In 2017, when Intel CEO Brian Krzanich (Brian Krzanich) was interviewed at the Code Conference, he expressed his views on the then-rumored Apple self-developed Mac processor.
He said: "As an engineer, if they don't try this, they must be stupid." The implication is that Apple will make its own computer chips sooner or later, and now his prediction has been confirmed. .
For Apple, the chip is the most important backing for hardware differentiation. Whether it is iPhone, iPad, AirPods, and the latest M1 core Mac, many of the selling points on these Apple devices are driven by different series of self-developed chips.
However, after the M1 core, Apple's self-research business has not ended, and they have already set the next goal.
According to a report from Bloomberg , Apple recently launched a self-developed baseband chip project internally to replace the current Qualcomm baseband on the iPhone in the future.
Unlike the well-known mobile phone processors, the baseband chip is related to the experience of mobile phone calls and networking speed, and directly determines the basic communication performance of the mobile phone.
▲ The Qualcomm SDR865 radio frequency transceiver (yellow area) on the iPhone 12 and the Qualcomm X55 baseband chip (green area). Picture from: iFixit
Although there are many chips developed by Apple, they are basically not involved in the subdivision of partial communication such as baseband. Previously, the baseband chips used in the iPhone have been provided by manufacturers such as Intel and Qualcomm.
Now baseband has become one of Apple's biggest headaches. The previous generations of iPhone have been criticized for weak signal and poor network stability. At that time, many people pointed the finger at the Intel baseband. Since then, almost every generation of iPhone will be complained of poor signal.
This year, the iPhone 12 has re-used the Qualcomm baseband, but for Apple it is just a "choice of no choice." In accordance with Apple's style, self-research and development of its own baseband is the most thorough solution.
Even if this is not an easy task.
Baseband development, nine deaths
What is the main focus of a baseband chip? Usually we use Speedtest to measure the speed and compare the upstream and downstream speeds. This is part of the performance side, but the integrity of the frequency band, the system, and the support are related to whether your phone can receive the signal normally.
Nowadays, if you just click on the introduction page of a mobile phone, you can see several rows of frequency band numbers in the column of network standard. Although we are talking about 5G now, this does not mean that mobile phones can abandon the previous 2/3/4G network, so for baseband, they must also be used for the frequency bands under the previous GSM, WCDMA and LTE branches. Do compatible.
How to ensure that the mobile phone can make calls and receive signals wherever it goes? The safest way is to support as many frequency band combinations in various regions and countries as possible to achieve seamless switching, and even global roaming. There are too many, and none of them can be left behind.
The terms 5 mode, 7 mode, 17 frequency and 19 frequency that we often hear refer to the network standards and the number of frequency bands supported by a mobile phone. Now that 5G is coming, the modes and frequency bands have increased dramatically, and sub- 6 and millimeter waves, obviously also bring more challenges to chip design.
This is not over yet, the next is the problem of base station equipment. Telecom operators in different regions and countries have different network deployment strategies. As a result, there is more than one brand of base station equipment, which may involve multiple brands such as Ericsson, Huawei, Nokia, etc. This requires solving the problem of baseband and different networking Compatibility issues between devices.
Second, the operator's network. More than one hundred commercial mobile networks in the world may require chip manufacturers to personally test them and then adjust them slowly based on data, which is a huge investment of manpower and material resources.
The complicated communication standards and the increasing number of frequency bands indirectly increase the difficulty and complexity of the development of baseband chips. In short, it not only tests the technological process or later mass production, but also values the accumulation of long-term experience. If you can't come up with a complete solution all at once, the mobile experience will ultimately be affected.
This is why it took 8-10 years for self-developed baseband manufacturers such as Huawei and Samsung to keep up with the progress of the first echelon. Now these barriers, Apple has to personally step over.
Patent in the hands of a few
Another issue involved in baseband chips lies in patents.
This has been fully reflected in CDMA in the 3G era. At that time, almost every manufacturer that made baseband chips had to purchase licenses from Qualcomm or Verizon (later sold to Intel), because most of the CDMA-related technology patents were in their hands, and without CDMA support, The inability to realize the full Netcom of mobile phones led to the fact that when we bought mobile phones that year, we often saw single-network versions such as the mobile version, the Unicom version or the telecom version.
▲ Picture from: Founder Securities-Semiconductor Industry Special Report
A report by Founder Securities also pointed out that although Qualcomm entered the WCDMA market a bit late in the 3G era, it paid more attention to the quality of patents rather than quantity. Therefore, with only 27% of WCDMA patents, it gained about 55% of the market. WCDMA market; in 4G, Qualcomm only owns 16% of LTE patent share, but at most occupies 96% of the LTE market.
In the face of such a strong patent pool, current mobile phone manufacturers naturally tend to use "Qualcomm Core." Excellent chip technology is definitely on the one hand, but on the other hand, buying a chip is tantamount to obtaining Qualcomm's patent authorization, and then getting the ticket to enter the mobile phone industry.
In other words, if Apple wants to develop its own baseband, it will be difficult to bypass Qualcomm.
