The U.S. is looking for every loophole that can help Huawei in the advanced chipmaking process. A recent report said that the Biden administration may soon add more Chinese firms to the Entity list, blocking the growth of the chip network in China.

Apart from keeping an eye on China, the U.S. is now investigating other foreign suppliers who can lend a hand to Huawei in semiconductor manufacturing.

Inputs reveal that the U.S. Commerce Department started probing TSMC for chip dealings with Huawei. The investigation is in the early stages and aims to check if the chipmaker has been making any AI or phone chips for Huawei for the time being.

If TSMC is found breaching the U.S. export regulations, it might face huge penalties or even a prohibition on using advanced chipmaking tools or techs temporarily.

Under this investigation, the U.S. authorities have targeted the Mate 60 series’ Kirin 5G 9000s chip and recently developed AI Ascend processor. Both these chipsets significantly contribute to China’s progress in the semiconductor industry.

The Commerce Department inquired TSMC about having any direct or indirect (via different names) chip shipments to Huawei in the past or at present. Investigators are also looking into other related aspects like “know-your-customer checks”.

Issues for TSMC

Further details note that the investigation could trouble TSMC to some extent. The Taiwanese chipmaker is one of the key chipset suppliers for U.S. firms.

Huawei AI chipsets (Image Credits: Huawei)

For instance, Apple and Nvidia mainly rely on TSMC for their processors. Also, the Biden administration was planning a $6.6 billion subsidy for the chipmaker to develop its third fabrication unit in Arizona. However, an investigation could halt benefits for some time.

TSMC responded to this investigation and said that it follows all the U.S. export regulations without fail. Although if there is a problem, then it will take action to guarantee compliance in all possible aspects.

“TSMC is a law-abiding company and we are committed to complying with all applicable regulations. If we have any reason to believe there are issues, we will take prompt action to ensure compliance.”

HuaweiCentral’s Take

The U.S. will try every effort to push back Huawei as done in 2019. But this time, the Chinese OEM is focusing on being self-reliant. A recent report claimed that Huawei Kirin 7nm chip performance can match the level of the 5nm processor of TSMC.

How far the U.S. will stretch this investigation on TSMC chip dealings with Huawei would be worth looking forward to.

U.S. investigates TSMC on chip dealings with Huawei (Image Credits: Huawei/TSMC)

(source)

The post U.S. investigates TSMC on chip dealings with Huawei appeared first on Huawei Central.

China Telecom is one of the world’s largest mobile network service providers and the second-largest wireless carrier in China. Besides providing useful services and seamless experiences, it also develops large AI models with advanced capabilities.

The telecom recently developed two large language models that were trained using Chinese AI chips. One of these is TeleChat2-115B while the other one is unnamed.

In a statement, the telecom operator said that its open-source large AI models rely on tens of thousands of domestically produced chips. One of these models has 1 trillion parameters. The effectiveness of AI models mostly depends on the parameter scales.

The AI Institute at China Telecom shared the news via WeChat and further stated: “… the achievement indicates that China has truly realized total self-sufficiency in domestic LLM training.”

China Telecom reportedly chose Huawei AI chips over Nvidia (Image Credits: Huawei)

Huawei AI Chips

China Telecom didn’t reveal the chip supplier’s name. But it has a cooperation with Huawei. Both firms have developed several network solutions that can benefit both operators and consumers, providing a rich and efficient network experience.

Earlier, the telecom operator mentioned that it is making new AI models with Huawei Ascend chips. China has already started sidelining Nvidia and other foreign chipmakers to a greater extent, promoting native suppliers and self-reliance.

This will also help to counter the U.S. sanctions that have been a major hurdle in China’s progress in the chip field. Thus, China Telecom would likely have used Huawei chips for its newly developed AI models over Nvidia and any other U.S. alternative.

Apart from Huawei, another AI chip supplier could be the Cambricon. These companies would have helped China Telecom in building the new AI models with immense efficiency.

Huawei has started sampling the new Ascend 910C chips to some Chinese AI firms. The company is expected to officially introduce the new processor this month. Ascend 910B is expected to extend the Chinese customers for the company in the high-end AI chip market.

