In a struggling laptop market, analysts predict not only the consolidation of Apple’s M chips, but above all the success of Qualcomm and MediaTek. Three representatives of ARM in “classic” computing, whose chips could account for 25% of the portable PC market in five years. A realistic vision?
A success for Qualcomm and MediaTek: this is what a report analyzing the laptop PC market published by Counterpoint professes. The firm specializing in the analysis of electronic equipment sales predicts that portable PCs, a fairly difficult market, will see the market share of ARM chips increase to 25% by 2028. And even predict that ARM chips will already weigh 10% of the global market from 2025 “at the expense of Intel and AMD”.
After the world of personal computers was entirely driven by x86 instruction sets from Intel and AMD, the tide started to turn in 2017 when Qualcomm and Microsoft introduced the first PC with Windows on Snapdragon. A version of the operating system specially compiled for an ARM smartphone chip (the Snapdragon 835). If Qualcomm continued its work, everything really changed when Apple made the transition to its own chips in 2020. Designed on the basis of ARM instruction sets and an internal design, the M1 then M2 chips have since integrated all machines from Apple, Macbook like iMac and Mac Mini – with the notable exception of the Mac Pro.
Read also: Qualcomm unveils Oryon, the CPU core of its anti-M1 weapon, from Apple (Nov. 2022)
Counterpoint’s rather sensational announcement echoes a previous forecast by Canalys which, last December 2022, expected no less than 30% of the market within four years. An even more optimistic analysis for ARM, both in market share and in speed of “conquest”. The ARM instruction set has its strengths, but what generally attracts the most is the possibility of designing chips “à la carte” rather than having to settle for the designs imposed by Intel and AMD.
Analyzes committing only those who listen to them, let’s take a small step aside here. While five years represents both an eternity in the world of its semiconductors but also, and above all, a complete development cycle, let’s take a look at what could derail the pro-ARM scenario that analysts foresee.
Fantasy architectural gaps
Counterpoint and Canalys undoubtedly see the traction forces of future ARM chips for PC based on analyzes at a given moment. /watt, the promises of the Oryon chip from Qualcomm or the connectivity integrated into ARM SoCs which should benefit Chromebooks. Real successes, tangible promises, but which seem to forget a little that Intel and AMD are not sleeping. And that the current strengths of upcoming competing chips are not so decisive.
Let’s first talk about the oppositions of microarchitectures – we also say “ISA” for instruction set architecture – and fineness of engraving. If in the past the RISC and CISC worlds opposed each other, now chip designs are hybrid: x86 chips have simplified their fundamental bricks, ARM chips have made their instructions more complex. ARM still has the advantage of being easy to add cores – but too many cores does not really benefit the general public – and the performance/watt ratio remains better than x86…
But less than in the past. One of the major reasons for Apple’s advantage with its first M1 chip was clearly TSMC’s 5nm process. Which happened at a time when the bulk of Intel chips were either 14nm (ok: 14nm+++), or first-generation 10nm for mobile chips. Since then, Intel has not only progressed (Intel 7, which is equivalent to an improved 10nm) but AMD has also demonstrated the importance of nodes by engraving chips, at up to 4 nm. When we see the 14 hours of autonomy of a laptop like the Asus Zenbook S OLED whose Ryzen 7 6800U chip is engraved in 6 nm, we can see that the manufacturing node has a lot to do with it. And out of this finesse of engraving, we must also remember that AMD and Intel have great assets.
Power of Intel, Flexibility of AMD
Intel jostled, Intel in turmoil: this is what analysts are (too?) quickly pointing out in view of Intel’s latest sales figures. A giant that continues to make money (lots of money), but still had to tighten the purse strings, lay off employees and cut branches (RISC-V development kit in particular) to focus on concrete goals.
What analysts seem to forget is that after years of weak or haphazard management, Mr. Gelsinger took the reins of Intel exactly two years ago by imposing a major strategic change: reorganization of business units, priority on microarchitecture developments and development of an independent foundry service… while giving itself the permission to look elsewhere for its own products. A 180° turn from what Wall Street demanded. And above all, a first year to thoroughly review the technological roadmaps – in the midst of the COVID crisis. However, laying the deep foundations for a technological renewal takes between three and five years. As proven by AMD who took the time to develop the foundations of Zen well which is still at the heart of its Ryzen chips.
Intel is already giving hints of its evolution towards modular architectures (disaggregated in the jargon of the company), and above all, more efficient… And less expensive, as evidenced by recent processors such as the N100, N200 and N300. Not to mention upcoming chips like Meteor Lake. A processor that could throw a (sacred) stone in the pond…
The other representative of the x86 world, AMD, also has some very nice cards up its sleeve. Among which are successive Zen architectures (Zen 2, 3, etc.) as well as a “chiplet” design which now allows it to offer inexpensive and efficient chips. As well as a graphic know-how which gives it the latitude to increase in power both from a gaming (Zen 2, Zen 3) and multimedia point of view. All with unique software compatibility, due to decades of in-game driver work.
Another advantage and not the least: AMD is the king of collaborations. Xbox Series, PlayStation 5 and other SteamDeck, the company powers the bulk of the consoles because it is able to offer “custom” chips. A customization that could grow with PC manufacturers as its design in technological “bricks” improves. Here, the interest of ARM is eroding since if we can make a chip with small onions, the costs of designing and validating chips in 5 nm and less explode. Where AMD can literally play Lego to offer unique processors.
The unknown of the impact of connectivity
PC chips from Qualcomm and MediaTek have a hidden promise (because we’re waiting to see them): network integration. Based on chips for smartphones, these chips could therefore have advanced connectivity Wi-Fi but also 5G. This simplifies the design of motherboards, general integration. And could, in some cases, make the difference, especially in professional fleets or state orders – school equipment, etc.
The argument, however, still did not catch on. Qualcomm is obviously pushing 5G to the limit, as is its Taiwanese competitor MediaTek. But Windows PCs (Qualcomm Snapdragon 8cx Gen 3 and other Snapdragon 7c) and other Chromebooks (Kompanio 1000) have so far failed to really break through. The success of the deployment of 5G by the world and the uses that will be made of it (which depends more on network equipment manufacturers than on PC manufacturers!), will undoubtedly dictate part of the success of their chips. That as well as the price of their solution: why choose such an expensive chip which risks being less compatible? Compatibility being the last key technical argument of x86.
Finally, analysts base part of their conclusions with the simple look of pure technology. However, that would hide the fact that both Intel and AMD are also industrial, technological… and marketing partners. From the design of chips to the card, through the supply of turnkey electronic designs, through the financing of advertising campaigns and the support of brands through their distribution channels, these two giants have many assets. And there is no doubt that they will use it to contain – and counter? – this ARM “threat”.
Counterpoint Research
