OPPO's first self-developed chip MariSilicon ®️ X
The industry has not waited for OPPO's self-developed mobile phone processor to be mass-produced as scheduled. Recently, OPPO announced the closure of its chip design company Zheku Technology business.
As soon as the news came out, the industry generally felt stunned and regretted. OPPO's self-developed image NPU MariSilicon X released at the end of 2021 uses the 6nm process and enters the category of advanced processes. The Bluetooth SoC MariSilicon Y, released at the end of 2022, is the first self-developed SoC of the Zheku team. With the experience of operating advanced processes and SoCs, it seems logical to move towards self-developed mobile AP (application processor). In January this year, there was news in the industry that OPPO's self-developed mobile AP (application processor) will be tape-out in the second quarter of 2023 and mass production in the third quarter. OPPO said that in the face of the uncertainty of the global economy and mobile phone market, after careful consideration, the company decided to terminate the Zheku business.
"Cross-border core making" is no longer new. Mobile phones, home appliances, automobiles, and Internet manufacturers have successively entered the chip industry in order to reduce costs internally and seek profits externally.
However, companies in different fields face different core manufacturing situations. Internet companies have a large number of scenarios to undertake, as long as there is a demand for computing power, chips can come in handy. This is also why Baidu Kunlun Core and Ali Pingtou have launched general-purpose processor chips to cover more computing scenarios.
The "core-making" of terminal manufacturers such as mobile phones is mainly for self-use, and the purpose is to carry out differentiated customization according to brand positioning, and the sales volume is required to cover the funds of "core-making". And this is precisely the difficulty of mobile phone manufacturers today. Zheku has been established for 4 years, and 3 years have been under the influence of the global epidemic. Affected by multiple factors such as insufficient momentum for global economic recovery, sluggish demand for consumer electronics, and extended replacement cycles of mobile phones, global mobile phone shipments have declined to the lowest point since 2014. According to Canalys, global smartphone shipments in 2022 will be 1.19 billion units. OPPO shipments were 113 million units, down 22% year-on-year. In 2019, when Zheku was founded, global smartphone shipments were still 1.37 billion units, and OPPO shipments were 120 million units.
SoC design costs increase as process nodes scale
(in millions of United States dollars)
Source: International Business Strategy, AnySilicon, TSMC
Changes in the market environment make it difficult for mobile phone manufacturers to rely on sales to continue to transfuse the huge cost of "core making", and also reduce the cost performance of "core making". According to the observation of EDA leader Synopsys, the development of a chip involves the purchase of a content library composed of third-party IP, the purchase of EDA tools, the cost of chip manufacturing, time costs, order delivery and other expenses, the content library license cost can be tens of millions of dollars, EDA operation requires powerful computers and large-capacity storage, involving the cost of investing in local or cloud computing infrastructure.
At the same time, the more advanced the chip manufacturing process, the higher the design cost. The data shows that the design cost of 7nm chips is $298 million, and the design cost of 5nm chips reaches $542 million. Zheku's first self-developed chip uses 6nm, and the self-developed AP aims at a more advanced process, and the design cost and tape-out cost will increase accordingly. In addition, the chip industry is a typical high-paying industry, and the size of the Zheku team is more than 2,000 people, and as of April this year, it has been carrying out social recruitment and school recruitment, and the preparation and expansion of the talent team also need corresponding investment.
Even if the film is successful, it is only a ticket to the marathon of "core making", and it is a process that lasts for years, ten years or even longer to achieve the leap from "usable" to "easy to use", and continue to improve performance according to the positioning of subsequent models. Apple established the PowerPC Alliance with IBM and Motorola in 1991 to jointly develop and produce PowerPC series processors, and only launched the first self-developed mobile phone processor A4 in 2010, which is still tackling baseband chips. Samsung has both the dual identity of the world's top two mobile phone sales manufacturers and old chip foundries, and has launched its self-developed mobile phone processor Exynos since 2011, which has not covered all flagship models and still relies on third-party manufacturers to supply.
To win this protracted "core-making" tug-of-war, it is necessary to have stable investment before tape-out, a virtuous cycle of mutual promotion and improvement of chip performance and product performance after tape-out, and a high degree of coupling between chip design and product positioning. Since the chip research and development and iteration process will go through multiple industry cycles, there are extremely high requirements for the stability of enterprise capital flow, real-time adjustment of market strategy, ingenuity of product definition, and supply chain coordination and supporting capabilities. This is a war that requires a bold approach to the underlying core technology, and it is also a long march that requires calm study and judgment.
Although the Zheku team was disbanded, on the road to pursuing self-developed mobile phone processors, OPPO has reserved technology, accumulated algorithms, and has experience in team building and management, and has also tempered talents for the industry. Driven by OPPO, Xiaomi, vivo and other mobile phone manufacturers, self-developed image chips and power management chips once became the standard equipment of flagship phones, and differentiated competition around hardware has been fully rolled out, bringing consumers a more diversified product experience. It is also worth looking forward to the next market opportunity and the next industrial outlet, and technology companies with self-research awareness and technology accumulation will occupy more opportunities and produce more possibilities.
Arm personally made the "core"?
From ISP to AP, how difficult is it for mobile phone companies to advance in core manufacturing?
Author丨Zhang Xinyi
Editor丨Qiu Jiangyong
American editor丨Maria
Executive producer丨 Zhao Chen