Arm in desktop computing: is it all in Apple's hands?

This week, following a Bloomberg article, rumors swirled that Nvidia was preparing to drop its purchase of Arm for more than $40 billion. And this following the signals sent by the Federal Trade Commission on its intention to stop the merger.

The merger of Nvidia and Arm would have created a potential second superpower in semiconductors, behind Intel, and a strong competitor for Qualcomm and Samsung in mobile and embedded processors. As well as a silicon supplier for the next generation of Arm-equipped Windows laptops from Microsoft and other vendors.

So many industry watchers may wonder: Does Arm's future in desktop computing exist outside of Apple? At present, only Apple has succeeded in using Arm silicon by upgrading its Mac platform to the M1.

Apple is clearly in the lead for Arm-based desktops

Let's be frank: Apple is so far ahead of other companies when it comes to putting a desktop platform on Arm-based silicon that it's not even funny. The M1 has now been effectively rolled out to the entire Mac product line except for the Mac Pro. And we know it will be available soon. The M1 Pro and M1 Max are no exception. I've been using one for about a week, and let me tell you, there's no way I'm going back to an Intel machine after experiencing such performance.

Granted, Qualcomm is making progress with SQ 2, and Microsoft has made Windows 11 run better on Surface Pro X. Still, there's no comparing what these two have done to make Windows run acceptably on a single product with what Apple has done to make Apple Silicon work across its entire product line. Not to mention iPad, iPhone, Watch, TV and who knows what else Apple is working on.

And what about the entire PC industry? HP? Dell? Lenovo? ASUS? Acer? Samsung? The first three alone occupy a 64% share of the personal computer market, while Apple represents only 7.6%.

Apple's share is small compared to the PC market as a whole; it is only slightly higher than that of Acer, which is 7.2%. But, there is a but.

Apple occupies a position that is reminiscent of the difference between Tesla as an automaker and the rest of the auto industry. Tesla has delivered more electric vehicles in a single year than the entire traditional auto industry in 10 years.

However, Tesla in the United States represents only 2.3% of the sector. But it has validated the technology to replace the internal combustion vehicle. And Apple did the same for the Arm architecture, which will eventually succeed the x86.

Like several years ago, Mac's overall market share in business desktops remains relatively small, but now we see that it has enormous potential, the power of the product is unassailable.

Today, we all say: “my God, look at what Apple has achieved”. The Arm desktop is real, the Arm is powerful, the x86 can't top what this thing can do in terms of performance per watt.

Apple's M1 isn't a miracle, it's the product of a massive strategic investment

The fact that Apple has achieved so much in what seems to be a relatively short period of time is a testament to the investments it has made in its own chips for more than 10 years. Before Apple announced Apple Silicon as its future platform, the idea of ​​using Arm chips as PC platforms was considered theoretical, bordering on science fiction.

We saw a first attempt at Windows on Arm with the Surface RT, which failed. It used an Nvidia Tegra 3 chip. It was slow, ran an incompatible version of Windows 8, and could not run existing Win32 x86 apps other than those preloaded by Microsoft and Windows Store "Metro" apps.

It has certainly been rumored for years that Apple was working on an Intel successor for Macs. But Arm on a desktop OS hadn't really been put into practice unless you were an engineer from a company like Microsoft or Qualcomm and knew that in a lab, Windows ran pretty well on that architecture . And yes, with a few years of work and billions of development dollars, we might be able to replace x86 one day.

We knew that Windows ran smoothly as a server operating system on special Qualcomm chips in Microsoft's data centers, and that some specialized Azure workloads ran on it.

But, for the typical business PC user, it took years, if not a decade, for that to become a reality.

Ultimately, the issue is not Windows readiness, but overall silicon maturity. Not only does Windows run on Arm, it also performs great if you run it in a Parallels window on a Mac and run native apps on that system. So, Windows on Arm as an operating system is ready, and Microsoft has all the developer tools to port applications if silicon were ready tomorrow in the PC maker ecosystem.

