Notebookcheck Logo

Computex 2024 | Analysis: Qualcomm puts Intel in panic mode and superstar Jensen lets Moore's Law die

The heads of Intel and Asus laughing at the Computex keynote. (photo: Andreas Sebayang/Notebookcheck.com)
The heads of Intel and Asus laughing at the Computex keynote. (photo: Andreas Sebayang/Notebookcheck.com)
It was an eventful Computex Taipei indeed. Intel visibly on the defensive and a little stubborn, Qualcomm omnipresent and Nvidia's superstar Jensen was everywhere to begin with. Far-reaching changes are on the horizon and the buzzword AI may even really reach maturity.

This past Computex Taipei was something special: The start of something new or the end of the reign of something old. Intel was clearly on the defensive at Computex. The reason: Nvidia and, above all, Qualcomm. The former has attacked Intel's important and high-margin server segment, and Qualcomm covers all bases as a smartphone chip manufacturer within the field of laptops.

Let's start with Qualcomm and the laptops: A whole number of devices were announced with a bang, all relying on the Snapdragon X Series. Elite as well as Plus, with the Elite version leading the way. In addition, this wasn't just a simple paper launch—the first devices have been announced for June 18, 2024, so they must have already been in production for ages. 

Moreover, the support from the industry was huge. There was hardly a manufacturer who didn't want to be at the launch. And Qualcomm proudly showed off its laptop designs in a display on stage. Of course, various manufacturers were invited to the stage and Microsoft's CEO Satya Nadella also spoke about the new future with Qualcomm.

An impressively large selection of Snapdragon laptops. (photo: Andreas Sebayang/Notebookcheck.com)
An impressively large selection of Snapdragon laptops. (photo: Andreas Sebayang/Notebookcheck.com)

And Intel? The chip giant put on a show that was slightly embarrassing. Before the keynote presentation, Intel introduced a good dozen important people from the very top decision-making level within the field of laptop production. They sat dutifully in the front row the day after the Qualcomm keynote. Acer, Asus, Compal, Wistron, to name but a few. 

They were then allowed to listen as Intel presented its Lunar Lake—but oftentimes having quite a lot of details left out. Above all, Intel couldn't yet say how Lunar Lake is supported to live up to the Snapdragon X Elite. This raised quite a few concerns.

Pat Gelsinger showcased Lunar Lake to the public. (photo: Andreas Sebayang/Notebookcheck.com)
Pat Gelsinger showcased Lunar Lake to the public. (photo: Andreas Sebayang/Notebookcheck.com)

Intel CEO Pat Gelsinger relied more on the past and on Taiwan. IT stands for Intel and Taiwan. Quite remarkable in view of the current tensions between PR China and the Republic of China. This was even part of the positive tone at Computex this year. Other companies emphasized the importance of the small island state, too. 

Next year marks 40 years that Intel has been linked to Taiwan, as Gelsinger said at Computex.  

There were also a few references to the x86 history of past decades. These were probably intended to say: "Hey, we've always been here and we're not afraid." However, we got the opposite impression. This was not a relaxed keynote proving to be unimpressed by the new competition.

Lunar Lake will be launched in the third quarter of 2024. (photo: Andreas Sebayang/Notebookcheck.com)
Lunar Lake will be launched in the third quarter of 2024. (photo: Andreas Sebayang/Notebookcheck.com)

Lunar Lake will only become a competitor later on

The display box for the Lunar Lake laptops—keep in mind that the chip isn't being launched until the third quarter of the year—didn't make the situation better. It seemed a little like it was hastily ordered from Ikea and put together. "Look at us, we have more laptops than Qualcomm!", could be heard from the white shelf. However, not on June 18, but a good few months later. It goes without saying that actual announcements of Lunar Lake laptops at Computex were limited. 

In the third quarter of the year at the earliest, it should also become clear that the high performance requirements of Intel's x86 processors are a myth. Lunar Lake is the "MythBuster" among the x86 processors and x86 technology is "at its finest", as Gelsinger put it almost defiantly at the keynote.

