For well over a decade, Qualcomm and MediaTek have been the default choices for the best smartphone makers. Almost every smartphone or tablet you see is powered by either a Qualcomm Snapdragon or MediaTek chipset. But the tremendous growth in the performance of mobile devices over the last few years implies that even the slightest of differences in performance can significantly impact a buyer’s decision. Companies such as Apple have played this to their advantage by hailing the performance of its custom chips over counterparts from Qualcomm and MediaTek.
With swarms of phone manufacturers exploring ways to design their own custom smartphone chipsets in a bid to offer a customized experience to the users. the phenomenon appears to be on the rise. Will Qualcomm and MediaTek succumb to this pressure from companies designing their chips in-house?
Samsung and Huawei have been making chips for a while
Smartphone brands already design their own custom ARM-based chipsets — although at a relatively smaller scale than Qualcomm or MediaTek. Samsung is unquestionably the first brand that comes to mind when we think about phone brands that design or even develop their own processors. Another phone brand that does so is Huawei, which is relatively lesser-known for its in-house chips.
Even though one is most likely to recall Samsung’s Exynos lineup of smartphone chips, the company has been making processors for electronic devices like handheld multimedia devices, personal digital assistants, and candy bar phones for more than two decades. In fact, the first Apple iPhone used a Samsung processor — although the Exynos series had not been conceived until then. It must be noted that Samsung’s mobile and semiconductor divisions are separate entities operating under the same brands, and Samsung Semiconductors, which manufactures the Exynos line of chipsets, also manufactures the Snapdragon 888/888 Plus and the latest Snapdragon 8 Gen 1 system on a chip (SoC).
The Exynos branding was first introduced in 2011 with the Samsung Galaxy S II, which ran on an Exynos 4210, a dual-core processor that featured ARM’s Cortex-A9 cores and was built on a 45nm process. Fast forward to early 2021; Samsung announced the Exynos 2100 just two days before the Galaxy S21 series was officially released. The Exynos 2100 was praised for being the first ARM chipset from Samsung in several years that matched up to Qualcomm’s flagship Snapdragon 88 chip in terms of performance, but its limited availability on just Samsung’s flagship series has precluded it from dominating Qualcomm or Snapdragon, and the same goes for Huawei’s HiSilicon Kirin chips.
Apple fuels the buzz around custom silicon
In 2020, Apple sparked the buzz around “custom silicon” as it broke ties with Intel for its Mac lineup. The company revealed Apple M1, the first ARM-based chipset for Macs, which consolidated the CPU, GPU, memory, and storage on a single chip. It’s only fair to say Apple broke the internet by choosing a custom solution and ending its dependency on Intel. The Apple M1 chip has been seen to offer much better performance and thermal efficiency than the latest Intel x86-64 chips used in the previous Mac machines. The upgraded versions — Apple M1 Pro and the M1 Max — that launched with the latest 2021 14-inch and 16-inch MacBook Pro models are much more efficient in terms of CPU and GPU performance, and offer up to 64 gigabytes of bundled RAM.
While the performance and power efficiency of Apple’s custom silicon chips is something to be envied — and is being attempted to be replicated by Qualcomm through its ARM solutions — the biggest takeaway from this announcement appears to be the marketing of custom chips. Besides Chinese companies that include Oppo, Vivo, and Xiaomi, Google also happens to be riding this wave and has announced its entry into the world of custom silicon solutions, beginning with the Google Tensor chipset announced earlier this year.
Meanwhile, Apple has also been designing its custom A-series chips for iPhones and iPads and is also exploring options to manufacture its own 5G modems after acquiring Intel’s chipmaking business.
Google Tensor inspires other manufacturers
In 2021, Google launched its custom silicon Tensor for the Pixel 6 and the Pixel 6 Pro. The chipset shares its name with Google’s Tensor Processing Units (TPUs) used inside the giant’s data centers. The Tensor SoC is designed by Google for Pixel smartphones and is produced in partnership with Samsung. While based on the Exynos platform, the chipset is the first non-Samsung chip to use an Exynos 5G modem, and is designed to cater to on-device A.I. and machine learning capabilities. As per Google, the objective behind the custom chipset is to bypass the limitations of chipsets from Qualcomm and enable workloads that were previously not achievable on “standard” mobile platforms by other silicon manufacturers.