China Telecom reportedly chose Huawei AI chips over Nvidia (Image credits: China Telecom)

|| Source ||

The post China Telecom reportedly chose Huawei AI chips over Nvidia appeared first on Huawei Central.

U.S. restrictions have turned many Chinese firms to Huawei in the search for AI chipsets. And ByteDance is no exception to this theory. A new report states that ByteDance plans to develop and train its large language AI model with Huawei chips.

In this AI model-making, the Ascend 910B chips are in consideration. The respective processors can outperform Nvidia’s A100 and claim to offer better GPU performance. It provides 20% more efficient computing power than Nvidia’s chipsets.

The report further notes that ByteDance’s new model will likely be less powerful in computing parameters and capability than its Doubao. Yet it would be efficient.

AI training models need huge amounts of data and a powerful AI chip that can easily manage it. Even though Nvidia premium GPUs are the perfect choice for AI models, the U.S. sanctions forced China to look for other alternatives over Nvidia Chinese chip variants.

ByteDance hasn’t confirmed the news yet. Although three of the persons related to this matter said that the app developer could take this big step to pace ahead in the AI race. Another person said that it’s still unclear if Huawei chips will be used.

Meanwhile, a TikTok spokesperson – Michael Hughes responded on this matter, saying: “The entire premise here is wrong. No new model is being developed.”

Ascend 910B is familiar for its efficiency. Hence, if ByteDance ever prefers making an AI model then it might opt for Huawei’s chips. The company has already ordered 100000 Ascend 910B chips starting this year. But till now, it July received 30000.

This tight supply of Ascend 910B ultimately made ByteDance halt the development of its new AI model. The company has been one of the biggest customers of Huawei’s AI chips. On the other hand, it’s the top user of Nvidia H20 AI processors.

The overall situation regarding the ByteDance partnership with Huawei over a new AI model is quite unclear. However, the matter may soon come into clear light with more details. ByteDance AI model running on Ascend 910B will surely help Huawei reach new heights in the AI chip market.

ByteDance to build new AI model using Huawei chips, says report (Image Credits: Huawei)

(Source)

The post ByteDance to build new AI model using Huawei chips: Report appeared first on Huawei Central.

Financial Times notes that many Huawei customers complained about the buggy performance of Ascend AI chips, resulting in stability issues. An AI engineer stated that the processor frequently crashes leading to difficult and unstable work.

A major reason behind the stability issues of Huawei AI chips is its buggy software called “Cann”. The Huawei-developed software launched as an alternative to Nvidia’s Cuda. It was expected to help in enhancing the chip performance for developers.

But it seems the software has run into a strange bug that caused the AI chip to perform poorly in various training models. It also showed a slower inter-chip connection. Several Huawei employees also poured out their opinions and said:

“When random errors occur, it is very difficult to find out where it comes from due to poor documentation. You need talented developers to read the source code to see what the issue is, which slows everything down. The coding is imperfect.”

A Huawei researcher mentioned that bad results are a part of the journey and mostly occur when developers have the least knowledge of how the hardware works.

Stability issues hold back Huawei AI chips to outpace Nvidia: Report (Image Credits: Huawei)

Even after several attempts, the AI chip performance remained the same for developers. An AI engineer said that Huawei AI chips still lagged far from Nvidia due to stability problems. They aren’t contributing well to the initial training of models.

Solution:

Good news is that Huawei has started sending engineers that can help its customers to tackle the problem. The report reveals that experienced developers transfer the training code written on Cuda into Cann to ease the overall operation.

Baidu, Tencent, and iFlytek are the major companies that now have a new team of engineers to solve the AI chip-related problems. On this matter, a tech analyst said:

“Huawei’s advantage over Nvidia is it can work closely with its customers. Unlike Nvidia, it has a large team of engineers to help solve clients’ problems and get them to transition to their hardware.”

Apart from sending teams, the company also opened an online portal that can help customers with constant services and feedback to improve its buggy software.

The Chinese tech giant has a huge workforce that can help it resolve the AI chip problem to some extent. However, the overall scenario suggests that Huawei AI chips still have room for improvement. To hold back Nvidia, China needs more strength and technical support.

Huawei AI chipsets (Image Credits: Huawei)

[Source – Financial Times]

The post Stability issues hold back Huawei AI chips to outpace Nvidia: Report appeared first on Huawei Central.

