Intel and Qualcomm have a problem. And it’s probably fatter than originally thought. Starting this Tuesday, March 8, Apple’s former chip supplier has to deal with the M1 Ultra, Apple’s latest silicon creation. Based on two M1 Max chips joined by a technology called UltraFusion, the result can be described as a technological milestone.
Looking to 2023 to beat the 2020 M1
Intel and Qualcomm are the two big chipmakers if we exclude Apple. Both are specialized in different markets, the first in PCs with its x86 architecture and the second in smartphones with ARM architecture. Both are going to have to make a big effort to recalibrate and face Apple’s M1 chips, as they combine smartphone ARM architecture in a chip for traditional computers: the Mac.
The challenge is not small. The problem for both manufacturers is that time has caught up with them. Late last year, Qualcomm announced an ARM PC chip “competitive” with Apple’s M1.
The case of Intel is similar. After dismissing Apple’s threat and mocking Mac’s M1, it wasn’t until a few days before the Mac Studio keynote that his plans were (conveniently) leaked. Intel’s roadmap against M1 also places its response in 2023.
It’s funny how both companies, Qualcomm and Intel, put their answers to Apple’s M1 in 2023. Intel can still say their plans are done with the 2021 M1 Pro and M1 Max on the market. But Qualcomm’s is based on the 2020 M1. Two and three years of difference between the product with which they are preparing and when theirs see the light.
Both Intel and Qualcomm have a 2020 chip in their sights to beat it in 2023
None of them knew what Apple would present last Tuesday. But the first benchmarks of the M1 Ultra leave no room for doubt: they are even ahead of a Mac Pro with an Intel Xeon, yes, from 2019. Later this year, we expect the first Macs with an M2 chip, the first iteration of this successful generation. What will the M2 Pro and M2 Max of 2023 look like? And the M2 Ultra of 2024? It is to them that they should put their sights, not to the M1.
Performance per watt as the enemy to beat
Beating Apple’s M1 chip in power is easy. In fact, the Intel Alder Lake already outperforms Apple’s M1 Max. The problem is the high consumption of these chips. Because what is the use of overcoming a chip in power if you consume double or triple it along the way? What Apple has brought back to the table is a variable that has been forgotten for too long: performance per watt. It’s the same one Steve Jobs mentioned in his transition from PowerPC chips to Intel in 2005:
Here we see how Jobs explained with a single slide the reason for the transition to Intel. For 2006, 15 performance units per watt were expected on the PowerPC compared to 70 performance units per watt for Intel. The choice is obvious and with the Apple silicon too.
The switch to Apple processors is also motivated by performance per watt, among other reasons. We’ve seen graphics like this one at every new Mac model unveiled. And those graphics are more than enough to prove the superiority of Apple’s chips.
Intel and Qualcomm have a problem. Many will think that this war is not going with them because the Apple silicon is only in Apple equipment. But it is a mistake to think that the PC user is captive to these computers and will not look at Apple’s Macs.