In 2020, “cutting supply” of Huawei will cause headaches for Qualcomm and Intel, and as Apple, Google, etc. have confirmed or been rumored to develop their own chips, their headaches will be even worse.
In 2020, Huawei encountered the biggest difficulty on the road of self-development of chips. The supply of upstream core components and advanced processes was cut off. Now, it can only buy chips from Qualcomm, which does not include 5G chips. Obviously, because of a ban, Huawei was “trapped” in 4G.
From this experience of Huawei, a point of view has been confirmed again: mastering core technology can not be controlled by others.
In the industry, the incident of Huawei’s entrapment has also become a fuse, forcing more companies to think about the issue of “autonomy”. In fact, as early as the past few years, several major manufacturers including Apple have had the idea of ”chip self-development”, but in this context, the Huawei incident has further heightened this concern.
With Apple taking the lead, Microsoft and Google join the “self-research army”
On June 22 this year, due to the epidemic, Apple moved its global developer conference online for the first time. It was also at this conference that Apple officially announced that Mac computers would switch to using self-developed ARM architecture chips, saying goodbye to its 15-year partner Intel.
On Double Eleven, when everyone in China was busy making payments and emptying shopping carts, Apple also officially unveiled its first self-developed Mac chip, as well as a Mac chip equipped with an M1 chip. From the parameters to the subsequent evaluations, they all communicate a message to the outside world: the new generation of Mac is very “fragrant”! The M1 is very “strong”! The system is very “nice” to operate!
Surprisingly, not only Apple, but in December, news of Google and Microsoft’s “self-developed chips” also began to appear in media reports:
On December 10, foreign media Axios reported that Google’s SoC chip codenamed “Whitechapel” has been successfully taped out. In addition to being used in its own Pixel phones, this chip may also appear on notebook Chromebooks in the future;
On December 21, Bloomberg reported that Microsoft is designing its own ARM-based processor chips for its servers and future Surface devices.
In fact, it’s not just Google and Microsoft. Just a month after Apple announced it was ditching Intel chips, Samsung was rumored to be working on an Exynos 1000 chip, which could potentially be used in upcoming Windows PCs.
Huawei’s bloody warning is ahead. Self-developed chips means that they can largely get rid of their dependence on Intel, Qualcomm and other manufacturers, and are not subject to the delay of the latter’s process and the delivery of products.
At the same time, in addition to enhancing product autonomy, the benefits brought by self-developed chips also include cost and innovation.
Take Apple as an example. Sumit Gupta, head of IBM AI strategy, once calculated that the unit production cost of Apple’s M1 chip is between $40-50, and the cost of purchasing an Intel Core i5 processor is $175-250. Around $2.5 billion a year, when Apple replaces Intel processors with M1 chips in the new MacBook Pro, MacBook Air and Mac Mini, it will save $2.5 billion.
Another example is innovation. Apple relies on A-series chips to prop up the iPhone and iPad product system, and Huawei takes off rapidly with Kirin chips… If it is only based on general or Intel’s general chip solutions for redevelopment, it is difficult to make a difference at the bottom. competitive. On the contrary, self-developed chips can combine the performance requirements of their own products to the greatest extent to create suitable chips, carry out technological and functional innovation to the greatest extent, improve user experience, and at the same time form differentiated competition with other competing products.
Whose cake was moved behind the self-research of the big machine manufacturer?
As we all know, the chip is a high-threshold track. Large development investment, long profit cycle, and high failure probability can be used as a summary of the chip industry from the outside world.
In the past mainstream models, major machine manufacturers such as Apple, Google, and Microsoft mostly purchased general-purpose chips directly from Intel, Qualcomm, and Nvidia, and then developed and built targeted frameworks based on their own product systems and ecosystems. However, as the complete machine manufacturers embarked on the road of self-research one after another, this product model has to evolve this year.
In the latest evolution of the chip supply model, complete machine manufacturers have partially replaced the original position of IC design companies. From the perspective of market share and revenue, this impact on third-party IC design companies cannot be underestimated.
Take Qualcomm, for example, which has held the top spot in the smartphone processor market in the past. Before adopting Kirin chips, Huawei was also one of Qualcomm’s important customers. With the full adoption of Kirin chips in Huawei’s products, Qualcomm’s Snapdragon processors have lost “sales” on Huawei’s smartphones.