Wall Street Journal reported that Huawei is readying the Ascend 910C AI chip in China. The company has started testing the chipset with different Chinese internet and telecom partners to inspect its overall potential as well as performance efficiency.

The report reveals that firms like ByteDance, Baidu, and China Mobile have seemingly obtained the Ascend 910C chipset. With a few test results, the Chinese tech maker informed its clients that the latest AI processor can easily rival with Nvidia H100.

Note that the company has been doing these trials for the past few weeks. It means that the new chipset is almost ready. Adding more details, WSJ notes that the Chinese manufacturer could begin shipping the chip by October this year, at the earliest.

WSJ further states that Huawei and its potential clients have negotiated for 70000+ chip orders. These are likely to be worth $2 billion. With a new AI processor, Huawei would be able to draw level with its competitors in the smart AI chipsets race.

Huawei silently testing Ascend 910C AI chip to rival Nvidia: Report (Image Credits: Huawei)

Huawei vs Nvidia

It seems Huawei is making efforts to fill the gaps left by Nvidia in its home ground. Launched last year, the Nvidia H100 never came directly to China, due to U.S. ban. The chipmaker was commanded not to ship any advanced processor in the region.

Even though the company introduced three new chips for China, they were of low quality and didn’t meet industrial needs. As a result, Huawei made a step. The OEM is still struggling with production delay issues, yet the latest report implies that the tech giant is managing to break off the U.S. sanctions and become independent.

Nvidia, on the other hand, is working on a new China-oriented chip B20. But in case, it doesn’t comply with U.S. regulations, it won’t be shipped to China. Over this, an analyst at SemiAnalysis research firm says that he sees Huawei Ascend 910C as an advanced chip that could perform better than the Nvidia B20 processor.

Huawei AI chips (Image Credits: Huawei)

|| Source – Wall Street Journal ||

The post Huawei silently testing Ascend 910C AI chip to rival Nvidia: Report appeared first on Huawei Central.

The CEO of Huawei Cloud – Zhang Pingan attended the World AI Conference in Shanghai on Thursday (July 4). While delivering the keynote, he cited that the lack of AI chips won’t pull back China from its ambitions towards AI technology.

Zhang said that though Huawei is facing limited computing power in the region, this doesn’t make it weak. The company can’t completely depend on advanced manufacturing processes as the ‘one and only’ foundation for its AI infrastructure.

“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 Pingan.

Huawei Cloud CEO has said these words amid the news of AI chipset production problems. A recent headline read that the company is struggling with a lack of AI chips yield rate at SMIC foundry. Four out of five chips appeared to be defective.

Shortage of advanced chips will not affect Huawei AI goals (Image Credits: Huawei)

Due to tighter U.S. restrictions on advanced chip-making tools, the tech maker has to rely upon the outdated machinery that now started to fail. Huawei also has to compete with rivals like Nvidia that are looking to capture the Chinese market.

Many Chinese firms are supporting Huawei by purchasing Ascend AI chips rather than Nvidia A100. Yet the lack of Huawei advanced AI chips might cause new partners to reluctantly turn to other options available in the market.

Zhang further talks about cloud innovation. He hopes that advanced and innovative cloud products can pay back for the shortage of AI chips in the native region.

According to him, innovation in other infrastructures such as the Cloud, network solutions, and more can help address the AI chip lack issues. A combined approach can help reduce energy consumption as well as boost efficiency.

But how long the shortage of advanced AI chips will persist and what measures Huawei will take to overcome these problems would be worth looking forward to.

[Source – Reuters]

The post Shortage of advanced chips will not affect China’s AI goals: Huawei appeared first on Huawei Central.

Whether it’s Kirin processors or Ascend AI chips, Huawei is working hard on both fronts. Before checking the latest developments, let’s go through a small flashback.

The “U.S. bans Huawei” is not a new story to Chinese consumers. In 2019, the American government blacklisted Huawei in the name of U.S. security concerns. This ultimately cut the supply of highly advanced chipmaking tools to China.

Before this ban, Huawei used to print its chip through TSMC. However, foreign trade restrictions led the Taiwan chip maker to give up on new Kirin processor production. As a result, Huawei put a halt on its high-end Kirin chipset project.

