Apple is worried and Intel has given up. Is baseband chip so difficult to make?

Countless facts tell us that baseband chips are indeed very difficult to implement, much more difficult than general-purpose CPUs.

Last year, the iPhone 11 was released. However, as Apple's main mobile phone version released once a year, there was no 5G version in the world at this time point. At the same time, there was news that Intel gave up the development of baseband chips, which really surprised many people. This is inconsistent with Apple's position as an industry leader in mobile phones and Intel's position as a giant in the chip industry.

Many Apple fans say that this is Apple's business strategy, and that 5G has not yet been built, so it is not too late to release the 5G version in 2020. I think this is all self-consolation. After all, when other major mobile phone manufacturers have launched 5G versions of mobile phones, and China Mobile has officially launched 5G commercial packages in 50 major cities on October 31, 2019, Apple has completely missed the first wave of 5G user market in China, and also missed the first wave of 5G user market in the world.

So what exactly is the baseband chip that Intel has abandoned that has made Apple cry?

What is a baseband chip?

The baseband chip can encode the baseband signal to be transmitted according to the wireless communication protocol, and then convert it into electromagnetic waves through the radio frequency circuit and send it through the mobile phone antenna; at the same time, the electromagnetic wave signal received by the antenna is decoded according to the protocol to restore the required audio or data. This is to perform forward and reverse protocol encoding and decoding.

The baseband chip is a SOC with a very complex integration. The mainstream baseband chip supports multiple network standards, that is, all mobile networks and wireless network standards are supported on one baseband chip, including 2G, 3G, 4G, 5G, WiFi, Bluetooth, etc. Only by integrating all protocols can seamless roaming between multiple mobile networks and wireless networks around the world be achieved.

To put it simply, the baseband chip is specifically responsible for wireless communication functions. Without the baseband chip, you cannot make calls or surf the Internet.

Those fallen baseband chip companies

More than a decade ago, many well-known chip companies sold smartphone CPUs, but eventually withdrew from the market. One of the important reasons was that they could not keep up with the trend in baseband chip development, and the 4G LTE baseband was a threshold.

These failed companies include: Texas Instruments, Marvell, Nvidia, Freescale, Broadcom, ADI... and so on. Even old communications companies like Nokia failed.

To date, the major baseband chip developers include Qualcomm, Huawei, Samsung, and MediaTek. Even Intel, the leader in the chip industry, announced its closure after releasing several unsuccessful baseband chips and was later acquired by Apple.

Among the remaining four companies, only Qualcomm and Huawei can produce qualified high-end basebands. Samsung and MediaTek only have a small share in the low-end market, which is already very difficult.

It can be seen that baseband chips are indeed a highly difficult technical job. Many great companies have tried it, but only a few have succeeded.

The protocol is extremely complex and the scale is extremely large

Netizens must not ignore complexity and large scale as difficulty. When the complexity reaches a certain level and the scale reaches a certain level, it will exceed the limits of what human companies can bear, and even exceed the affordability of some small countries, and it will become very difficult.

First of all, as a chip in a communication system, it must support all modes and all frequency bands, including 2G/3G/4G/5G, which are divided into many branch protocols such as GSM, GPRS, EDGE, TDSCDMA, WCDMA, HSPA/+, TDD-LTE, FDD-LTE, CDMA, and none of these standards can be left behind.

In addition, global frequency bands such as the Americas band, Europe band, and China band also need to be supported.

This is the list of main frequency bands specified by 4G protocol.

This is a list of global 4G frequency band usage. Since the list is too long, only a part of it is included. All three of these baseband chips must support it, otherwise you will not be able to use it when roaming in that country.

The above is only for 4G. What if 2G/3G/5G are added? Don’t forget that the complete set of documents for each protocol must be transported by car.

What is even more troublesome is that in addition to so many network standards, there is also the compatibility of different communication equipment. You don’t know what equipment the operator uses when building the network, such as Ericsson, Nokia Siemens, Alcatel-Lucent, Huawei, and ZTE. Therefore, the safest way is to support full device compatibility.

This created a very serious problem, requiring a lot of coding and testing work, with tens of millions of lines of code and hundreds of thousands of test scenarios and test cases.

Not to mention anything else, first of all, you need a R&D team of tens of thousands of people, all of whom must know programming, one thousand people who understand chips, and one thousand people who understand communication protocols. This will make it difficult for a lot of companies, and only a major technological power can organize so many computer professionals.

You may not feel deeply about this, so let me give you an example. More than 20 years ago, when Huawei was developing the 2G GSM protocol, the protocol documents were pulled over by a car and piled up in half a room. Everyone was dumbfounded when they saw it. It took four and a half months to organize dozens of people to understand the protocol, not to mention the development. It took two years until Huawei's first GSM call was connected, and thousands of people worked hard for two years. However, GSM is the simplest protocol, and it is simply incomparable to the complexity of the current LTE protocol, which is more than ten times larger. At that time, Huawei's 3G protocol development took more than 10,000 people five or six years. If Huawei had not maintained a research and development team of tens of thousands of people for more than 20 years, it would really collapse from working on these protocols.

Sometimes, scale and complexity can overwhelm a large group of people, which requires a company to have very high organizational capabilities and accumulated communication technology.

