The Next Front in the U.S.-China Battle Over Chips

bnew

Veteran
Joined
Nov 1, 2015
Messages
57,570
Reputation
8,519
Daps
160,381
.https://www.nytimes.com/2024/01/10/technology/risc-v-china-united-states-chips-security.html?unlocked_article_code=1.Mk0.iR7l.PFiJCmnGgCh9&smid=url-share

The Next Front in the U.S.-China Battle Over Chips

A U.S.-born chip technology called RISC-V has become critical to China’s ambitions. Washington is debating whether and how to limit the technology.


09CHIPS-RISCV-videoSixteenByNine3000.jpg

CreditCredit...Nick Little

By Don Clark and Ana Swanson

Don Clark, who covers the chip industry, reported from San Francisco and Santa Clara, Calif., and Ana Swanson, who covers trade, reported from Washington.

Jan. 10, 2024, 5:01 a.m. ET

NASA has chosen the technology to help it land future spacecraft on unmapped planets. Meta uses the technology for artificial intelligence. Chinese engineers have turned to it to encrypt data.

And it could represent the next front in the semiconductor trade war between the United States and China.

The technology is RISC-V, pronounced “risk five.” It evolved from a university computer lab in California to a foundation for myriad chips that handle computing chores. RISC-V essentially provides a kind of common language for designing processors that are found in devices like smartphones, disk drives, Wi-Fi routers and tablets.

RISC-V has ignited a new debate in Washington in recent months about how far the United States can or should go as it steadily expands restrictions on exporting technology to China that could help advance its military. That’s because RISC-V, which can be downloaded from the internet for free, has become a central tool for Chinese companies and government institutions hoping to match U.S. prowess in designing semiconductors.

Last month, the House Select Committee on the Chinese Communist Party — in an effort spearheaded by Representative Mike Gallagher, Republican of Wisconsin — recommended that an interagency government committee study potential risks of RISC-V. Congressional aides have met with members of the Biden administration about the technology, and lawmakers and their aides have discussed extending restrictions to stop U.S. citizens from aiding China on RISC-V, according to congressional staff members.

The Chinese Communist Party is “already attempting to use RISC-V’s design architecture to undermine our export controls,” Representative Raja Krishnamoorthi of Illinois, the ranking Democrat on the House select committee, said in a statement. He added that RISC-V’s participants should be focused on advancing technology and “not the geopolitical interests of the Chinese Communist Party.”

Arm Holdings, a British company that sells competing chip technology, has also lobbied officials to consider restrictions on RISC-V, three people with knowledge of the situation said. Biden administration officials have concerns about China’s use of RISC-V but are wary about potential complications with trying to regulate the technology, according to a person familiar with the discussions. The Department of Commerce and the National Security Council declined to comment.



The Global Race for Computer Chips​



The debate over RISC-V is complicated because the technology was patterned after open-source software, the free programs like Linux that allow any developer to view and modify the original code used to make them. Such programs have prompted multiple competitors to innovate and reduce the market power of any single vendor.

But RISC-V is not code that can directly be used to make anything. It is a set of basic computing instructions that determine the calculations a chip can perform. Engineers can download these instructions and incorporate them in the much more complex task of creating design blueprints for parts of a semiconductor. Many companies sell RISC-V chip designs, and some universities and other institutions distribute them free.

As with Linux — but not technologies from companies like Arm and Intel — engineers around the world can make suggestions to enhance the underlying instructions. That process is overseen by RISC-V International, a nonprofit with more than 4,000 members — including the Chinese Academy of Sciences and Chinese firms like Huawei and Alibaba, as well as Google and Qualcomm — in 70 countries.

The group changed its incorporation from the United States to Switzerland in 2020 to calm “concerns of political disruption” and control by any single country. Its leaders said their model mirrored that of other international groups that govern standard technologies like Ethernet and Wi-Fi.

“Open standards have been around for 100 years,” Calista Redmond, chief executive of RISC-V International, said in an interview. “This is no different.”