Before the Pixel 6 launch event, Qualcomm could be seen letting off steam by subtweeting.
"We've decided to make our own smartphone SoC instead of using Snapdragon" 🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩
— Snapdragon (@Snapdragon) October 13, 2021
Unlike any other flagship chipset from Qualcomm, MediaTek, or brands like Samsung and Huawei, the Google Tensor uses two ARM Cortex X1 cores for high-performance threads to satiate the requirement for demanding A.I., as well as machine learning applications. This custom solution also allows Google to use low-powered and older-generation middle cores in an attempt to keep power demand and heat dissipation under control — although there have been numerous complaints about overheating. The chipset also uses a custom 20-core Mali-G78 GPU to prioritize graphics tasks such as instant photo and video processing as well as AR applications.
Chinese companies vie to make in-house chips
Chinese smartphone brands such as OPPO and Xiaomi have also been vying to replace Qualcomm and MediaTek chips with their own custom silicon. Xiaomi launched its first custom chipset, called Surge S1, for the higher midrange of smartphones. The chipset was specifically designed for the Xiaomi Mi 5C. Although structurally similar to Qualcomm’s Snapdragon 660, the Surge S1 chip featured a custom Mali GPU and was primarily designed to offer a more optimized software experience on Xiaomi’s custom Android skin, MIUI, and to help it cut its dependence on Qualcomm and MediaTek, especially for devices to be sold in mainland China.
The Surge S1 was to be followed by the S2, but the proposed successor never materialized, and Xiaomi eventually abandoned (via Weibo) its endeavors to make a custom chip. However, a year after it reportedly pulled the plug on its custom chipset lineup, Xiaomi seems to be working on rebuilding a team and seeking IP for its next chipset, as per sources from China (via Android Headlines).
At the same time, Oppo has also been reported by Nikkei to be working with Taiwan’s TSMC to release custom chipsets in 2023 or 2024. The company has already embarked on its custom silicon journey with its custom MariSilicon X neural processing unit (NPU), which was showcased at the Oppo Inno Day event in December. MariSilicon X essentially combines the NPU with the image signal processor and features a dedicated memory subsystem. All of these components combined will give Oppo phones the ability to process HDR in 4K video recording and A.I.-related camera optimizations in real time.
The first device to feature Oppo’s custom MariSilicon X chip is the Find X4, which launches in the first quarter of next year. Meanwhile, because Oppo shares its ownership with other prominent phone brands including Vivo and Realme — and has recently been remerged with OnePlus, we can expect to see more devices from these brands also benefiting from Oppo’s developments.
Along with Oppo, Vivo has also been weighing in on different Snapdragon alternatives. Last year, it announced the Chinese versions of the Vivo X60 and the X60 Pro with Samsung’s Exynos 1080 chipset. Meanwhile, Vivo also announced its custom V1 image signal processor (ISP) for its flagship Vivo X70 Pro Plus. The custom ISP — in conjunction with Zeiss Optics — is claimed to offer professional-quality photographs with some A.I. processing on board.
Looking at the legion of Chinese smartphone companies, it is fairly evident that they are trying to reduce their dependence on Qualcomm and MediaTek. It’s natural to question why that is so important. And to answer that, we have two potential explanations for why it is a must for companies to have capable firepower against Qualcomm.
Evading Qualcomm’s monopoly
For many years, Qualcomm and MediaTek have been the frontrunners in terms of the chipset market share. Of the two leading chipmakers, Qualcomm has had a stronger hold, and it’s easy to correlate this dominance with the historic notion about its chipsets outperforming the competition, including MediaTek, Samsung’s Exynos, and Huawei’s HiSilicon Kirin. However, the story has more caveats than one would naturally imagine.