High performance requirements

As for being able to make baseband chips, some companies are able to do so, but it is very difficult to ensure that they are usable and easy to use.

It is far from enough for your chip to be able to make calls. Your performance must be ahead of your competitors. When the mobile phone signal is good, you must be faster than others. When the mobile phone signal is weak and others cannot make calls, you must be able to make calls easily. You must be faster than others to surf the Internet and have lower power consumption than others. Even if you have achieved all these and are ahead in performance, it is not enough. Your cost must be low enough.

The performance tuning of the baseband chip is a real professional job. It is not so easy to tune the quality of the mobile phone signal.

This is the electromagnetic wave beam from the antenna in the case of multiple reception and multiple transmission. Each petal must be adjusted to ensure the effective use of energy.

This is to choose different signal coverage solutions according to different scenarios.

This is to evaluate and optimize mobile signal coverage

I just randomly captured a few pictures for you to get a feel for. Wireless communication is a professional field that requires a lot of expertise and experience.

How strong the signal is when it passes through the woods and walls and enters the house, and what will it look like after refraction? How to deal with the signals that pass through the wall and the signals that come in through the window at the same time in the house? All kinds of coverage scenarios need to be adjusted to the best state.

Manufacturers without sufficient experience in mobile communications cannot do this well, such as Intel.

Competition is too fierce, if you don't succeed, you will die.

Competition in the baseband chip field is fierce, and only the top chips can survive. There are almost no market segments here, and it is the most competitive field with no room for luck.

Because baseband chip R&D is difficult, time-consuming and requires high investment, if the chip fails to gain sufficient market share after release, or fails to gain a global top three market share, it will basically lose money. How can a company persist?

You have finally made a baseband chip that is stable, easy to use, has top-notch performance, and has good cost control, but there is still one final hurdle to overcome, which is the high patent moat of the baseband. Qualcomm has done its best in protecting baseband patents, and companies with weak patent foundations are easily eliminated by Qualcomm.

So we need to understand why companies like Xiaomi don't want to make baseband chips at all, because even if they can make baseband chips, with Xiaomi's current patent strength, it will be abolished by Qualcomm in a matter of minutes. Please note that Pengpai is not a baseband chip, but a general-purpose CPU chip.

So why isn't Huawei afraid of Qualcomm's patents? Because Huawei has accumulated experience in the communications field for more than 20 years. In particular, as the world's number one communications equipment manufacturer, Huawei itself has accumulated a large number of communications patents. Huawei has to use Qualcomm's patents, but Qualcomm cannot avoid Huawei's patents. This creates a situation where you are in me and I am in you. When the strength of the two companies is similar, they will compromise with each other and authorize each other's patents, so that they can survive.

Huawei has ranked first in the world in the number of international patent applications for two consecutive years, ensuring its own security.

Let’s take another look at how Intel’s baseband chip died, and everyone will feel it more deeply.

As the leader in the chip industry, Intel's chip manufacturing capabilities are unquestionable, but baseband chips are different. Not only do you need to know how to make chips, you also need to understand mobile communication protocols. You may be confused about how Intel can understand communication protocols without any experience. The answer is that Intel doesn't understand them because it bought Infineon in 2010. Obviously, this has become the biggest shortcoming of Intel's baseband.

After switching to Intel baseband, Apple phones have been plagued by problems such as poor signal quality and degraded user experience, which have been frequently complained by users. However, because Intel itself does not have the ability to adjust mobile communication performance, or in other words, it does not have a deep understanding of mobile communication protocols, Intel basebands have no ability to resolve these user complaints, nor do they have the ability to further optimize chip performance. Apple just had a falling out with Qualcomm over patent fees, and it really has no baseband chips to use. It may be able to endure for a year or two, but it cannot see the future. If this continues for a long time, Apple's own market will be lost, and even if Intel is given another year, there is no hope of solving these problems. Apple cannot afford this consequence.

Intel's baseband has fully proved that baseband chips that cannot achieve leading performance cannot survive because they cannot occupy enough market share. Even if they barely enter, they cannot hold on. Mobile phones using your baseband chips are not competitive and cannot be sold. If the sales volume is insufficient, the huge R&D costs of baseband chips cannot be recovered and will lose money. The key is that if you don't understand the mobile communication protocol, you can't improve it, and if you can't improve it, you can't see the future.

What's worse is that Intel could still barely play in the 4G era, but in the 5G era, the complexity of the protocol is completely beyond Intel's capabilities, and the development progress has been delayed too much, and it can't keep up with the progress of Huawei and Qualcomm. This has led to Apple not having 5G chips available for a long time, and Apple had no choice but to go back to Qualcomm for chips. Qualcomm then settled the score and started to use patent weapons to attack people everywhere, which almost made both Apple and Intel cry. Intel had no choice but to give up.

Even a giant like Intel cannot afford the fierce competition in the baseband chip market.

In summary, baseband chips must have strong chip design capabilities, deep mobile communication capabilities, sufficient patent capabilities to protect themselves, and a lot of money and scientific and technological talent to support them. There are not many companies that meet all these conditions at the same time, so it is very difficult to make baseband chips.