Open-source technologies have generally been granted exceptions to U.S. export controls. Any change to that treatment “is certainly going to raise thorny legal issues and important public policy concerns,” said Daniel Pickard, a lawyer specializing in trade and national security at Buchanan Ingersoll & Rooney.

U.S. regulations limit Arm and RISC-V companies from exporting chip designs to China based on certain performance limits. But trying to restrict the underlying instructions is like trying to control words or letters, Silicon Valley executives said.

“It is absolutely silly,” said Dave Ditzel, the chief technology officer of Esperanto Technologies, a chip start-up that uses RISC-V. “It’s like saying, ‘Well, the Chinese can read a book on nuclear weapons that’s written in English, so let’s solve the problem by banning the English alphabet.’”

As RISC-V helps Chinese firms including Huawei design more of the world’s semiconductors, some U.S. officials have raised concerns that Beijing could use Chinese foundries to insert cyber vulnerabilities into chips that may be used to cripple American electrical grids and other critical infrastructure.

RISC-V backers counter that technologies with inner details that can be openly studied are much more secure. Any new restrictions, RISC-V backers said, would weaken U.S. influence over the technology while doing little to hold China back because the instruction set is already widely distributed.

The original inspiration for RISC-V was saving money. Starting in 2010, a professor and two graduate students began developing a new instruction set based on technology pioneered by David Patterson, a computer science professor at the University of California, Berkeley, who had helped invent reduced instruction set computing, or RISC. The aim was to help study the inner workings of computing without having to pay Arm, which charges royalties for every chip that uses its technology.

“I just wanted to learn how to build computers,” said Yunsup Lee, one of the graduate students, who now works at SiFive, a start-up that sells RISC-V designs. Then the goal evolved “to benefit everybody in the world,” he said.

The RISC-V variant swiftly attracted interest among engineers. Having a standard set of instructions can allow software programs to work on all chips that use them.

In China, engineers and officials were also quick to see the potential, viewing open-source technology as a way to become self-sufficient and counter risks like embargoes and supply interruptions, Ni Guangnan, a researcher at the Chinese Academy of Sciences, wrote in an article about RISC-V in June.

In 2019, Mr. Patterson, who now works at Google, helped establish a RISC-V lab in Shenzhen, China, which was supported by an institute set up earlier by Berkeley and Tsinghua University in China. Representative Gallagher, in a video his committee released in November, expressed concerns about the professor’s work and collaboration between the institute and organizations with links to Chinese military and intelligence activities.

Mr. Patterson declined to comment through a Google spokeswoman.

A U.C. Berkeley spokesman said that the university’s work with the institute had been basic research that was unrestricted, and that the university was responding to requests for information from Congress.

More than 100 “significant” Chinese companies are designing chips with RISC-V today, as are at least 100 more start-ups, said Handel Jones, an analyst at International Business Strategies. Many of the applications are in fairly mundane consumer products, but engineers believe the technology will eventually take over some of the most demanding tasks.

Chinese aerospace scientists have proposed using RISC-V to develop high-performance spaceborne computers. Other Chinese companies and institutions are aiming to string together RISC-V processors to run bigger jobs in data centers, including A.I. applications.

At a RISC-V conference in Silicon Valley in November, T-Head, Alibaba’s semiconductor subsidiary, discussed RISC-V designs that Sophgo, another Chinese company, used in a chip powering a large server deployed at Shandong University in China. It’s the first instance of RISC-V technology’s running a cloud-style computing service, the companies said.

“We just made a small step, but we put RISC-V on the starting line,” David Chen, ecosystem director at Alibaba, said at the event.
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
57,570
Reputation
8,519
Daps
160,381

Report: Black market keeps Nvidia chips flowing to China military, government​

Unknown suppliers keep Nvidia's most advanced chips within China's reach.​

ASHLEY BELANGER - 1/15/2024, 2:01 PM

An Nvidia H100 graphics processor chip.