Turning point:

Although the tables turned in August 2023 when Huawei introduced the Mate 60 series with a 7nm Kirin 9000s 5G chip. The smartphone launch not only caused a stir in the Chinese market but also gave a headache to the U.S. Commerce Department.

Even though the chip has been developed using old equipment and tech materials, it was enough to show that Huawei is coming back on track. The U.S. became busy finding why, where, and how the chip was made. Meanwhile, Huawei started looking at how to increase its chip yields.

Major aids:

The Chinese phone maker started teaming up with native partners like SMIC for its chip development. As per the details, the Kirin 9000s uses an N+2 7nm tech from SMIC. More new Kirin chips based on SMIC tech are also rumored to be underway.

Kirin chipset (Image Credits: HiSilicon Kirin)

Huawei is also taking help from the stocked chip-making equipment by ASML to develop new processors. The Chinese government, on the other hand, has eventually begun backing up the company with more funding for chip development.

Huawei further believes that if they don’t start using self-made chips now, the tech gap will continue to persist. It seems the U.S. sanctions appeared as an opportunity for the firm to become independent.

New development:

In April 2024, Huawei introduced the Pura 70 Ultra with a new Kirin 9010 chip. TechInsights found that the new processor belongs to the SMIC foundry. This news has once again created a buzz in the U.S. on Huawei’s progress.

Both Huawei and SMIC are now in the headlines for developing a 5nm chip. The latter is said to lay chip manufacturing lines in Shanghai for mass production. Rumors regarding 5nm and 3nm chip development are also flying in the air.

Although Huawei put a full stop to these talks by saying that the company is currently focusing on resolving 7nm issues instead of making new developments.

Huawei AI chip is another end of the rope. Ascend 910B has become quite popular in China. Chinese firms are constantly picking the respective chip over the U.S. alternatives such as Nvidia A100. Tests even proved that Huawei AI processors are more efficient than that of Nvidia.

Huawei AI chips (Image Credits: Huawei)

A recent rumor has added that Huawei is now testing powerful Taishan cores that are 175% faster than the Kirin 9000s. Hence, we might see a dynamic 5nm processor with the Mate 70 series later this year.

Rivals like TSMC says that Huawei will take years to catch up in the chip race. But regardless of the tough U.S. ban and trade challenges, Huawei is trying its best to rise in the chip production game. It would be interesting to see how Huawei will win this game!

Despite U.S. ban, Huawei trying well to catch up with rivals in the chip race (Image Credits: Huawei)

The post Despite U.S. ban, Huawei trying well to catch up with rivals in the chip race appeared first on Huawei Central.

A few industry analysts reported that Huawei is recovering in its server business despite being chained in U.S. sanctions for years. The credit goes to the heavy demand for the company’s AI chips over other alternatives in the home ground.

The Chinese OEM’s server business depends on two chips: Kunpeng for common servers and Ascend for AI products. Many computer developers like Talkweb, Huakun, XFusion, and more extended their hands to Huawei, giving it strength.

This has ultimately impacted the growth of other tech players in the market. Shenzhen Stock Exchange listing of firms like Digital China Group suddenly dropped by 10% from their daily limit this week and continues to remain down.

Huawei on the other hand, is expected to step into the Chinese server sector as a major assembler rather than being a component (like AI chips) supplier. Although the company has not decided any such adjustments to its business so far.

But if true, this move by the phone maker could shake the performance of other companies, making them just a distributor. Digital China Group is one example that teamed up with telecoms to offer effective server facilities to its customers.

Huawei AI chipsets (Image Credits: Huawei)

Flashback

Back in 2018, Digital China joined forces with Huawei and initiated its ‘Greater Huawei’ scheme to act as a global distributor for the company. Some major products like Kuntai run on Kunpeng and Ascend chips.

Things drifted apart when Huawei faced U.S. sanctions in 2019, which made it dispose of its server subsidiary since it couldn’t purchase the Intel x86 chips.

Digital China says Kunpeng and Ascend products only have 3% of its total revenue. But Huawei has gradually stepped on the progress path and may rise back in this sector.