Today's baseband chip market is completely oligopolistic. Many semiconductor manufacturers, including Texas Instruments, Nvidia, and Marvell, are all forced to withdraw from the pressure of this patent competition, or profit issues, or they are merged.
In the end, apart from Qualcomm, which occupies half of the country, and MediaTek and Ziguang Zhanrui, which do mid-to-low-end business, only Huawei and Samsung have strong financial resources.
▲ In 2019, Huawei released the Barong 5000 baseband for 5G
Huawei's baseband chips have also started from the 4G era. It self-developed the first TD-LTE baseband chip: Barong 700 in 2010, which got rid of its dependence on Qualcomm baseband, plus Huawei's emphasis on 4/5G patents, it is now one of the best in the world.
But as mentioned in the previous article, the 5G baseband is not only for 5G, but also for previous generations of compatibility, so as long as part of the 3/4G patent is still in Qualcomm's hands, then Huawei still has to pay a certain amount of patent license fees to Qualcomm. If you are a small manufacturer, the cost will be higher.
As for Apple, it chose Intel at the beginning. It also didn’t want to be controlled by Qualcomm. Therefore, it adopted a dual-vendor strategy at the end of iPhone 4G, switching to Intel basebands on some mobile phones, and then developing to switch to all of them. 5G has cultivated a solid ally.
But it is 5G that really makes Apple give up the Intel baseband. I can only say that Intel has more than enough energy.
In 2018, it was reported that Intel is making slow progress in the development of 5G basebands and cannot meet Apple's power consumption requirements. The worst result is that Apple cannot mass-produce the 5G version of the iPhone as planned. This is a very fatal problem.
As a result, we have also seen that after fighting with Qualcomm for 3 years, Apple still chose to compromise, and signed a 6-year patent licensing agreement and chip supply agreement with Qualcomm, while paying a large amount of fees. Because at this stage, both in terms of performance and patents, Qualcomm baseband is Apple's best choice, and it is also the only choice.
Only in this way can Apple keep the iPhone from falling behind other manufacturers in the early stage of 5G. In the long run, self-research is still the most suitable way out for Apple if it wants to leave Qualcomm.
Apple and Qualcomm have reconciled, and Intel baseband lost its largest customer. On the same day, Intel announced its withdrawal from the 5G baseband chip market. By July 2019, Intel packaged related teams and assets as a whole and sold it to Apple for $1 billion.
▲ The disassembly diagram of iPhone 3G that year, you can see the baseband chip supplied by Infineon
In fact, the Intel team also has a lot of connections with Apple.
In the first few generations of iPhone, Apple used Infineon's baseband chips before turning to the better-performing Qualcomm. Infineon lost its gold master and was acquired by Intel for $1.4 billion. Now, this team that integrates Infineon and Intel is in Apple's hands again, and I don't know if it is fate or coincidence.
But it is precisely because of this team that Apple can quickly cut into the development of self-developed basebands. Experienced talents can increase Apple's right to speak in the formulation of standards, and more than 17,000 baseband patents are expected to allow Apple to skip the starting point and directly enter the development of 5G baseband chips, shortening the time to market.
In addition, Apple's business model is also different from Intel and Qualcomm. The purpose of its self-developed chips is mainly for its own use, not for export. This is actually very similar to the situation of Huawei and Samsung. They are all using their own huge hardware sales to smooth out R&D costs, and then invest profits in the next generation of chips. In research and development, a virtuous circle is formed.
So the most critical question is, what can Apple's baseband land? If, according to Bloomberg, Apple’s internal R&D work has just started, then we may have to wait 3-5 years to see the results. Even in Apple’s style, the self-developed baseband has not reached the performance of Qualcomm's same-generation products Under circumstances, Apple will not let the iPhone use it rashly.
We might as well refer to the development path of Apple's M1 chip. The first news that Apple was going to make a Mac chip was in 2011. In 2014, MacRumors revealed that Apple was testing a computer prototype based on the ARM architecture; in 2017, Bloomberg’s Mark Gurman revealed more accurate information. He called Apple’s Mac. The processor project has been underway for almost a year.
A conservative estimate is that it took at least 4-6 years from Apple’s idea of building a Mac chip, to the establishment of the project, and then to the chip landing. The baseband chip is likely to follow a similar timeline, so you must wait until At the end of 5G, even when 6G is commercially available, we have the opportunity to see the appearance of Apple's self-developed baseband.
However, the returns of self-developed chips have always been very positive. Whether it is cost reduction or power consumption reduction, it will ultimately benefit a series of Apple hardware such as the iPhone. The improvement of the power consumption of the new generation of Mac by the M1 chip is the best example, and there will be great room for development in the future. .
Some industry analysts also believe that if Apple can complete the task of integrating the baseband in the SoC, then not only the iPhone, but also various miniaturized wearable products such as Apple Watch and AirPods in the future will also have the opportunity to move toward independent operation and achieve direct networking. , Instead of always relying on the iPhone.
How important the self-developed core is can be seen, and the baseband is the same. Perhaps only time will tell Apple whether this investment is worthwhile.
Source of title picture: Galus Australis
#Welcome to follow Aifaner's official WeChat account: Aifaner (WeChat ID: ifanr), more exciting content will be provided to you as soon as possible.