Many demonstrative Intel laptop designs (photo: Andreas Sebayang/Notebookcheck.com)
Many demonstrative Intel laptop designs (photo: Andreas Sebayang/Notebookcheck.com)

At least there was one impressive figure: Lunar Lake is said to require 40 per cent less power. With a small asterisk, because this is based on a YouTube benchmark. A "simple" reference to potential battery life would certainly have been more helpful. 

And of course, there is an NPU in Lunar Lake which, at 48 TOPS, is a whole 3 TOPS faster than Qualcomm and 2 TOPS slower than AMD. Yes, AMD is also still around and somehow also plays along with its Ryzen AI. AMD was even the first company to present its AI CPUs at the opening keynote. This was only theoretically so because Nvidia technically already made its announcement before that. But more on that later. 

In any case, there was little to be seen of AMD at Computex, while Intel and Qualcomm not only did a lot of advertising in the exhibition halls themselves, but were also suspiciously often seen inside the Taipei metro. If you took the subway, you couldn't miss one or the other. Last year, there was no such exchange of blows at Computex.

Qualcomm advertising inside the Taipei Metro (Nangang Exhibition Center). (photo: Andreas Sebayang/Notebookcheck.com)
Qualcomm advertising inside the Taipei Metro (Nangang Exhibition Center). (photo: Andreas Sebayang/Notebookcheck.com)
Nvidia advertising at the Tainex 1 exhibition hall (photo: Andreas Sebayang/Notebookcheck.com)
Nvidia advertising at the Tainex 1 exhibition hall (photo: Andreas Sebayang/Notebookcheck.com)
Asus advertises with Snapdragon and Ryzen AI at the Tainex 2 exhibition hall (photo: Andreas Sebayang/Notebookcheck.com)
Asus advertises with Snapdragon and Ryzen AI at the Tainex 2 exhibition hall (photo: Andreas Sebayang/Notebookcheck.com)

This exchange of blows then caused a real embarrassment at Intel's keynote. At the end of its own presentation, Intel once again showcased all the CEOs in the video, who were then allowed to say "I'm with Intel". Quite a few of them had been "with Qualcomm" the day before and were quite proud of their new ARM products.

Snapdragon is becoming an important phrase in the field of laptops. (photo: Andreas Sebayang/Notebookcheck.com)
Snapdragon is becoming an important phrase in the field of laptops. (photo: Andreas Sebayang/Notebookcheck.com)

The battle for the laptops is getting exciting 

One thing that definitely became clear after the processor manufacturers' presentations is that things are getting very exciting in the laptop sector. There haven't been that many designs with Qualcomm SoCs to date and the previous generations of Qualcomm have been somewhat lost to history. This is already evident when it comes to the NPU and its missing display within the Task Manager. Anyone who has bought an old Qualcomm-based laptop with the 8cx Gen3 may be a little annoyed. The old platform doesn't look like the future. But this is exactly the problem that Snapdragon X Elite and X Plus are supposed to solve. 

The first signs strongly suggest that this has been a success. Within the laptop segment, it seems Qualcomm will be able to take some market share from Intel and its attack on the desktop segment is still to come. However, we suspect it will rather be the mini PC segment it aims for. Qualcomm has not yet shown that it can make high performance possible via strong heat development—a domain of Intel and AMD.

According to Qualcomm's head, laptops are but the beginning. (photo: Andreas Sebayang/Notebookcheck.com)
According to Qualcomm's head, laptops are but the beginning. (photo: Andreas Sebayang/Notebookcheck.com)
A side blow to Intel: The NPU of the old Intel generation is said to get very warm. (photo: Andreas Sebayang/Notebookcheck.com)
A side blow to Intel: The NPU of the old Intel generation is said to get very warm. (photo: Andreas Sebayang/Notebookcheck.com)

They are now "here to stay in the neighborhood", said Qualcomm CEO Christiano Amon accordingly at his own keynote.