Enlarge / An Nvidia H100 graphics processor chip.

31

China is still finding ways to skirt US export controls on Nvidia chips, Reuters reported.

A Reuters review of publicly available tender documents showed that last year dozens of entities—including "Chinese military bodies, state-run artificial intelligence research institutes, and universities"—managed to buy "small batches" of restricted Nvidia chips.

The US has been attempting to block China from accessing advanced chips needed to achieve AI breakthroughs and advance modern military technologies since September 2022, citing national security risks.

Reuters' report shows just how unsuccessful the US effort has been to completely cut off China, despite repeated US attempts to expand export controls and close any loopholes discovered over the past year.

China's current suppliers remain "largely unknown," but Reuters confirmed that "neither Nvidia" nor its approved retailers counted "among the suppliers identified."

An Nvidia spokesperson told Reuters that the company "complies with all applicable export control laws and requires its customers to do the same."

"If we learn that a customer has made an unlawful resale to third parties, we'll take immediate and appropriate action," Nvidia's spokesperson said.

It's also still unclear how suppliers are procuring the chips, which include Nvidia's most powerful chips, the A100 and H100, in addition to slower modified chips developed just for the Chinese market, the A800 and H800. The former chips were among the first banned, while the US only began restricting the latter chips last October.

Among military and government groups purchasing chips were two top universities that the US Department of Commerce has linked to China's principal military force, the People's Liberation Army, and labeled as a threat to national security. Last May, the Harbin Institute of Technology purchased six Nvidia A100 chips to "train a deep-learning model," and in December 2022, the University of Electronic Science and Technology of China purchased one A100 for purposes so far unknown, Reuters reported.

Other entities purchasing chips include Tsinghua University—which is seemingly gaining the most access, purchasing "some 80 A100 chips since the 2022 ban"—as well as Chongqing University, Shandong Chengxiang Electronic Technology, and "one unnamed People's Liberation Army entity based in the city of Wuxi, Jiangsu province."

In total, Reuters reviewed more than 100 tenders showing state entities purchasing A100 chips and dozens of tenders documenting A800 purchases. Purchases include "brand new" chips and have been made as recently as this month.

Most of the chips purchased by Chinese entities are being used for AI, Reuters reported. None of the purchasers or suppliers provided comments in Reuters' report.

Nvidia's highly sought-after chips are graphic processing units capable of crunching large amounts of data at the high speeds needed to fuel AI systems. For now, these chips remain irreplaceable to Chinese firms hoping to compete globally, as well as nationally, with China's dominant technology players, such as Huawei, Reuters suggested.

While the "small batches" of chips found indicate that China could still be accessing enough Nvidia chips to enhance "existing AI models," Reuters pointed out that US curbs are effectively stopping China from bulk-ordering chips at quantities needed to develop new AI systems. Running a "model similar to OpenAI's GPT would require more than 30,000 Nvidia A100 cards," research firm TrendForce reported last March.

For China, which has firmly opposed the US export controls every step of the way, these curbs remain a persistent problem despite maintaining access through the burgeoning black market. On Monday, a Bloomberg report flagged the "steepest drop" in the value of China chip imports ever recorded, falling by more than 15 percent.

China’s black market for AI chips​

The US still must confront whether it's possible to block China from accessing advanced chips without other allied nations joining the effort by lobbying their own export controls.

In October 2022, a senior US official warned that without more cooperation, US curbs will "lose effectiveness over time." A former top Commerce Department official, Kevin Wolf, told The Wall Street Journal last year that it's "insanely difficult to enforce" US export controls on transactions overseas.

Part of the problem, sources told Reuters in October 2023, is that overseas subsidiaries were "easily" smuggling restricted chips into China or else providing remote access to chips to China-based employees.

On top of that activity, a black market for chips developed quickly, selling "excess stock that finds its way to the market after Nvidia ships large quantities to big US firms" or else chips imported "through companies locally incorporated in places such as India, Taiwan, and Singapore," Reuters reported.

