life after the college entrance examination. Choosing a major deliberately avoided the Internet, "At that time, there was already a downward trend, and the Internet legend was that if you couldn't be a leader at the age of 35, you would be laid off."

After comprehensively considering the city where he lives, as well as the scale of the enterprise, the track and the source of funds, he finally chose Zheku, "There are more than 3,000 people in all the projects, which feels relatively stable, and it is a wholly-owned subsidiary of OPPO, so I don't have to worry about getting money."

For a young person who has not yet set foot in society, these considerations are thorough enough. In the following time, it seems to confirm the correctness of this choice. When he joined the company, Wang Wei'an was sent to Shanghai for a month of full-time training and lived in a five-star hotel. The company seems to have abundant funds and sufficient employee benefits, including coffee machines and ice machines, afternoon tea, weekly club activities such as badminton and basketball, and team building, as well as lucky draws during the New Year's holidays.

Not only that, in the domestic chip industry, Zheku is like a rising star. Backed by OPPO, the chip design company that aims to develop its own flagship mobile phone SoC (System on Chips) (mobile phone core chips integrating CPU, GPU, communication and other modules) continues to show its ambitions to the outside world.

● Mariana MariSilicon Y on display during Mobile World Congress in Barcelona, Spain, 2023. Source: Visual China

It named the chip self-development plan after the world's deepest trench "Mariana" to describe the most difficult thing to do "top chips". The parent company, OPPO, transfused blood and built a complete industrial supporting facility for it. In 2019, OPPO continuously invested in more than a dozen upstream and downstream companies related to semiconductors. After that, Chen Mingyong, founder and CEO of OPPO, announced that he would invest a budget of 50 billion yuan for the research and development of underlying technologies such as chips.

It was a period when domestic chip companies were competing, and there was a shortage of talents in the industry. In order to roll the chain as soon as possible, Zheku recruited talents from leading companies such as Huawei HiSilicon, Zoomlion, and UNISOC with the highest salary level in the industry, and built a team of more than 2,000 people in less than four years. According to the open recruitment data, nearly 80% of Zheku's engineers come from top universities at home and abroad, with more than 5 years of industry experience.

It is goal-oriented. From the beginning, the company focused on the most difficult SoC chips and baseband chips, hoping to reduce product costs and improve profits and competitiveness by using self-developed chips like Apple. Wang Wei'an remembers that at the newcomer training meeting, the company's future vision was repeatedly mentioned, and "the whole blueprint has even been planned for products in the next few decades."

In 2021, Zheku released the first self-developed NPU chip for imaging, "Mariana X", which can help mobile phones optimize the photo effect, and announced that all OPPO flagship mobile phones will be equipped with this self-developed product in a year. In the second year, Zheku launched the second Bluetooth audio self-developed chip "Mariana Y", which can support Bluetooth headsets to transmit lossless audio. The chip is made using the N6RF process, which is the only company in the world other than Apple to use this advanced process. In the same year, OPPO invested nearly 4.5 billion yuan in the R&D center project in Jiaoyi Bay, Dongguan.

According to the plan, after the image and audio chips, Zheku will launch the core SoC chip of the mobile phone. The road of core making seems to be moving forward steadily, but the reality suddenly swings a stick. In May 2023, due to the "security system upgrade", all members of Zheku were asked to work from home, and the next day an online all-staff meeting was held, and the team directly announced its disbandment. This new star, which is soaring into the sky, fell in an instant, and smashed to the ground, including the jobs of more than 3,000 employees.

From calm to crazy

After the dissolution of Zheku, until the winter of 2023, the sea of chips has been frozen. Under the cold weather, Xiaoran, a 25-year-old headhunter in the chip industry, has not negotiated a list for a long time, and "the income is very bleak". The number of corporate jobs has dropped from dozens to single digits, or even zero, and the mentality of job seekers has also changed to "stable", "candidates are more cautious, they will worry about instability and choose to stay in the original company".

This headhunter, who has been in the industry for less than two years, has always been chasing the wind. She first devoted herself to real estate and the Internet, and when the twilight of the industry came, she felt the wind blowing from the chip industry. searched for a few professional videos on station B, wrote down those awkward professional terms, and turned into a headhunter in the chip industry in the second half of 2021.