How big is the loss? The following data set is enough to illustrate:
A Qualcomm briefing cited by a report mentioned that the export ban would not allow Qualcomm to sell chips to Huawei, which would cede the $8 billion-a-year market to other competitors. $8 billion, which is equivalent to 34% of Qualcomm’s total revenue in fiscal 2020 ($23.531 billion).
Another example is Intel. Based on Gupta’s bill, we learned that after losing Apple as a customer, Intel lost an order worth at least $3.5 billion. In the server market, Microsoft and Intel’s “Wintel” combination can be described as sweeping the world. Today, Microsoft has announced a chip self-research plan, and it has joined the ARM camp. For Intel, the big machine manufacturers have fled its ecology, which is a major setback.
2020 will be over, the ultimate two questions about the self-developed chips of the big machine manufacturers
As mentioned earlier, the chip supply model of the big machine manufacturers is evolving. When the self-developed chips of the big machine manufacturers start to be applied, it means that they no longer need to buy chips from companies such as Qualcomm and Intel, which cannot be ignored for the latter.
In this future conjecture, we cannot help but ask two questions:
· Will self-developed chips by major manufacturers become a trend?
We have already analyzed the reasons why the big machine manufacturers embarked on self-research and the advantages.
In fact, the intention of the big machine manufacturers to develop their own chips has been revealed one after another a few years ago, not only foreign Apple, Google, etc., even in China, in addition to Huawei, which has successfully built chips, including Xiaomi, OPPO and others all have chip self-research plans.
However, in previous years, the “voices” of larger chips were mostly concentrated in cloud chips, especially cloud AI chips. It is still relatively rare to report on manufacturers’ self-developed terminal chips as intensively as this year.
It is undeniable that the development and implementation of chips are not achieved overnight. Even Apple, with the experience of building several generations of A-series chips, has taken at least two years for the M1 chip project to be officially released, and the follow-up still needs to experience For a transition period, leave time for a large number of apps to transform and run-in chips and systems.
However, judging from the current situation, as far as machine manufacturers are concerned, Apple, Huawei, and Samsung are the first cases, and the benefits that self-developed chips will bring are also visible to the naked eye. That being the case, on the condition that the money is not bad, Jianshui has already been moved a few years ago, and it is time to make a foray.
Not bad for money, chip products also have product applications, not to mention the long-term, in the short term, the self-developed chips by major manufacturers of the whole machine will form a trend and become a trend.
· Will third-party independent IC design companies continue to exist?
If classified, IC design companies can be divided into multiple categories, such as HiSilicon, which is dominated by big machine manufacturers, and Qualcomm, which has core technology and focuses on providing product solutions for external customers, and Enterprises that use third-party tools and lack core technical support.
In the wave of self-developed chips by big machine manufacturers, companies such as HiSilicon need not worry too much. Their main products are provided to the parent company, and there is no need to worry about sales. The only thing to worry about is the stability of the supply chain such as the supply of core components.
As for the likes of Qualcomm, it is undeniable that the impact of the self-development of the whole machine manufacturers cannot be ignored.
In terms of product revenue and market share, Qualcomm will definitely be affected. But you don’t need to worry too much, even if the whole machine manufacturers develop their own, they will only be squeezed in some chip markets. At the same time, even if they do not build chips, relying on years of patent accumulation, etc., Qualcomm will not lose too much in core technology.
For example, Qualcomm, in addition to relying on chip sales to obtain income, also obtains patent royalties from customers at the same time – charging a patent fee equivalent to about 5% of the selling price for each mobile phone that has been sold, which is called “Qualcomm” in the industry. Tax”. For most mobile phones, the “Qualcomm tax” is an unavoidable hurdle.
As for enterprises that use third-party tools and lack core technical support, they face much greater risks than the former two. Originally, because of the use of third-party tools and technologies to build chips, they were much weaker in confidence, and because there were many competitors, after chip products entered the market, in order to win more orders, they also needed to increase their price. concession.
When complete machine manufacturers start their own research and development, the originally small market will be further compressed. At this time, “sales ability” may replace technical strength and become the key to their continued survival.