OPPO Chip Manufacturing: Long-term and Short-term Concerns on the Edge of Top 5

The phrase "be Chinese people's own XXX" has always been able to penetrate industry barriers and resonate with public opinion. Chips are a representative example. From the Hanxin and Ark incidents in the past to the pressure on ZTE and Huawei today, 20 years have passed and the people are still unhappy.

It cannot be said that there is no achievement. With the huge demographic dividend, domestic mobile phones have achieved a curve overtaking in the application innovation level. In just a few years, they have accumulated huge market advantages and used this as capital to increase investment in the basic technology field of chips.

But to use a common saying, there is no harm without comparison. For example, Huawei HiSilicon has the strength to compete with Qualcomm in high-end products; for example, Xiaomi Pinecone has disappeared after installing the Surge S1 into the Xiaomi 5C three years ago.

I think everyone knows that mobile phone chips are a pitfall. Huawei and Xiaomi are not the only ones who have fallen into the pit, but OPPO is also trying to get involved.

Let's review the events. First, on the evening of February 16, the media reported that the special assistant to OPPO CEO released an internal article titled "Some Thoughts on Building Core Technologies", which proposed the "Mariana Plan" involving software development, cloud, and chips, which then sparked heated discussions in the digital technology circle.

On February 18, OPPO officially responded to the report, saying that "the core strategy is to make good products, and any R&D investment is to enhance product competitiveness and improve user experience", which was considered a positive affirmation of the rumor.

Compared with the plan itself, the outside world is no less interested in the name "Mariana". Naming it after the "Mariana Trench", the deepest part of the ocean known to mankind, at least shows that OPPO has not underestimated the difficulty of "making chips".

So the question is: Why this plan? And why this time?

OPPO's "battery life anxiety"

“Charge for 5 minutes, talk for 2 hours”, OPPO, which has solved the battery life problem of mobile phones, is actually facing battery life issues at the brand and enterprise levels. I am certainly not saying that OPPO, the world's fifth largest mobile phone manufacturer, has reached the edge of the cliff.

But if we sort out the TOP5 manufacturers, we can see that Samsung has Exynos, Huawei has HiSilicon Kirin, Apple has the A series, and Xiaomi's Pinecone has at least made products. Only OPPO still has zero in the field of self-developed chips.

The year-on-year growth is even more interesting. Samsung, which accounts for more than 20% of the market, can still maintain a slight positive growth, while Huawei is closely following with a double-digit growth. Apple's decline is a commonplace, Xiaomi can also record 5.5%, while OPPO's growth rate is less than 1%.

What do these numbers mean? Under the premise that the overall market has only increased by -2.3%, manufacturers with larger growth rates such as Huawei and Xiaomi will most likely continue to eat up more market share by inertia. The first to bear the brunt will naturally be OPPO, which has a low market share and growth rate.

In other words, OPPO, which does not have the ability to develop its own chips, is very likely to fall out of the TOP5. However, this still fails to explain the underlying logic, that is, can the addition of self-developed chips improve OPPO's competitiveness?

In The Innovator's Solution, Christensen once prescribed a prescription for enterprises to promote disruptive innovation: choose the right production architecture. The so-called production architecture is roughly divided into modular and integrated types. Among them, the integrated type is completed by one company for most or even all core links, thereby breaking the constraints of the value network to achieve differentiation and innovation, which is more suitable for the mature stage of the industry.

So, to answer the question above, Huawei is a typical case. For example, last year's flagship Mate30 Pro, with its self-developed SoC, not only became one of the earliest flagship models to support NSA/SA dual-mode 5G; but more importantly, in the field of IC design, self-developed micro-architecture is the embodiment of strength, and HiSilicon has started to increase its investment since Kirin 950;

This is why the Mate30 has enhanced imaging performance by leveraging the powerful throughput performance of ISP5.0, achieving 7680 frames of slow motion, supporting 4K time-lapse photography with defined single-frame exposure parameters, and stronger ultra-dark light noise reduction, greatly extending its "long board" beyond its competitors.

In other words, in the competition of the stock market, the demand for the combination of software and hardware has further increased. It is difficult to achieve brand premium by relying solely on the original "hardware procurement + software innovation", and the pace of R&D and mass production will be restricted by upstream chip manufacturers.

In fact, OPPO's exploration into upstream fields is not without trace.