It's a pity that she only grabbed the tail of the tuyere. She has heard a lot of the scenery of the chip industry, "At that time, you could really pick up money everywhere." So far, she regrets that "I have been in the industry for a short time, and I have no industry precipitation, so I am still late."

stepped into it, even if it was a novice in the industry, he was quickly involved in the crazy battle for people at that time. According to public data, the number of registered chip companies in China has risen sharply since 2018, and the number of registered chip companies in 2020 will be 59,793, nearly 5 times that of 2014 and nearly 100 times more than 10 years ago.

In order to be able to stake their ground quickly, these start-ups are looking for talent. "If a small company wants to raise money, it must hire people to do things," Xiaoran said, "but the scale is small, and it can't attract people, so it can only rely on headhunters to introduce advantages." They charge a high headhunting fee for this, and the headhunting fee for the most basic engineer position can be 50,000 or 60,000 yuan, compared with the headhunting fee of only high-level managers and engineers in large companies, and it is only a few thousand yuan.

Not only that, but the boss of the startup company even went down in person and flew to another place to persuade the candidate. Many headhunters simply bypass the process, as long as they get the candidate's phone number, they will directly enter it into the company's system, and seize the quota first, "without even contacting the other party".

At the other end of the spectrum are the job seekers who stand tall on top. Many of them changed careers halfway, spent money to take a crash course for a few months, and then uploaded their resumes to recruitment websites, and dozens of phone calls poured in every day, holding seven or eight offers at the same time. When communicating the position, a candidate replied, "The registered capital of this company is not as much as my family's deposit", so Xiaoran was deleted. The HR of a chip company recalled that because the communication offer was a day later than expected, the fresh graduate on the other side of the phone accused, "Why are you so slow?" Aren't we God? ”

Of course, most of the time, as long as you give enough money, everything else is easy to say. In the chip industry, the school recruitment in 2021 and 2022 is almost piled up with money. Zheku's method of grabbing people is money, and on top of the salaries of other companies, it directly doubles to "pinch the tip". The industry was miserable, but it had to join the war, "single-handedly raising the salary level of the entire industry," said an HR at a mobile phone chip company.

● Huge investment in chip R&D is an industry where money is piled up. Source: Visual China

These seemingly endless amounts of money are all passed down from the upstream. According to Yunxiu Capital's "2020 Interpretation of China's Semiconductor Industry Investment", the investment amount in the semiconductor industry in 2020 exceeded 140 billion yuan, an increase of nearly 4 times over 2019. The start-up company Biren Technology created a "chip speed" of "one round of financing every 3 weeks" in that year.

Investor Bob doesn't know anything about chips, and it doesn't prevent him from embracing chips, "more than 90% of semiconductor investors have a failing knowledge base." But Bob has his own set of logic: the night before Baidu's professional terms, the next day at the dinner table can also talk to the entrepreneur, "I want to understand what this word is, but also to be able to take a sentence, seem to understand very well, must not be timid."

The investment logic is not difficult, when all companies do not have products, you can only look at the background of the founder, "the most favored is the work experience of foreign and relatively top semiconductor companies".

Recalling the madness of the time, Bob said, "A lot of people's mindset is not that they are afraid of making a mistake, but that they are afraid of missing out. An investor who insisted on not getting involved in high-end chips said that the biggest pressure came from partners, "Why are all institutions investing but you not investing?" ”

These madness were all seen by Wang Li. He is a chip design engineer who worked in the chip design department of a communications company after graduating in 2013. Before the arrival of the outlet, Wang Li's work was 9 to 6, "similar to that of a civil servant". In his experience, although the chip manufacturing process continues to advance, the design field is still relatively traditional, the R&D methods are old, and the process is fixed, "nothing particularly new, just like the automotive and machinery industries". In the era when there was no "lack of cores", chip self-development lacked a lot of motivation and enthusiasm, "it's a big deal to buy someone else's, it's also good, anyway, it's the price of cabbage."

At that time, the better career options were to go to the Internet. Wang Li was assigned to the chip department, he can only "come and be safe", sometimes he sees the kind of enterprise outsourcing development App project, one makes hundreds of thousands, he is also very excited. I thought I had missed the outlet, but I didn't expect that after 2018, chips suddenly became popular, hot money came, headhunters came, and the most exaggerated time was when more than a dozen or twenty headhunters contacted him a day, all of them were start-ups recruiting people, and some positions even offered millions of annual salaries, and everyone wanted to get a piece of the pie from "domestic substitution".

