laitimes

The sad moment of OPPO's four years of core making, a boom waking up

The sad moment of OPPO's four years of core making, a boom waking up

Author|Ma Hui

After four years of high-profile core manufacturing, OPPO's chip subsidiary Zeku was finally shut down by Liu Jun, the CEO who founded it. After announcing the decision to shut down, Liu Jun's face was full of sadness. Facing 3,550 employees off camera, after a few seconds of silence, he read a poem, the second half of which reads, "Good dreams are the easiest to wake up."

The brief, 18-minute meeting, chaired by four executives, announced the end of OPPO's four years of core manufacturing.

It was a dissolution without warning, and no one knew when or by whom the decision to shut down would be made. It was also a neat layoff, with no business contraction, no partial layoffs, but all employees losing their jobs together.

In the past four years, OPPO's investment in ZEKU has been a lot of expenditure, CEO Chen Mingyong once said that "the money is not small", but there is no sale plan, there is no divestment, the chip company ZEKU, which has existed for four years, will no longer exist from the dissolution.

For OPPO, which hopes to reduce the cost of components and optimize the performance of mobile phones by manufacturing cores, so as to improve the company's profits and competitiveness in the long term, this is an expensive trial and error. The difficulty of core manufacturing has been warned early: ZEKU's entry into the system-on-chip SoC integrating 5-7 chips, and only Apple, Huawei and Samsung have come up with products among mobile phone manufacturers.

OPPO has also repeatedly expressed its determination to make a core, that is, it will invest 50 billion yuan in technology research and development expenses by 2023, and chips are the most concerned one, and named the core building plan "Mariana", which is the deepest trench in the world. But this business to enhance the company's differentiated competitiveness has become a heavy burden on OPPO's shoulders: global mobile phone shipments and replacement frequency are declining, mobile phone company revenue is peaking, and core manufacturing investment is still bottomless, and there is no benefit in sight.

Looking back four years ago, the intensified Sino-US conflict and the executive order signed by Trump made Huawei's high-end chips fold, affected by this, Chinese companies are willing to purchase self-developed chips to cope with the uncertain future, which has blown up a wave of core manufacturing in China, not only giving birth to a number of chip startups with the background of domestic substitution, but also allowing mobile phone manufacturers, Internet manufacturers, and automobile manufacturers to join the core team, engineers' salaries have risen year by year, and more and more headhunters have poured into the chip industry.

OPPO is the largest and most resolute one in the core manufacturing team of large manufacturers. Around its subsidiary ZEKU, there are not only more than 3,000 employees, but also more than 50 headhunting suppliers. ZEKU is one of the largest customers in the chip talent market. After learning of ZEKU's shutdown, a headhunter whose performance has come from ZEKU for three consecutive years felt unprecedented panic, "If I don't make a few orders this time, I will lose my job."

The panic comes not only from the loss of big customers, but also from the unease of the chip talent market that has cooled since this year.

Unlike the previous two years, when chip startups emerged one after another and engineers continued to jump jobs, the market is calmer this year, and more than 3,000 employees have been released at ZEKU, and the market relationship between the company's demand exceeding the supply of engineers will be temporarily inverted. Although the establishment and shutdown of ZEKU is essentially about OPPO's business decisions, the startup that has attracted the most people in the past four years inevitably caused a crisis of confidence.

Engineers and headhunters are finding their companions and reassuring each other, "just as if they paid a windfall in the past two years", because it will be difficult later.

01 Disbanded without warning, there is a problem of core making with early warning

Usually, companies always reveal some wind before a major decision occurs. BUT ZEKU IS NOT THIS TIME. This is an ordinary All hands meeting, except that on the evening of May 11, the company suddenly informed all employees that because of the server IT and access control upgrade, they did not have to come to the company to work the next day, and they could not log in to the server. There was not the slightest suspicion among the employees, not a trace that the company would be dissolved the next day.