This brings us to the end of the laptop segment. Intel has quite a lot to fear here. Maybe this is also the reason why Intel wants to prepare the industry with Panther Lake. Within the desktop segment, we don't see too big a threat for the time being—especially as Arrow Lake, a potentially fast SoC, is still set to be launched in 2024. This is Intel's way of also enabling AI use within desktop PCs. We can assume it will feature a fast NPU.

On the flip side, Intel is likely to face stiff competition within the server segment.

Christiano Amon announced the rebirth of PCs. (photo: Andreas Sebayang/Notebookcheck.com)
Christiano Amon announced the rebirth of PCs. (photo: Andreas Sebayang/Notebookcheck.com)

Intel versus Nvidia  

Nvidia's CEO Jensen Huang, who is something like a superstar in Taiwan, was also able to take a hit at Intel. This is because the manufacturer simply set up its own keynote before Computex and a few days later, it even hosted hosted its own conference—the Nvidia AI Summit. Nvidia's GTC was clearly not enough.

To be fully transparent: Intel also held its own tech tour in the run-up to Computex. However, Nvidia's efforts were more extensive. The AI Summit was primarily intended to offer Taiwanese companies a glimpse into the future. Most of the presentations were held in Chinese. Only a few were in English.  

In turn, Nvidia rented an entire floor inside the Grand Hilai—which opened just last year and is located directly next to the exhibition center in Nangang—and occupied three lecture halls, which were generally filled well. 
Jensen held a large-scale presentation showing in detail what the company had prepared. And it was thoroughly impressive.

Despite the pouring rain, Nvidia managed to fill the NTU Sports Center with a capacity of over 4,000 seats. At Nvidia, the focus has clearly shifted to Huang the person, while its products have been put on the back burner—at least for the time being. There were already hints of what has long been common practice at many server companies such as Dell or HPE: solutions instead of explicit hardware.  
But while Dell and HPE reveal almost nothing about their hardware and position this information in hard-to-reach data sheets nowadays, Nvidia and especially Huang still remained true to themselves: Technical data at a keynote? It still exists. And even a roadmap for future products.

On the other hand, some people criticized the fact that they didn't announce many new things. The show surrounding Nvidia's CEO was a little unusual in this otherwise very tech-oriented industry.

One of the few concrete announcements: Performance data on the upcoming Ruby platform. (Photo: Andreas Sebayang/Notebookcheck.com)
One of the few concrete announcements: Performance data on the upcoming Ruby platform. (Photo: Andreas Sebayang/Notebookcheck.com)

Not to mention gamers in particular are likely to have been disappointed by the presentation. Is Nvidia actually still a company that produces graphics cards? Well, that's a matter of definition, because GPUs are now much more than just accelerators for games in Nvidia's eyes. "This is a GPU!" joked Nvidia's CEO during his keynote, standing next to a full 19-inch rack that was drawn on the screen inside the nearly full National Taiwan University Sports Center in Taipei. 

He only briefly showed a "real" graphics card on stage once—without announcing anything new.

So, there's nothing for gamers. Intel has to worry about that. The different approaches are very interesting in this aspect: Nvidia with its proprietary technologies and Intel with its open concepts. Intel emphasized the latter during its presentation of Gaudi 2, Gaudi 3 and the Xeon 6. Whether it's the Ultra Ethernet Consortium or the Open Compute platform, anyone who relies on Intel relies on standards.

This alone apparently isn't enough, as Intel additionally countered with competitive prices. You could almost say that Intel is becoming to AI hardware what AMD is to desktops and laptops. Is this also a sign that Intel is under threat? After all, it has been able to dictate prices for decades, especially within the high-end sector.

Furthermore, Intel's chief Pat Gelsinger reacted almost petulantly to Jensen Huang's statements on the end of Moore's Law. The law is still applicable and will continue "as long as the periodic table has not been exhausted", according to Gelsinger.