Before we go over those caveats, here’s a brief history of Qualcomm making its way into the chipset market. The chipmaker began its journey in 1985 as a satellite communications and telecom equipment provider. It invested hugely in revamping and marketing CDMA wireless technology and collected well over 100,000 patents for wireless communications via phones. Qualcomm also developed CDMA-based handheld phones in the late 1990s but eventually sold its business to Japan’s Kyocera Corporation to cover losses.
Qualcomm actually became relevant in the chipset industry with its introduction of the Scorpion series mobile chipsets in 2007. This series is what we know as Snapdragon now. By adopting an SoC design, Qualcomm bundled its baseband processor — or modems — into the mobile chips that it eventually began selling to phone manufacturers. Given Qualcomm’s control over the CDMA standard, it was able to employ aggressive licensing strategies with smartphone companies, including single-handedly controlling the terms for licensing deals and threatening to cut off supplies abruptly. These tactics were identified as part of an antitrust ruling against Qualcomm in 2019 — a case that had initially begun with the Federal Trade Commission (FTC) suing Qualcomm for anticompetitive practices in 2017.
Days after FTC’s lawsuit against Qualcomm, Apple sued Qualcomm for another $1 billion for not paying the promised rebates in licensing fees if it chose Qualcomm exclusively for its modems. Apple also highlighted the latter’s “no license — no chip” policy, which implies phone companies had to agree to pay license fees for its entire portfolio of patents.
Instead of just selling their SoC’s, Qualcomm arranged licensing agreements for the use of CDMA-based wireless technologies where licensees had to pay a fee based on the price of the handsets. Qualcomm made a good profit overall irrespective of whether it was only selling these components as part of an SoC or not.
Despite its legal binding, Qualcomm refused to share its patents with competitors, including Intel. This meant that Qualcomm was for several years the sole provider of CDMA modems and forced companies including Apple to rely on it for these chips. Incidentally, it shared its patents with smaller — at that time — competitors including MediaTek. But since Qualcomm was the leading manufacturer of high-end chipsets during the early smartphone years, smartphone companies had to rely on it for the entire SoC, and not just the modem.
In some other cases, Qualcomm also offered discounts in its licensing fees to smartphone companies if they agreed to use Snapdragon chipsets on at least 85 percent — and in certain cases, 100 percent — of their smartphones. Besides the U.S., these monopolistic practices got Qualcomm on the radar of — and hefty fines from — anticompetition agencies in Korea, Japan, China, Taiwan, amd the European Union.
With the 2019 ruling against Qualcomm, the chipmaker was forced to renegotiate its licensing terms with most phone makers without threatening to cut off supplies for modems.
However, with Qualcomm still supplying over 50 percent of the chipset modems (as of 2018), its dominant monopoly is difficult to refute if companies keep relying on the Snapdragon lineup. Therefore, using custom solutions is a solid mechanism to skirt and thwart Qualcomm’s stronghold.
Fear of U.S. government
While trying to fight its monopoly is surely one of the reasons why smartphone companies are preparing to stock up against Qualcomm, the U.S.-China trade war might have also led phone companies — especially Chinese ones— to be prepared for any sudden action against them. One of the major blows in the trade war came when the U.S. government under former president Donald Trump issued executive orders barring Huawei from engaging in business with any American company on grounds of national security.
Huawei, which was barred from using Android or engaging in commercial activities with companies like ARM, was suitably prepared to mitigate the impact of this blow because of its custom Kirin lineup of mobile chipsets. In the face of such adversity, business would come to a halt overnight for any other company that does not have its own range of chipsets ready to replace Qualcomm’s offerings.
MediaTek, Samsung close in on Qualcomm
Despite Qualcomm’s early advantage in mobile computing, competitors MediaTek and Samsung have been closing in on it in terms of chipset performance. Samsung announced its Exynos 2100 earlier this year with the Galaxy S21 series, which boasted performance and power efficiency on par with Qualcomm’s then-latest flagship chipset, the Snapdragon 888.
