In just over a year and a half there have been two pretty big blows to the industry that are going to completely change it to the level that the first-generation Ryzen architecture changed it. And it is that efficiency now is going to be the great trigger for the future of CPUs. Qualcomm knows this and has been relatively quiet until now, until Apple and Intel have come to eat their cake. Therefore, the movements are going to happen to be competitive and the first could be to leave TSMC for Samsung.
If you’re clueless about behind-the-scenes moves for this particular industry, the simplest question is what’s going on? Well, it is simple and at the same time explaining it serves as an input for the topic of this article. Basically and in short: Apple and Intel have fully immersed themselves in the terrain that Qualcomm dominated with ARM, where it was the absolute leader and with its own weapons they are surpassing it.
Buying Nuvia to compete in CPU
In February of this year Qualcomm in a clever market move bought the Nuvia company, knowing that it was one of the “cheapest” ways to enter the CPU sector. Logically they were aware of the movements of Apple and Intel after their little family dispute, so they knew perfectly well that the water was coming to their neck and the purchase was a way not only to protect themselves, but to start the counterattack.
It is currently being developed where Qualcomm has had to make a difficult decision for the second half of next year or the beginning of 2023, since knowing that Apple and TSMC are “fingernail and flesh”, that those of Cupertino have absolute priority for the new node that each generation comes out from the Taiwanese, what options does Qualcomm have to be able to beat its rivals and be for in front of?
Qualcomm and Samsung: the first ARM GAA CPUs?
The problem here is evolution and time, a difficult compendium to square when you are competing to arrive first. Apple will continue to rely on TSMC for its 3 nm, but these will continue to be FinFETs, Intel is somewhat late and the RibbonFETs will arrive after the named dates, so the only viable option is Samsung, which could sign a contract with sufficient volume thanks to the new FABs that it will create from looking forward to the next few years.
For all this, Samsung claims that in the first half of 2022 it will mass produce its new 3 nm node with GAA transistors that will presumably achieve better efficiency and consumption, not so much performance per se, at least in the first versions of it. Rumors point towards this outcome, where Qualcomm would have a more advanced node than Intel and Apple / TSMC, maybe not that dense, but they could compete and maybe win in the efficiency battle if they make an ARM CPU design to match.
And it is that the delay that Samsung now has in front of its two rivals will be an advantage next year, arriving first to the ring and having more experience with the creation of wafers and the engraving of these transistors, from which Qualcomm could benefit and get at the height of its rivals.