The industry is still that industry, but the water temperature here is already boiling.

Money is the least important

The first project Wang joined in Zheku was the Mariana Z3. In the vision of the company's top management, this SoC equipped with 5G baseband, using TSMC's most advanced 4nm process, will become the first generation of Zheku's top chip, and the former Zheku's first architect publicly said that it will be "an industry-leading SoC chip".

This chip, which has high hopes, drives the continuous rotation of 3,000 people up and down Zheku. In Wang Wei'an's impression, after joining the company, all departments are preparing for the tape-out (trial production) of this chip, and he can only get off work after 9 o'clock every night, and he can only take a single day off on weekends.

There is still a shortage of people. At that time, the biggest problem for the leaders of Wang Wei'an's department was that they "couldn't recruit people," and the upper echelons even directly issued orders to increase the number of people in his department from the original 20 or 30 to 60 or 70. As a newcomer, Wang Wei'an is destined to be just a marginal role, not only unable to access the core work, but also to be deployed at any time.

Not only is there a mismatch of talent, but rapid expansion also brings management challenges such as resource competition, resignation, and delivery delays. People from different corporate backgrounds converge in Zheku, giving rise to factional disputes and run-in styles. A former senior employee of Zheku mentioned in an interview that there is often a phenomenon where two departments are developing the same function, and "then there will be a competition for resources".

Under the undercurrent, the plan seems to be proceeding in an orderly manner according to the blueprint. In the spring of 2023, this pioneering SoC chip will be sent to TSMC for tape-out, that is, for small-scale trial production.

Strictly speaking, a mature high-end chip that can really be accepted by the market requires a long period of technical accumulation. Taking Huawei HiSilicon as an example, it took more than 10 years from its establishment to the launch of the Kirin 9 series chips, and finally realized the 7nm process. The Qualcomm Snapdragon chip, which now dominates the market, traces its technology back to the 2G era.

Founded less than four years ago, Zheku has launched the high-computing power image processing chip Mariana X and the high-speed Bluetooth processing chip Mariana Y. It then leapfrogs the basics of the technology and challenges the most difficult 5G SoCs.

● 2020, Beijing, a self-owned chip display board at a fair. Source: Visual China

This is one of the reasons why Zheku is widely criticized. In the chip industry, its failure does not seem to have brought much of an impact. "It's expected," said Wang Li, a chip design engineer. In Wang Li's view, the biggest problem of Zheku is that "there is no technical foundation, and it is difficult to make such a difficult chip as soon as it comes up".

Wang Li explained that the technical foundation is difficult to pile up with money. Mobile phone processor chips involve camera, image processing, WIFI and other modules and interfaces, which need to be developed and accumulated one by one. If you develop it yourself, it will need to iterate, test and evaluate constantly, "which is a very long, painful process".

Zheku tried to make up for the shortcomings with money, such as poaching people at a high cost, but it was difficult to succeed. In the process of chip development, the most basic thing is the design code. Wang Li took Huawei's Kirin processor chip as an example, the front-end design involves hundreds of millions of lines of code, and a simple file size is hundreds of gigabytes, and they are generally stored in an encrypted way on a large server. ”

Purchasing an IP may be an effective way to accelerate. This is also the "secret" in this round of chip self-development in China. But it also means a higher cost, "which is unimaginably exaggerated," Mr. Wang said. "Without a technical foundation, all the costs are amortized, and in the end, it will definitely not be sustainable."

Not only that, even if the problem of technology accumulation is solved, these chip projects that rely on capital stacking still need to be verified for several years. Taking a 7nm chip as an example, from birth to mass production, it has to go through a series of complex processes such as design, verification, tape-out, and function and performance testing after reproduction.

According to traditional experience, if the tape-out is successful once, it will take a year at the earliest. The reality is that "no one can succeed at 7nm, including Qualcomm," said Xu Qing, an investor. "Because there will always be bugs, which have to be fixed and fixed."

On one end is the slow iteration and accumulation of technology, and on the other end is the hot money that has suddenly plunged into the chip industry and needs urgent returns. What laymen didn't expect was that the so-called outlets, financing and dividends failed in the ancient industry of chips, and hot money fell here.

Money is the least important thing in this industry, Xu Qing said, "So, strictly speaking, it is impossible to accelerate this thing, not that you can accelerate it if you have money." ”

Helpless choice

The clock is turned back to 2023, and multiple shocks are intertwined in this year.