Until May 12, ZEKU had been operating as usual: HR was still busy communicating with multiple suppliers about candidate requirements; The recruitment process on the website and the onboarding process for new people are all going on normally. Three weeks ago, ZEKU also recruited an executive who jumped from GPU star startup Bicheng to be responsible for ZEKU's NPU chip (image chip). Executive job hopping is also a kind of weather vane, proving the reliability and strength of ZEKU.

More than 1,600 engineers in the AP chip department (including SoC and NPU), one of the four major chip departments of ZEKU, are still waiting with hope for the 4nm SoC that has been developed for 2 years to return the chip from chip manufacturer TSMC next month, and the test light up after the return film proves that this is a chip without bugs and can run through, and then it is on the machine for trial and preparation for mass production. When mass production is successful, they will be the fourth mobile phone manufacturer in the world to develop an SoC. This journey of core making, which lasted for four years, can be regarded as a small progress.

All the engineers are waiting for this moment, and it looks like things are moving in a better direction. For example, OPPO is as generous and tolerant as ever to the ZEKU team. Two months ago, many OPPO employees decided to defend their rights because they were dissatisfied with the year-end bonus discount, but most of ZEKU's engineers received more year-end bonuses than OPPO employees. This makes ZEKU engineers more convinced of OPPO's investment in core manufacturing, and it is more difficult to imagine that ZEKU will be abandoned.

On May 12, many employees were happy to get a free day off. An engineer who is celebrating his birthday in two weeks is planning a trip next week: a concert in Hangzhou and a vacation to Sanya on annual leave. Before attending this meeting, he was entangled in how to ask the leader for annual leave at this busy project node, and when he heard Liu Jun suddenly choke up and say "shut down ZEKU and give up self-developed chips", his first reaction was unbelievable, "Three thousand people are leaving?" Then he was anxious about the heavy mortgage after losing his job, and after learning that he could get "N+3" compensation, his tight heart relaxed, and this relatively generous compensation allowed him to rest for at least a few months. In just 18 minutes, most engineers have experienced this shock, anxiety and even a little relief.

According to the data of the recruitment website, the average annual salary of ZEKU is 400,000-500,000 yuan, and the average person can get 20-300,000 yuan in compensation, and ZEKU has to pay 6-900 million yuan in compensation. "It didn't deteriorate people's moods," said a headhunter who has worked for Zeku for three years.

Sudden, trance, is the most mentioned word when engineers chat after going offline, because the incident happened suddenly, Liu Jun mentioned at the meeting The reason for the shutdown was not accepted by everyone - "The global economy and mobile phone industry are now extremely unoptimistic, the company's entire revenue is far from meeting expectations, and such a huge investment as a chip will be unaffordable for the company." ”

Some speculate that the reason may be greater force majeure. After all, OPPO has been making cores for four years, and now it is only one foot away from the door.

In fact, the core manufacturing of large factories has always been an uncalculable account in business. Especially in the mobile phone market with a mature supply chain, upstream chip companies have steadily provided SoC, NPU and Bluetooth chips to mobile phone manufacturers for many years, which can dilute the huge chip research and development costs with large shipments, so that the quotation of a single chip is lower. In just a few years of hard work, it is difficult for mobile phone manufacturers to catch up with chip companies that have been deeply engaged in the chip field for more than ten years.

Among them, OPPO's in-depth challenge SoC is a chip that integrates a large chip, and its manufacturing difficulty and process complexity are the sum of the difficulties of several chips. Huawei began to make processors in 2009, and developed the first chip K3V2 in 2014, but encountered extremely poor performance, the reputation of "pole card", its Kirin chip iteration for 5 years, only to gradually become the core competitiveness of Huawei's high-end flagship machine, during which Huawei invested more than 100 billion yuan in HiSilicon semiconductor.

It's a time-consuming and expensive chase.

An engineer said with regret on Zhihu: "This generation cannot reach the current flagship (chip) of Qualcomm and MediaTek, but iterating two or three products, we believe that we can make excellent Android flagship SoC chips." ”

This means that compared with this compensation of about 1 billion yuan, OPPO will pay at least 10 times more research and development expenses. An SoC tape-out requires hundreds of millions of yuan, and the annual labor cost is about 2 billion yuan. This is indeed in line with Chen Mingyong's preset investment scale at that time: 50 billion yuan.

