On the one hand, Samsung will use Qualcomm Snapdragon processors for its next Galaxy smartphone. On the other, we hear rumors of a new chip from Samsung for 2025 with an AMD “super” GPU. In the middle, doubts about Samsung’s strategy. Who asks the question: have the Exynos chips reached the end of the road?
During its Snapdragon Summit, the chip giant for smartphones Qualcomm validated a rumor that had been running for a few months : Samsung’s future Galaxy smartphones would be well equipped of Snapdragon chips in Europe. A micro earthquake – and good news? – for the European market, which was only entitled to Samsung’s in-house chips, the Exynos. In a strange ballet, Qualcomm and Samsung have (obviously secret) technological sharing agreements that allow the Korean to partially rely on certain blocks of Qualcomm. The goal is for Samsung to continue to offer powerful processors in smartphones. A Samsung which is still the first supplier of iPhone SoCs before Apple develops its own chips.
From the point of view of journalists and analysts testing the products, the switch to Qualcomm seems like good technical news. Because whatever Samsung does, its partner’s chips are always a cut above, both in terms of pure power and in terms of energy consumption. The Galaxy smartphones delivered to Europe were thus systematically a small notch below those found in the USA or South Korea (yes, in its own national market, Samsung delivers terminals equipped with American chips!) question has always been: why does Europe alone have to have a slightly less efficient processor? A question that is now evolving towards a probable “death”, even partial, of Samsung’s high-end Exynos chips. Because recent history and the challenges ahead are enormous.
A first abandonment in CPU cores
Samsung has been developing SoCs since the beginning of smartphone history. And even designed custom hearts (custom in the jargon) in a dedicated R&D unit in Austin, Texas (USA). In November 2019, Samsung must face the reality: the work provided by this team does not make it possible to make a difference compared to the original plans provided by ARM. Samsung fails to offer chips that are more efficient than those of Qualcomm, and often finds itself behind. Faced with the impossibility of these engineers to do better than the reference design of ARM, Samsung decides to close the unit.
A few months earlier, the Korean firm announced a partnership as surprising as it is attractive with AMD. It was on the plans of the American GPU, which was then (and still!) on the rise, that Samsung indicated that it was basing its next generation of GPUs integrated into its mobile processors. And the more the months passed, the more AMD’s RDNA technology made people salivate: we learned in the spring of 2020 that both the Xbox Series and the Playstation 5 would use this RDNA architecture (in version “2” in the case of consoles living room).
As 2020 and 2021 progressed, Samsung’s two-year road plan was unveiled. Until the announcement on January 18, 2022 of the Exynos 2200 chip which you can find today in the European Galaxy S22. A 100% ARM chip on the CPU side, but totally customized on the GPU side. For a profit that we know…
The AMD GPU hasn’t (yet?) made a difference
If AMD begins to seriously threaten Nvidia in the PC and controls all home consoles, its first cooperation in the field of mobile GPUs did not produce the expected “wow” effect. The Exynos 2200 which equips the European Galaxy S22s has indeed difficulty in comparing with the Snapdragon 8 Gen 1 and, a fortiori, the Gen 1+.
According to an analyst we spoke to at Snapdragon Summit who wished to remain anonymous, “ the fact that AMD’s GPU cores are the basis of high-power cores makes the operation of transformation into low-power cores very difficult. “A vision shared by Qualcomm engineers when asked if they will be able to really scale up graphics power in their future computer SoC: “ In addition to Oryon, our GPU will scale up because our architecture is very scalable* and so it’s easy for us to add cores and scale up.” The x86/ARM battle proves them right: it’s easier to start with a low-power architecture and then scale up than the other way around. If we now find ARM processors in computers and supercomputers, x86 has never managed to find a place in the world of mobility.
The impossibility of even matching the Qualcomm chip, especially in 3D, is a total snub for Samsung. For the unit that designs the SoC, for the one that manages the partnership with AMD, but also for the one in charge of manufacturing. Samsung’s 4nm was found to be less efficient than TSMC’s 4nm. So much so that Qualcomm has changed dairy for its “+” version as well as for the new Snapdragon 8 Gen 2.
(*: scalable means “extensive”, “scalable” and is said of a technology that supports very well the increase in power by adding calculation units)
The Exynos do not convince either Samsung or its customers
In full overhaul of its SoC design division in 2019, Samsung also announced that it wanted to change its business model. In addition to designing chips for its own terminals, Samsung had indicated that it wanted to sell its Exynos chips to other manufacturers. A strategy that seems to have fizzled. If the Koreans have delivered entry and mid-range chips to Chinese manufacturers, the transplant has not taken. It’s even the opposite: the market share of Exynos continues to decline.
Here, it’s not just Qualcomm hurting Samsung, but also MediaTek. The Taiwanese continues to expand its ranges and gain momentum. Taking advantage of its status as an independent chip designer, MediaTek not only produces chips for smartphones, but also for internet boxes and televisions. Markets that have pushed its R&D to a very high level. Allowing it not only to sell chips for the entry level as in the past, but also for high-end devices of certain brands. Who undoubtedly prefer to buy good chips from an independent player rather than a competitor!
Rotating governance that prevents consistency
Samsung did not become the world’s number 1 smartphone by accident and has some excellent engineers in its ranks. With such a global weight in electronics – Samsung has just relaunched its computer sales in many European countries, including France – why is the Korean struggling so much in the design of a high-end chip? For several analysts interviewed on condition of anonymity, Samsung is paying here for its governance system. Many high positions indeed rotate every two years – and it is sometimes every year for other profiles. All of this without asking too much of their opinion. If the fact of preventing his troops from getting “encrusted” can be a good thing, this mobility also has negative effects.
Because in the world of semiconductors, the development cycle of a processor is about three years and requires around 10,000 people. A huge payroll that must be managed and moved in the right direction for thirty-six months. A huge challenge, which only a few rare companies in the world manage to meet… and which hardly seems compatible with games of musical chairs that are too close together.
By letting it filter a “return” with an “AMD super GPU” by 2025, Samsung seems in the nails. If we count the three years to design a SoC from scratch and the integration of a more efficient (and adapted!) graphics part, this horizon seems possible. But Samsung has an interest in staying the course and avoiding changing its mind every four mornings. On pain of ending up, one day who knows, having to close his division. And settle for Qualcomm and MediaTek chips.