Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/huawei-exec-rejects-idea-that-advanced-chip-shortage-will-hamper-chinas-ai-ambitions.20570/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Huawei exec rejects idea that advanced chip shortage will hamper China's AI ambitions

Daniel Nenni

Daniel Nenni

Admin
Staff member
fbd929ee44ef8acd430e1dd5116c563f

FILE PHOTO: Huawei's booth at the Mobile World Congress in Shanghai, China

BEIJING (Reuters) - A senior executive at Chinese technology giant Huawei on Thursday dismissed the idea that a shortage of the most advanced artificial intelligence chips will hinder the country's aim to be a leader in AI, but said innovation is needed to address the issue.

The comments by Zhang Ping'an, CEO of Huawei Cloud, comes amid tighter U.S. restrictions on advanced AI chip shipments to China including a ban on sales there by companies such as U.S. giant Nvidia.

"Nobody will deny that we are facing limited computing power in China... But we cannot rely solely on having the AI chips with the advanced manufacturing process nodes as the ultimate foundation for AI infrastructure,” Zhang told a forum at the World AI Conference in Shanghai, a three-day event that kicked off on Thursday.

"If we believe that not having the most advanced AI chips means we will be unable to lead in AI, then we need to abandon this viewpoint," Zhang said.

Huawei, which has been placed on the U.S. Entity List barring it from purchasing advanced chips from U.S. companies, has developed its own AI chip product called Ascend, which is now used by many companies in China to train AI models.

However, the Ascend AI chip, along with many others from Chinese companies, is considered to be significantly inferior in terms of computing power compared to the offerings from Nvidia.

Zhang called for innovative approaches that place more focus on the cloud, which he said can help to compensate for the lack of advanced AI chips through innovation in computing architecture.

He also said that a converged approach is needed to combine cloud, edge, and networks in ways that can be used to reduce energy consumption and improve overall efficiency. Zhang touted Huawei Cloud as being among the leaders in providing such innovative solutions.

www.yahoo.com

Huawei exec rejects idea that advanced chip shortage will hamper China's AI ambitions

A senior executive at Chinese technology giant Huawei on Thursday dismissed the idea that a shortage of the most advanced artificial intelligence chips will hinder the country's aim to be a leader in AI, but said innovation is needed to address the issue. The comments by Zhang Ping'an, CEO of...
www.yahoo.com www.yahoo.com
 
A key requirement for AI chips, especially those with large dies like NVIDIA's, is for redundancy to be built into the design. Then the naturally low yields for these large dies won't matter (as much) as they can be repaired. So it begs the question of whether Ascend has the redundancy built in.
 
lilo777 said:
As far as AI is concerned, all innovation there is in software (although more powerful hardware obviously helps) and in AI software Chine is doing very well. China beat the U.S. in generative AI patents by 6-to-1 for the past ten years.
Click to expand...
I’m not sure patent counts or paper counts mean much when it comes to actually delivering value. IBM is leading the US in Gen AI patents, but nowhere close in value or deliver.


And there seems to be a horde of innovation on the hardware side, from amazing new design and process breakthroughs at companies like Cerebras, to packaging breakthroughs all over, to optical computation breakthroughs like LightMatter.
 
Fred Chen said:
Then the naturally low yields for these large dies won't matter (as much) as they can be repaired. So it begs the question of whether Ascend has the redundancy built in.
Click to expand...
There are some indicators that that might be the case. The highest end variant boasts fewer cores than the original 910, though the 910B cores include an extra vector unit. Interesting analysis here:

cset.georgetown.edu

Pushing the Limits: Huawei's AI Chip Tests U.S. Export Controls | Center for Security and Emerging Technology

Since 2019, the U.S. government has imposed restrictive export controls on Huawei—one of China’s leading tech giants—seeking, in part, to hinder the company’s AI chip development efforts. This data snapshot reveals how exactly Huawei’s latest AI chip—the Ascend 910B—improves on the prior...
cset.georgetown.edu cset.georgetown.edu
 
soAsian said:
That doesn't mean anything. What is important is the quality of the patent
Click to expand...
Do you know something that suggests low quality of Chinese patents? It's possible that they have more and higher quality patents. Besides, the fact that China is one of the leaders in AI is widely recognized. For example: China leads the U.S. as a top producer of research in more than half of AI's hottest fields, according to new data from Georgetown University's Center for Security and Emerging Technology (CSET) shared first with Axios.
 
lilo777 said:
Do you know something that suggests low quality of Chinese patents? It's possible that they have more and higher quality patents. Besides, the fact that China is one of the leaders in AI is widely recognized. For example: China leads the U.S. as a top producer of research in more than half of AI's hottest fields, according to new data from Georgetown University's Center for Security and Emerging Technology (CSET) shared first with Axios.
Click to expand...
Quantity of patents don’t matter when it comes to quality and value of commercial products. IBM is an excellent example of why the quantity of Gen AI patents is meaningless when it comes to successful commercialization
 
KevinK said:
Quantity of patents don’t matter when it comes to quality and value of commercial products. IBM is an excellent example of why the quantity of Gen AI patents is meaningless when it comes to successful commercialization
Click to expand...
Well said. I never understand why people think a lot of patents equals advanced AI. That's like saying a patent troll created the most valuable patent when, in fact, it can be overturned in court.
 
KevinK said:
Quantity of patents don’t matter when it comes to quality and value of commercial products. IBM is an excellent example of why the quantity of Gen AI patents is meaningless when it comes to successful commercialization
Click to expand...
Both of you are implying that Chinese patents have low quality without any evidence. I suggest that in absence of evidence it would be better to assume that the quality is the same and thus the quantity does matter. More importantly, there is sufficient evidence that your low quality suspicions are off base. China trounces U.S. in AI research output and quality.
 
lilo777 said:
Both of you are implying that Chinese patents have low quality without any evidence. I suggest that in absence of evidence it would be better to assume that the quality is the same and thus the quantity does matter. More importantly, there is sufficient evidence that your low quality suspicions are off base. China trounces U.S. in AI research output and quality.
Click to expand...
Read again - I’m not implying that at all. I’m pointing out that successful products and commercialization do not correlate or rely on patents and that number of patents, Chinese or US (IBM) are a poor indicator of how well companies / countries are going to do with a new technology.
 
Jert said:
how do you explain the desperation in China trying to get their hands on any of the NVIDIA chips especially the advanced ones,
Click to expand...
I'm wondering how many of those patents are on the back of CUDA infrastructure and NVIDIA hardware. As us hardware guys say about software, "It's gotta run on something !". And each hardware platform is quite different.
 
You must log in or register to reply here.
Back
Top