Ascend 910B AI chip demand has helped the company to turn back into its progressive mode. As said by a Huawei executive, the chip is far more efficient and reliable than its competitor – Nvidia A100.

Constant U.S. pressure is another reason that Chinese company turned their faces towards native products rather than relying on foreign ones.

Huawei regains power in Chinese server sector due to AI chips demand (Image Credits: Huawei)

|| Source – SCMP ||

The post Huawei regains power in Chinese server sector due to AI chips demand: Report appeared first on Huawei Central.

According to the details, Huawei invests fresh capital into the Shenzhen Habo Technology Partners subsidiary in favor of growing demand for 5G phones and AI chips. Last Thursday, the company injected around 7.98 billion yuan ($1.1 billion).

Huawei seems making efforts to proceed with its growth in the right direction. Established in 2021, the Shenzhen Habo Investment firm already gained 7 billion yuan capital. However, the tech giant is now adding some more amount to it.

On the other hand, estimations say that Huawei gained 9% sales revenue YoY to 700 billion yuan in 2023 due to its 5G handsets’ boosting demand among consumers. As of now, the company foresees a new scope through this investment.

The Shenzhen subsidiary is a joint venture developed by Huawei and Habo. The latter focuses on chipsets, enterprise services, and materials for various products. It has investors like Peifengtunan (automation software provider) and Open Valley (promoter of enterprise adoption of HarmonyOS).

Huawei and Habo (Source: SCMP)

Earlier, Habo used to help Huawei reform its chip supply chain when it was put behind sanction bars and restricted to trade in new and advanced technologies.

By the end of 2023, Habo made 97 investments, the majority covering start-ups in the semiconductor industry. This has further helped Huawei to widen its enterprise services and other technical fields in the native region.

Past achievements:

Huawei Mate 60 Pro with the Kirin 5G chipset has been written in the history of the company’s success. Besides, the Ascend 910B AI chip with high-end capabilities is also giving a tough spot to the Nvidia A100 GPU processor in China.

Perhaps, a new investment in the Shenzhen-based subsidiary will open new ways for Huawei’s growth as well as development in its home ground in the time ahead.

Huawei brand (Source: SCMP)

(Source)

The post Huawei invests more in its subsidiary to meet 5G phones and AI chips demand: Report appeared first on Huawei Central.

Exploring chips from Huawei and Nvidia, some analysts depict that the Ascend 910B processor is enough capable of competing with the Nvidia A100 AI chip. The former’s chipsets are superior in both theoretical and performance terms over the latter.

Notably, the A100 is just a level below Nvidia’s newly launched H100 GPU processor. Moreover, Huawei Ascend 910B has competitive capabilities for AI usage.

Huawei Ascend 910B is a self-designed processor, produced by using the SMIC 7nm process technology. Despite challenges and the U.S. ban, the tech giant aims to boost its semiconductor business in the home ground and continues to meet new suppliers.

On the other hand, Ascend 910B received a good response in the market. The Baidu firm placed a big order for these chipsets whereas, iFlytech created a large language model using these AI processors. It looks like those who can’t access Nvidia chips are now chasing Huawei.

Nvidia CEO (Credits: The JoongAng)

The information further reveals that Huawei is transforming itself into a powerful technology-based company rather than just being a phone maker or an equipment supplier. Perhaps, this may appear as a major threat to many AI players.

A few official executives in the semiconductor believe that even though China doesn’t have the latest technologies, it can still succeed due to its expertise in chipset design.

“China is more industrialist than Korea in terms of semiconductor design. Due to design capabilities, the region can produce as many AI chips as possible.”

Huawei and SMIC team up!

The Chinese government is constantly supporting its companies to grow in all fields including AI. Meanwhile, SMIC plans to begin the 5nm chips production line this year with Huawei’s Kirin processors. In addition, a few analysts predict that the chipmaker formed an internal research and development team for a 3nm chip.

SMIC plans to increase its facility investment by over 20% this year. At the same time, it sees the 5nm chip as a breakthrough this year. Hence, things seem to be running in favor of Huawei and Chinese companies at the moment.

(Source)

The post Huawei Ascend 910B AI chip seems as strong as Nvidia processor: Report appeared first on Huawei Central.