Intel's Gaudi 3 partner (photo: Andreas Sebayang/Notebookcheck.com)
Intel's Gaudi 3 partner (photo: Andreas Sebayang/Notebookcheck.com)
Gaudi 2 is intended to attack Nvidia on price as an entry-level AI accelerator. (photo: Andreas Sebayang/Notebookcheck.com)
Gaudi 2 is intended to attack Nvidia on price as an entry-level AI accelerator. (photo: Andreas Sebayang/Notebookcheck.com)

AI, AI, AI 

There was one thing that all participants had in common: an overemphasis on artificial intelligence. You could assume that nothing works without it. At Computex, you could see the strangest manufacturers with an AI logo. Apparently even cases and mice need something like this.

Still, while in recent years it has been possible to safely ignore artificial intelligence or to send ChatGPT into a tantrum with the help of suitably complex questions, AI is now slowly growing up.

The only problem is trying to pick out the important facts from all the buzzwords. One example of this was Qualcomm's AI vision. We are supposed to talk to our Snapdragon PC almost naturally and let it place orders for white goods if a repair is not worthwhile.

This reminds us a little of all the big promises the Google Glass project made, of which not much is left over these days.


Visions like this may be interesting, but small-scale progress is more likely to be made—not least because some areas (such as ones reliant on knowledge) suffer larger problems if the supplied information isn't enough for the AI. For example, we conducted various experiments with food packaging (Ecolean) and the history of a hotel. In both cases, it turned out that the internet couldn't provide enough information and the AI tended to make things up or stubbornly keep saying the same (wrong) thing.

In turn, applications like image generation already work quite well. It just depends on the field in which the AI is to work. Laptop NPUs should be able to power such tasks very well, while professional AI accelerators work primarily in the cloud or on large solutions.

For Nvidia, Blackwell is … (photo: Andreas Sebayang/Notebookcheck.com)
For Nvidia, Blackwell is … (photo: Andreas Sebayang/Notebookcheck.com)
… of enormous importance. (photo: Andreas Sebayang/Notebookcheck.com)
… of enormous importance. (photo: Andreas Sebayang/Notebookcheck.com)

Nvidia's vision of a second Earth is especially exciting. Nvidia aims to build a digital twin of the Earth piece by piece and will report on its progress every year. It is likely to become one of the references for its own GPUs. This copy of the Earth will then be suitable for predicting the weather, among other things. Taiwan has repeatedly been used as an example of this, as it has to contend with challenging weather events. The island nation is well prepared for a typhoon and infrastructure failures are manageable. But there is potential for optimization. With such a second Earth, weather forecasts down to an accuracy of one square meter should eventually be possible—worldwide.

And Nvidia has a second focus: According to Jensen's vision, the AI must learn physics. He calls this Physical AI. The AI will only become better once it can follow the corresponding laws. He is alluding to robotics in particular. Humanoid robots will likely be the next big step for Nvidia.

This is a field which you should keep a close eye on. A small field that will be surrounded by the buzzword AI. Because today (for marketing reasons) nothing works without AI. However, very few AI products are actually likely to provide any significant benefit.

Nvidia B200 … (photo: Andreas Sebayang/Notebookcheck.com)
Nvidia B200 … (photo: Andreas Sebayang/Notebookcheck.com)
… and B100 (photo: Andreas Sebayang/Notebookcheck.com)
… and B100 (photo: Andreas Sebayang/Notebookcheck.com)
The Earth is set to receive a digital twin. (photo: Andreas Sebayang/Notebookcheck.com)
The Earth is set to receive a digital twin. (photo: Andreas Sebayang/Notebookcheck.com)

Source(s)

Computex / own research

Read all 2 comments / answer
static version load dynamic
Loading Comments
Comment on this article
Please share our article, every link counts!
Mail Logo
> Expert Reviews and News on Laptops, Smartphones and Tech Innovations > Reviews > Analysis: Qualcomm puts Intel in panic mode and superstar Jensen lets Moore's Law die
Andreas Sebayang, 2024-06-10 (Update: 2024-08-15)