But at this time, OPPO's main business is not as optimistic as it was at the beginning. According to the latest mobile phone quarterly tracking report released by market research institute IDC, global smartphone shipments in 2022 fell by 11% year-on-year, the lowest level in a decade, and China's smartphone market shipments fell below 300 million units for the first time in eight years, the largest decline in history. OPPO's global market share fell 2 percentage points in the first quarter of this year from the previous year.

The Find X5 series, which has been equipped with the self-developed NPU chip Mariana X, has not achieved outstanding results, sending a strong vote for OPPO's high-end. Although OPPO did not announce the sales of the Find X5, in JD.com, the most important online sales platform for electronic mobile phones, the Find X5 sales are only in the thousands. An industry insider said that a chip shipment must reach at least 20 million pieces to recover the cost.

New energy vehicle companies and Internet manufacturers will face the same problems when investing in core manufacturing: high investment in chip R&D, slow results, difficult to dilute R&D costs for self-use alone, and single-chip bidding is not comparable to chip manufacturers. When large manufacturers make cores, their only optimistic expectation is that the sales volume of the main business will be sold to a scale that dilutes the cost of research and development. But once the market is down, they are in a dilemma.

ZEKU IS EVEN MORE EXPOSED ON THIS ISSUE. It invested more resolutely in core building and withdrew more thoroughly.

02 High-profile core manufacturing for four years, the fastest growing number of startups

OPPO's obsession with core manufacturing has traces to follow. As early as 2016, OPPO founder Chen Mingyong invested in a web search processor chip, after which OPPO established a Jinsheng Communication Co., Ltd. that increased its semiconductor business in the later period, but it failed. At the first OPPO Future Technology Conference in 2018, CEO Chen Mingyong proposed that R&D investment in 2019 would increase by 6 billion yuan, and then increase investment year by year.

It seems that OPPO was already preparing funds for core manufacturing at that time. In 2019, OPPO's Shoupu Technology (Shanghai) chip research and development project was officially established and the company was established, and a year later it was renamed Zheku Technology (Shanghai) Co., Ltd., which was the moment when ZEKU officially debuted.

This year, OPPO also recruited former Qualcomm technical director Chen Yan and established a chip TMG (Technical Committee) to provide support for chip manufacturing. At this time, OPPO basically determined the core, and in the recruitment information released by OPPO at that time, it was clear that SoC design engineers who could complete architecture definition and functional design, and architects were the most important talent reserve for an SoC.

At the end of this year, Chen Mingyong, who rarely appeared, proposed to invest 50 billion yuan in the next three years at the Future Science and Technology Conference, focusing on 5G, 6G and artificial intelligence, and paying attention to the underlying core technology, and chips are one of the projects that are focused on.

By February 2020, OPPO officially disclosed its self-developed chip plan. In the internal article "Some Thoughts on Building Core Technologies", three major plans are proposed, involving software development, cloud services, and the "Mariana Plan" about chips.

Its responsible persons also appeared one after another, and this year, the media disclosed that Zhu Shangzu, former chief operating officer of MediaTek, and Li Zonglin, former general manager of MediaTek's wireless communication business unit, joined ZEKU one after another. They have since become the leaders of ZEKU, and they also showed up at the last moment of ZEKU to say goodbye to the entire staff.

Among them, Zhu Shangzu left MediaTek at the end of 2017 as a partner of Xiaomi Industrial Investment Department, and managed a team of more than 6,000 people around the world in nearly 18 years at MediaTek, while Li Zonglin is the main driver of MediaTek chip research and development and the keynote speaker of every chip conference.

OPPO core manufacturing has become high-profile and a foregone conclusion. Since the autumn of this year, ZEKU, as Huawei HiSilicon, has been a strong takeover in the recruitment of major electronic universities, giving master's graduates a monthly salary of about 20,000 yuan and an annual salary of 300,000-400,000 yuan, making chip engineers an enviable job under the market of sluggish employment of fresh graduates. The fresh graduates who got the offer could not hide their excitement in the exchange group, "catching up with the golden age, catching up with the chip outlet", which also led to the chip training course of "studying for 3 months and getting an annual salary of 400,000".