Emulated x86 apps now work much better on Windows 11 for Arm. But the Qualcomm SQ 2 is the only capable Windows desktop chip on the market, and it's nowhere near as powerful as a first-gen M1. Not to mention an M1 Pro or an M1 Max or whatever M-something we'll definitely see at the next Apple event when the Mac Mini gets a refresh and other stuff is shown to us.

The performance of the PC Arm system? A chasm that will take years to fill

The main Arm silicon supplier to rival Apple, now that Nvidia is certainly sidelined, is Qualcomm. Microsoft seems to have exclusive rights to the company's SQ series and doesn't seem to have any intention of licensing it out to other OEMs, at least for now.

HP is testing something with the Surface Pro X. This laptop uses Qualcomm's low-power, 4G-capable Snapdragon 7c Gen 2 platform, but it doesn't match the M1 or even the SQ 2 in terms of performance. It is designed to compete with the Pentium Silver.

Lenovo has the Yoga which uses the 5G-enabled Qualcomm 8cX. It was launched in 2020 and refreshed in February 2021. However, if you look at Lenovo's announcements at CES, there is no Arm version of the Yoga announced for 2022. It's a 100% Intel Evo and AMD Playbook Ryzen.

So I think sales were bad for the Yoga Arm. The company opted for the low-power Intel Evo because x86 is a proven architecture with predictable performance and industry confidence. Dell is also going with Evo, and HP's low-power laptops are running Evo as well.

The 12-gen Intel Evo is pretty good. Not as powerful as the M1 though. It offers sufficient performance in terms of CPU and energy consumption with an announced autonomy of 9 hours in real conditions with a Full HD screen. Most mainstream business users will be fine with this.

The MacBook Air M1 has a battery life of around 15 hours. Its wattage consumption varies between 18W and 110W depending on the application load. For example, if you are processing videos in Final Cut or playing a very resource-intensive FPS game, the power consumption is high.

Like Apple Silicon, Evo is an SoC platform with i6 Wi-Fi and an integrated GPU, with a total system footprint of 35-45 watts. To be clear, this is a laptop roughly comparable to a low-end 2020 M1 MacBook Air in terms of features. But these Evo chips aren't nearly as powerful.

Finally, we should probably mention the new AMD Ryzen 6000 platform for laptops, which is also an SoC platform that is getting a lot of attention from PC manufacturers.

From a practical standpoint, the only worthwhile Arm processor right now is Apple. Will Qualcomm progress? Sure. But this is a company that has its hands in a lot of different verticals and businesses.

Qualcomm is a hugely profitable embedded and wireless company; it doesn't need to make desktop computer chips to make money.

The same can be said of Samsung Semiconductor. The South Korean electronics giant makes a lot of money making chips for everyone and although it has its own designs for its smartphones and tablets and has a computer business wearables, it partners with companies like Google and manufactures their smartphone chips, as well as those from Qualcomm and, of course, Apple.

So Samsung doesn't need to make chips for desktop computers to be profitable.

Is this a potentially important area of ​​expansion for these two companies? Yes. Will desktop chips make them or break them? No. They can afford to take their time.

Arm on desktop PC: where are we going?

If Nvidia had gone through with its purchase of Arm — assuming Bloomberg's article is true — the company would have been in a position to drive the future of the Arm architecture. And companies like Apple, Qualcomm and Samsung would have bought licenses.

The merger/acquisition would have given it an edge in its own chip designs to create chips that would eventually catch up with Apple Silicon. But it looks like Nvidia will have to do without Arm. Instead of this acquisition, it also appears that Softbank will take Arm public, making it a public company. However, Arm, without a major silicon maker behind it, remains on a licensing business model.

I continue to believe that the desktop computer has a future that is not tied to x86 PCs. But it might not take the form we expect. I firmly believe that the ultimate destination of the desktop is in the cloud, where data is accessed by low-powered, low-cost endpoints running minimal, embedded operating systems. And these PCs will probably ship open source operating systems. But this is an argument to be developed in a future article.

Source: ZDNet.com