The first is the unfavorable external transfusion environment. Data shows that global smartphone sales will fall by 12% in 2022, and OPPO will be even worse, with shipments falling by about 27% year-on-year. The cold winter of electronic consumption has directly affected the blood transfusion capacity of OPPO, the parent company of Zheku. OPPO, which relies heavily on the low-end thousand-yuan machine market, has a mismatch with Zheku's chips.

So, with the shrinking demand of the mobile phone industry, even if high-end chips are mass-produced in the future, the question in front of Zheku and OPPO is, how big is the possibility and market of commercialization in the future?

● In 2023, Shanghai, at the World Artificial Intelligence Conference, staff check computer chips and the word "autonomous". Source: Visual China

Many people have begun to realize that on the road to core manufacturing, technical barriers and cycles are just the beginning, and more importantly, the commercialization problems after self-developed chips and the lack of industrial ecology are the biggest obstacles in front of China's high-end chips. "The core problem of high-end chips is who uses them and how to use them, and if no one uses them, its ecology cannot be established." Investor Xu Qing said.

Xu Qing likened the ecology of high-end chips to "a pyramid". For example, he said that Intel, which has mastered the x86 architecture, is at the top of the pyramid, and below it is the technology that defines the digital infrastructure, Microsoft's operating system, and the lowest application, Adobe. Obviously, domestic high-end chips lack such a market ecology, which is also a problem that self-developed chips have to face under the wave of "domestic substitution". "It's progressive, and if you don't even have this pyramid, what's the point of making it?"

The problem of ecological lack is also gradually revealed in the upstream and downstream chains of high-end chips. From EUV lithography machine to chip design industrial software EDA, the technical code is in Europe and the United States, which has become a key link on the road of chip self-development. Every Chinese company that tries to challenge the technical problems of chips has to face patent blockades and various restrictions in Europe and the United States.

Multiple factors are intertwined, and China's high-end chip industry is ushering in a moment of deep quagmire.

The dissolution of Zheku was like a depth charge, and the aftermath of the shock was transmitted one after another. In August 2023, Meizu Group terminated its self-developed chip business, and in November, Moxing Semiconductor, a subsidiary of TCL Holdings, was dissolved. Beneath the waves, countless small boats were overturned. According to titanium media reports, a total of 10,900 chip-related enterprises will be cancelled or revoked in 2023, an increase of 89.7% over 2022.

Under the turmoil of the industry, the Mariana Z3, which was once full of hope for domestic substitution and crossed the ocean, has been forgotten before it even has time to send back the news. People in the chip industry are anxiously looking for their own way out.

Investor Bob and headhunter Xiaoran left the market. "In another five years, I will not be able to reach the height of professionalism, and I will always be a mediocre third-rate investor." Bob said. In his perception, at the beginning of 2022, the semiconductor investors around him began to do post-investment work one after another, that is, to find people to connect resources and improve the performance of invested companies, and some people are even considering the problem of stop-loss exit, "At this time, the wind direction has changed, not thinking about going in, but thinking about coming out."

The setbacks in the industry may also breed new opportunities. After the withdrawal of hot money, the industry has undergone baptism, and the people in it have begun to become rational. Taking advantage of the industry, a girl crossed over from the materials major to the chip industry, but quickly fell off the tuyere. She has been laid off twice, giving up the pursuit of "stability", "whether it is a large company or a small company, basically facing layoffs", in the latest job offer, she hopes that "you can learn a little something, the most important thing is to improve yourself", anyway, "you can't change the general environment." ”

After leaving Zheku, Wang Wei'an finally became an engineer in a small chip company in Xi'an with equal salary. Small companies don't seem to be standardized, for example, the progress of the project can only be decided by the leader, in contrast, as a large company, Zheku has a complete process and shared knowledge documents.

But he didn't have to pick. "After all, if you don't choose this opportunity, you won't know where the next opportunity is."

Passing by the original office building in Xi'an, Wang Wei'an found that the traces of Zheku had been wiped clean, and in the same building, which is now attracting attention, is Xiaomi, which also aims to develop its own mobile phone chips, "We just fell, and they went in again." The research and development of China's high-end chips has to be continued.

(Names have been changed)

Copyright Notice: The copyright of all contents of this article belongs to Mirror Studio, and may not be reproduced, excerpted or used in other forms without written permission, unless otherwise stated.

View original image 65K