Fresh graduates who joined OPPO this year and then transferred to ZEKU found that ZEKU has 800 people, which is the size of a medium-sized startup. Since then, after the school recruitment every year, it is the peak of ZEKU's entry, just like the fresh graduates majoring in integrated circuits in 985 and 211 schools before 2019 will get HiSilicon's Offer, and now everyone gets ZEKU's.

When recruiting in the society, ZEKU's high salary and stability behind OPPO also do not require headhunting "education" candidates. A headhunter who has served ZEKU for three years said that engineers who have been working for 5-7 years can get 70-1 million, and the salary for more than 10 years is more than 1.2 million. Compared with other startups, ZEKU's salary is not only 30% higher than the market increase, but also consists of cash, including signing fees, year-end bonuses, no equity, no options. When each headhunter recommends a company to a candidate, he adds ZEKU to the list, "If it is not added, the candidate will also ask, it is better to occupy a pit." ”

After joining ZEKU, engineers can feel ZEKU's desire to do something. For example, internally, ZEKU's goal is to be China's HiSilicon and benchmark Qualcomm.

OPPO is also pinning its hopes on ZEKU. At the OPPO 2021 Future Technology Conference, the highlight is again chips. Chen Mingyong emphasized the role of the underlying technology, he felt that all the problems came down to the underlying chip, "A technology company, without the underlying core technology, can not have a future; And the flagship product without the underlying core technology is even more a castle in the air."

By 2021, the number of ZEKU will exceed 2,000, and there is already an ISP chip in tape-out, and the research and development of SoCs has not stopped at recruitment, but has been carried out for a year. By December, OPPO launched the first self-developed chip Mariana X - image-specific NPU, which is equipped in OPPO's high-end Find series, compared to Qualcomm's 7nm process NPU, OPPO uses 6nm advanced process, research and development and tape-out costs will be more expensive. The improved performance is the enhanced energy efficiency of AI real-time computing. But how to show the mobile phone photo experience, there is not much feedback.

In 2022, OPPO released the second Mariana Y Bluetooth chip one after another. Jiang Bo, director of chip products, admitted in an interview that OPPO's self-developed chip has no price advantage, and making this chip is to make the team more familiar with the process when making cores, and also to let users have a better experience, so that the company has the ability to master upstream technology and obtain differentiated products.

At this time, there were more than a dozen subdivisions and fields within ZEKU, with nearly 3,000 people. However, a headhunter who has served ZEKU for many years feels the change, in the early years, Zeku did not have high requirements for the candidate's years, even if it was a social recruitment, as long as it was a candidate with an excellent resume, he could join the company in more than 3 years.

Before the abolition, the social recruitment demand sent by Zeku to headhunters had become more than 7 years of experience. Headhunters familiar with ZEKU revealed that nearly half of ZEKU's engineers are engineers with less than 3 years of experience, and the team urgently needs experienced engineers to take the helm. An entrepreneur with more than ten years of chips believes that a qualified chip engineer must work for at least 5 years and complete the process of two chips from design to tape-out and mass production.

It took HiSilicon ten years to cultivate such a team of tens of thousands, and it took longer for Zeku. OPPO also emphasized that it will make chips with the mentality of grinding a sword for ten years.

After the two chips came out one after another, Chen Mingyong gave many interviews, in which he mentioned that "although the money for core manufacturing is indeed a lot, I don't think it's too much, or I won't invest it again, but it's all planned." At the end of 2022, Chen Mingyong once again emphasized in an internal letter that "chips must be done, must be done well, and must move forward step by step and solidly." No matter what happens in the future, it makes sense for OPPO to make chips."

Until this year, OPPO's SoC, which has been polished for two years, was sent to TSMC and is scheduled for mass production in the third quarter. However, the engineers can't wait to come back.