Later on in the year, MediaTek also announced the Dimensity 9000 chipset — its greatest flagship chipset in many years — that is set to compete with the Snapdragon 888 and the more recent Snapdragon 8 Gen 1 chipsets. In 2020, MediaTek also surpassed Qualcomm as the largest supplier of smartphones chipsets, and the gap has only been growing with each passing quarter.
Meanwhile, Qualcomm is diversifying into multiple different markets beyond smartphones. It is catering to chipsets for wearables such as smartwatches and TWS headsets, extended reality (XR) headsets, 5G and Wi-Fi 6/6e connectivity solutions, connected cars, A.I., automation in industries, and Internet of Things (IoT) beyond the home in industrial and medical environments.
Despite rising competition, Qualcomm has been posting better financial results than forecasts. Even as companies such as Apple grapple to fulfill demand due to the ongoing chip shortage, Qualcomm sailed through with 55 percent year-to-year growth in revenue in the fiscal year ending in 2021.
MediaTek, on the other hand, has been focused on consumer electronics. Besides smartphones, the company has been working on its portfolio of chipsets for Chromebooks, smart and Android TVs, Wi-Fi routers, and 5G hot spot solutions. Alongside the Dimensity 9000, the company also announced Kompanio 820 and 828 chips for mobile PCs that offer higher power efficiency as compared to counterparts from Intel, as well as a 7nm Pentonic 2000 chip to offer up to 8K video playback with A.I. upscaling and Dolby Vision on smart TVs.
Qualcomm and MediaTek May be hard to replace
We spoke to leading industry analysts who shared their insights on why smartphone companies leaning into chip design may have a tough time uprooting Qualcomm and MediaTek. All of them unequivocally agree it is difficult to replicate the same success as Qualcomm and MediaTek instantly as it may take companies several years to gain the same expertise as these giants.
Avi Greengart, president and lead analyst at Techsponential, tells us: “Designing your own silicon is difficult and expensive, but it can be a great way to differentiate your device, and, if volumes are high enough, can improve margins. However, competition is fierce: Apple sets a very high bar, and Qualcomm and MediaTek are pushing hard to keep up. If your in-house chip does not provide the same performance, functionality, connectivity, and power efficiency than what you can buy from Qualcomm, MediaTek, or Samsung, your phone will be less competitive, not more. ”
Greenheart elaborates that chips limited to a single brand may not only be less fierce than the competition, but will also be more expensive to produce and scale. The only company succeeding at it well is Apple, which sets a high bar for performance throughput with its A-series Bionic chips that use custom designs, instead of simply rebranding IPs from ARM like other companies in the list.
Tarun Pathak, research director at Counterpoint Research, states: “Designing chips is complex. They have become multifunctional, and integrating CPU, GPU, DSP, ISP with optimized performance is a very challenging task for any [manufacturer]. Then there is also a challenge of integrating RF and modem with the SoC. [The] short answer is it might take several years and significant resources for [manufacturers] to reach … the level to compete with the likes of MediaTek and Qualcomm.”
Pathak also highlights that Qualcomm’s prowess in designing non-smartphone chipsets has given rise to opportunities worth $10 billion cumulatively. These include the company’s endeavors in RF front-end parts, automobiles, and other IoT components.
Navkendar Singh, research director at IDC India, says: “For established players like [MediaTek] or Qualcomm, it is not a big worry as of now. Bringing in-house chips to the market and scaling up is a costly and time-consuming process. And even after launch, it required integration, testing, and acceptability in the market. Furthermore, brands like Oppo and OnePlus merge, they can share more resources and eventually chips development as well. However, this still requires a few years of effort and investment, before their own devices are replaced with [their] own chips.”
Singh remarks that while Qualcomm and MediaTek benefit from their vast portfolio of clients, the two chipmakers should be wary of advancements and research and development in areas such as A.I., machine learning, and camera technologies, especially when it comes to custom solutions like the Google Tensor that are capable of enticing a significant population of loyalists. This drift will be a cause of worry for Qualcomm and MediaTek.