The earliest can even be traced back to 2017. At that time, OPPO CEO Chen Mingyong and Duan Yongping successively invested in chip company Xiongli Technology. The founder of the latter is an expert of the national "Thousand Talents Plan", and most of the team members are from Huawei, Cisco, etc. But in the end, nothing happened. Only at the end of the same year, "Shanghai Jinsheng Communication Technology Co., Ltd." was established in Shanghai.

In September 2018, Shanghai Jinsheng included "integrated circuit design and services" in its business projects. In November of the following year, OPPO announced that it had signed VOOC flash charging patent licensing agreements with six manufacturers, four of which were chip design companies, and OPPO was clearly accelerating.

At OPPO's first Future Technology Conference at the end of 2018, founder and CEO Chen Mingyong also said: "OPPO's R&D investment next year will increase from 4 billion to 10 billion, and will increase investment year by year in the future."

This is indeed reflected in the actions taken in 2019. In January, OPPO joined the WPC Wireless Charging Alliance; in August, Jiwei.com reported: Several chip industry practitioners revealed that "OPPO recently released a lot of chip design engineer positions. They set up a company in Shanghai and poached a group of grassroots engineers from Spreadtrum and MediaTek."

At the same time, many Zhihu users in the chip industry said they had received a favorable offer from OPPO. In November, Dutch technology media LetsGoDigital reported that the European Union Intellectual Property Office had approved the trademark application for "OPPO M1", and its description showed that the scope included chip integrated circuits, semiconductor chips, electronic chips used in integrated circuit manufacturing, and smartphones.

It can be seen that OPPO's decision to enter the field of self-developed SoC is not rash. Instead, it is building a research and development team while trying to lay out an ecosystem through standards and patents.

The long-term and short-term concerns behind chip manufacturing

Since we are talking about "Chinese chips", we can't avoid Huawei and Xiaomi. In my opinion, they just represent the two starting points of domestic mobile phone manufacturers to develop their own SoC: long-term concerns and short-term concerns.

In my opinion, Huawei is a typical long-term thinker, not to mention the "spare tire" that came later. On September 22, 2012, Yu Chengdong's Weibo post was very representative, directly pointing out several specific adjustments to Huawei's self-developed SoC:

Before that, Huawei's mobile phone business was actually a typical chicken rib: although it could be included in the so-called "China Cool Alliance", it essentially relied on deep binding with operators; and HiSilicon before, although it had some gains in the chip industry, it was more focused on the security chip field.

Therefore, the "foresight" is reflected in giving up the stable market and dying if they don't succeed. Even though K3V2 received a lot of negative reviews, it still forced to promote its own processors in high-end products at a huge cost, and gradually won back the reputation of Kirin 910, Kirin 920, etc.

Of course, this is indeed inseparable from the window period that Qualcomm won due to the poor performance of its products during the same period, as well as the market situation that was already in a chaotic period at the time; but it is mainly due to the huge investment in research and development over the past decade, which is rumored to have exceeded 480 billion yuan, keeping SoC at the strategic level.

Xiaomi, on the other hand, is more concerned about the immediate future. The first Xiaomi 5C equipped with the Surge processor went on sale on March 3, 2017, but was removed from the official website in October of that year, with a life cycle of only seven months. Since then, the Surge processor has never appeared in Xiaomi phones again.

The last time news about Pinecone appeared in Xiaomi's official channels was at the Xiaomi IoT Developer Conference in November 2017. At that time, Pinecone was already in the product line of NB-IoT modules and was expected to be released in Q2 2018, but there has been no further news about this product so far.

Then on April 2, 2019, Xiaomi Group's Organization Department announced the developments of its chip company Pinecone. It had nothing to do with the new generation of SoCs, but rather the reorganization of the Pinecone Electronics team: part of the team was split off to form a new company, Nanjing Dayu Semiconductor, and financed independently.

The official statement at the time was "to cooperate with the company's AIoT strategy to accelerate implementation"; but in the Sino-US entrepreneur dialogue in November 2017, a memorandum of intent was reached with Qualcomm to purchase components worth US$12 billion within three years. This to a large extent caused Pinecone, which had been repeatedly frustrated in research and development, to further miss out on market space.

In short, the difficulty of "making chips" lies in technology. On the one hand, the difficulty of SoC lies not only in the CPU, but also in the baseband. Huawei has a foundation in communications, Xiaomi relies on the deep binding of Unigroup, and even Apple, as strong as it is, has to rely on Qualcomm, and even acquired the baseband part of Intel, which has a bad reputation, before it dares to start self-research. This is for long-term consideration.