03 Chip on the air vent

The core OPPO is just a microcosm of the boiling chip market.

With the intensification of the Sino-US trade conflict in 2019, Huawei's high-end mobile phone chips were no longer produced by US sanctions, and in 2020, Cambrian and SMIC were listed, and Cambrian opened up nearly 290%. After SMIC became the semiconductor company with the highest A stock market value with more than 600 billion yuan, the market's enthusiasm for the chip industry was high again. According to CITIC Securities' September 2020 research report, within the scope of its selected targets, the valuation of semiconductor companies in the A-share market in the past four years averaged 70 times PE (PE of the year), and the average value of US semiconductor companies was about 20 times.

Until 2022, new companies in the semiconductor industry will complete an average of more than 500 financings and absorb more than 170 billion yuan per year. Some engineers who have been in the industry for decades have been very shocked, before 2019, chip companies to raise 1-200 million is a big thing, after which the valuation is tens of billions, and it is normal to invest in a billion yuan. An investor who looked at semiconductors exaggerated at the time that semiconductor investment, only looking at large customers and hot directions, "everyone's judgment of valuation is based on faith, I think there is a possibility of 50 billion or 100 billion yuan."

The situation continued in the next two years: a number of large chip companies such as CPU, GPU, DPU, and autonomous driving chips with the background of domestic substitution were born, mobile phone manufacturers such as Lenovo, vivo and Xiaomi joined the core, Internet companies such as Baidu, Ali, Tencent, ByteDance, Kuaishou and other Internet companies all set up chip teams, and new energy vehicle companies such as Wei Xiaoli and Pinguan BYD began to grab people.

Engineers are particularly sought-after, and statistics from talent solutions company Hudson show that chips are the industry with the highest salary increase in 2022, with an average increase of more than 50%.

Compared to joining startups, engineers prefer to join larger factories with more financial resources, they know that chips are a research and development that requires long-term investment, although they are also worried that large manufacturers will cut the chip business, but they can only seek clues from the attitude of large manufacturers to the chip department.

ZEKU SEEMS TO BE THE LEAST LIKELY TO CUT THE CHIP BUSINESS, AND EVEN IF INDUSTRY INSIDERS ARE WORRIED ABOUT ZEKU'S FUTURE, ZEKU IS OFTEN RUMORED TO BE FULL OF INTERNAL STRIFE — IN THE FINAL MEETING, A MANAGER ALSO MENTIONED FRICTION BETWEEN TEAMWORK. However, in the early days of the establishment of a chip startup, managers from different companies will always bring different management styles and technical routes, which naturally leads to internal inconsistency in product development and technology application.

Other problems have also been happening below the surface, and an engineer who has worked at ZEKU for three years pointed out that the project node delivery tasks of many departments of ZEKU are very heavy, and the degree of internal self-research is not high. This is also due to the short establishment of ZEKU, and if you want to make a flagship SoC chip in a short period of time, you rely on the solution of introducing core IP and complex modules from the outside.

A core executive search firm serving ZEKU believes that ZEKU is conservative about technology and ZEKU is more focused on certainty. THE HEADHUNTING COMPANY HAD RECOMMENDED CANDIDATES FOR THE GENIUS JUNIOR CLASS TO ZEKU, BUT DID NOT MAKE AN APPOINTMENT, AND ZEKU NEEDED SOMEONE WHO COULD DO THE JOB WELL.

When employees join ZEKU, they will also have a hint of hesitation, "Will it be quick to recruit and dissolve quickly?" But this concern was quickly dispelled, one is OPPO's high-profile statement on core manufacturing, and the other is that this gradually expanding startup has become less and less like a startup, for the talent-intensive semiconductor industry, 3,000 people is not a small volume. No one expected it to end this way.

This allows engineers to break the sense of security in the core manufacturing of large manufacturers. A headhunter began to take a large factory as an example, "** (an Internet giant) can also cut the chip department at any time." ”

This year, the semiconductor industry has rarely heard of financing news. The shutdown of ZEKU and the run on talents have chilled headhunters and engineers who once felt that the chip market is hot.

04 Hot recruitment group, fading talent fever

On the night that ZEKU announced its closure, even if it was the last night of the weekday, when people were least willing to work, the newly disbanded Zeku engineers and almost all semiconductor headhunters were very busy greeting around. Everyone sent out the QR code of the first and second recruitment groups in the circle of friends, and soon, a batch of recruitment groups appeared. Almost all chip companies have invited ZEKU engineers: "ZEKU exclusive communication group" and "ZEKU warm group", as well as direct employment group, exclusive channel group, internal push group, the number of people are in the hundreds. Once a new QR code appears, people immediately pour in, and everyone does not want to miss any possible favorable news.

Those who are keen to join this group chat recruitment are often three types of people: small and medium-sized and traditional chip manufacturers who were not favored by engineers in the past; ZEKU engineers with less than 3 years of work and the most difficult job finding in collective unemployment; As well as chip headhunters who feel panic and are urgently looking for engineers with 3-10 years of experience.

In the seemingly hot recruitment scene, the situation has been reversed. A headhunter who has been working for Zeku Xi'an has received notices from some companies when recommending resumes to companies for former candidates: carefully meet candidates' salary expectations and raise recruitment conditions. In the view of another senior headhunter, this is inevitable, when enough resumes are submitted, "the recruiter wants to screen it".

Due to his wide network and good reputation, the headhunter received nearly 40 resumes on his hands, and the engineer's attitude has become polite compared with the coldness of a year or two ago, and he almost immediately turned the resume and added a "thank you". And what he is troubled about is no longer the previous inability to collect the contact information of these engineers, but how to "sell" these resumes, "there are not many companies with high matching".

Three ZEKU engineers with an annual salary of about one million introduced him to some engineers, and he had a bit of a headache, "I don't know how to arrange these three people."

Some long-term headhunters in the chip field are more optimistic about today's development trend, "short-term is a setback, long-term is the beginning of maturity and rationality", the heat of the past few years, so that more and more money and people poured into the chip industry. There are too many startups, too many headhunters, too few experienced engineers, and too few projects that can innovate and iterate for a long time.

Engineers and headhunters alike have seen the crisis, and an anxiety is spreading. Some headhunters who have changed careers to join the chip field are anxious, "will the industry go back to three years ago", and Zeku engineers are worried about whether the next job will have a salary cut. Everyone comforted each other, "Just as if a windfall money was issued in the past two years", because it will be difficult in the future.

IN THIS SUDDEN, HASTY LAYOFF, FEW EMPLOYEES ARE NOSTALGIC FOR ZEKU, AND FEW EMPLOYEES ARE BLAMING ZEKU. SOME ENGINEERS AFFIRMED ZEKU, THINKING IT WAS A GOOD JOB, WITH A LOT OF MONEY AND A LITTLE SENSE OF VALUE. Some engineers miss its working atmosphere, thinking that everyone is very motivated to work, and questions will be patiently answered, "probably due to high salary."

ZEKU's culture is more like the culture of a chip foreign company. Before joining the company, fresh graduates received a signed gift package from ZEKU, including a letter from Liu Jun, "Thank you for choosing to embark on a 'Mariana' adventure with us". During the epidemic in Shanghai, ZEKU distributed a large number of gift packages to engineers, and after working from home, in order to let engineers adjust their emotions, they also took two days off. These cultures make it easy for engineers at ZEKU.

It is undeniable that ZEKU's entry into SoC chips is a courageous challenge. SoC chip system integration and energy efficiency requirements are high, this high-specification chip allows engineers to contact complex IP, only Huawei has done in China, ZEKU's experience is very beneficial to some engineers.

Unfortunately, OPPO core-making, walking in with awe of semiconductors, eventually gave up. Without a clear market and a long enough time investment, even large factories with strong funds are not enough money. This painful lesson is just another education for companies that want to enter the chip industry to make cores.

(Wu Jingjing, Xu Yun, Fan Wenjing also contributed to this article)

Read on