On the other hand, in the market, upstream baseband manufacturers are often also SoC manufacturers. For example, Qualcomm has a joke of "buy a baseband and get a CPU for free". From a commercial perspective, it excludes the sale of baseband alone. Once there are those who do not believe in it, they will often be "choked" by Qualcomm, which will punish them by giving the first launch of chips to competitors and controlling supply. This is a short-term concern.

Is it too late for OPPO to make chips?

Finally, let's talk about OPPO's "chip manufacturing". There are two voices that can be heard at present. One is the slogan-shouting type, which labels OPPO as "the pillar of the country" and "courageous"; the other is the pessimistic type, which believes that chips are a bottomless pit.

In fact, neither one of them is quite right. Many people who shout "China Chip" seem to always have a misunderstanding, that is, as long as they invest heavily at all costs, they will definitely be able to create a completely independent chip.

However, if we look back at the history of semiconductor development, we can easily find that the semiconductor industry has a trend of increasing concentration. Statistics show that in 2001, there were as many as 29 top chip manufacturers in the world, but now there are only five.

Where did the other 20 companies go? Take X86 as an example. Companies such as Transmeta, Cyrix, and VIA once occupied a large market share. Even IBM and Intel have launched many products that failed in both R&D and marketing. The investment in chip R&D and production is getting more and more expensive, and the technical barriers are getting higher and higher. In the end, it naturally became a game for big companies.

This is the rule of the semiconductor industry: the existing processes, ecology and structure are the result of gradual exploration after a large number of projects have failed and after continuous trial and error in technology and the market. This must be accompanied by huge R&D and production investments.

As for the events like Hanxin, Fangzhou and even Xiaomi Pinecone, there are countless of them in the United States, Japan and South Korea. It is precisely because a large number of such "front waves" were washed up on the beach that the "back waves" have a polished and mature path of design, manufacturing, packaging and testing.

A failed tape-out and rework means that all the previous investment has gone to waste. Therefore, in the semiconductor industry, the development process is recognized as the greatest asset. Scientific and strict development specifications and processes are the core of accumulating experience and accelerating R&D.

The reason why the semiconductor industry has become the "foundation" of many industries such as consumer electronics, Internet/mobile Internet, IoT, etc. in just a few decades is that it has continuously reshuffled through the cruel mechanism of survival of the fittest while providing the industry with a large amount of experience and cases, which has enabled the competitiveness of the strongest companies that remain to continue to improve.

So it is not surprising that they do not have their own SoC. This is just like the fact that foreign teams cannot beat the Chinese team in table tennis, it is a system problem.

Obviously, the more reasonable route is not to have to bypass the numerous patents from scratch. Instead, it is to pay the licensing fees and learn as the leaders of the semiconductor industry should. For example, HiSilicon Kirin, which is now gaining momentum, also pays money to ARM honestly.

Then, on this basis, we gradually accumulate experience, seize key nodes and seek opportunities. For example, Huawei HiSilicon's independent architecture takes imaging as a breakthrough point. Starting from Kirin 950, it integrates self-developed ISP to improve imaging quality from the bottom up, and finally pushes Huawei's flagship camera to the top of the world, giving it the confidence to feed back R&D with the market.

For OPPO at this time, it is not too late to develop its own SoC.

When Texas Instruments was dragged down by baseband and finally withdrew from the SoC market, there was an important background: at that time, mobile Internet was in its infancy, and the demand for basic communication performance such as calls was extremely high, and it was difficult to see other breakthroughs. However, when Huawei started, the development of mobile Internet and the rigid demand for photo sharing further increased, in a sense, created a "curve" for HiSilicon to overtake.

For OPPO, perhaps it is relying on a relatively higher premium and nearly 120 million shipments to cooperate with Qualcomm on baseband. Then, taking advantage of the high-definition audio and video dividends brought about by the 5G technology wave, it breaks through based on the advantages of audio and video accumulated in the past at the Blu-ray machine level. Judging from the various audio and video patents that Shanghai Jinsheng has applied for, this may be the case.

Going back a little further, Yu Jun, the father of Tieba, proposed a "user value formula": user value = (new experience - old experience) - replacement cost. In other words, as long as the experience upgrade brought by the new experience is large enough, it can always bring reorganization value to users. At least in this regard, there is still a lot of room for imagination for 5G.