The US has maintained that its plan is not to ensure that China has absolutely no access but to limit access enough to keep China from getting ahead. But Nvidia CEO Jensen Huang has warned that curbs could have the opposite effect. While China finds ways to skirt the bans and acquire chips to "inspire" advancements, US companies that have been impacted by export controls restricting sales in China could lose so much revenue that they fall behind competitively, Huang predicted.

One example likely worrying to Huang and other tech firms came last November, when Huawei shocked the US government by unveiling a cutting-edge chip that seemed to prove US sanctions weren't doing much to limit China's ability to compete.
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
57,570
Reputation
8,519
Daps
160,381

China Is Closing the A.I. Gap With the United States​


In recent weeks, Chinese tech companies have unveiled technologies that rival American systems — and they are already in the hands of consumers and software developers.



A.I. generated videos created from text prompts using Kling, a video generator made by the Chinese company Kuaishou.


  1. Prompt: “The astronaut jumps up from the moon’s surface and launches himself into space.”

    Kuaishou


  2. Prompt: “A giant panda is playing guitar by the lake.”

    Kuaishou


  3. Prompt: “A Chinese boy wearing glasses is eating a delicious cheeseburger in a fast food restaurant, with his eyes closed for enjoyment.”

    Kuaishou


  4. Prompt: “A couple is holding hands and walking in the starry sky, while the stars move dramatically in the background.”

    Kuaishou


  5. An A.I. generated video created from an archival photo without using text prompts.

    Kuaishou

By Meaghan Tobin and Cade Metz

Meaghan Tobin reported from Shanghai, and Cade Metz from San Francisco.

July 25, 2024
阅读简体中文版閱讀繁體中文版

At the World Artificial Intelligence Conference in Shanghai this month, start-up founder Qu Dongqi showed off a video he had recently posted online. It displayed an old photograph of a woman with two toddlers. Then the photo sprang to life as the woman lifted the toddlers up in her arms and they laughed with surprise.

The video was created by A.I. technology from the Chinese internet company Kuaishou. The technology was reminiscent of a video generator, called Sora, that the American start-up OpenAI unveiled this year. But unlike Sora, it was available to the general public.

“My American friends still can’t use Sora,” Mr. Qu said. “But we already have better solutions here.”

A.I. generated videos created from text prompts using Kling, a video generator made by the Chinese company Kuaishou.

  1. Prompt: “Mona Lisa puts on glasses with her hands.“

    Kuaishou
  2. https://vp.nyt.com/video/2024/07/23/121345_1_mosaic-ai-china-ai-cropped-23-2-627_wg_720p.mp4

    Prompt: “Einstein plays guitar.”

    Kuaishou
  3. https://vp.nyt.com/video/2024/07/23/121347_1_mosaic-ai-china-ai-cropped-23-4-724_wg_720p.mp4

    Prompt: “Kitten riding in an airplane and looking out the window.”

    Kuaishou
  4. https://vp.nyt.com/video/2024/07/23/121346_1_mosaic-ai-china-ai-cropped-23-3-440_wg_720p.mp4

    Prompt: “Cute shepherd dog running, tennis ball bouncing, warm atmosphere.”

    Kuaishou
  5. https://vp.nyt.com/video/2024/07/23/121348_1_mosaic-ai-china-ai-cropped-23-5-2-673_wg_720p.mp4

    Prompt: “A girl eating noodles.”

    Kuaishou

While the United States has had a head start on A.I. development, China is catching up. In recent weeks, several Chinese companies have unveiled A.I. technologies that rival the leading American systems. And these technologies are already in the hands of consumers, businesses and independent software developers across the globe.

While many American companies are worried that A.I. technologies could accelerate the spread of disinformation or cause other serious harm, Chinese companies are more willing to release their technologies to consumers or even share the underlying software code with other businesses and software developers. This kind of sharing of computer code, called open source, allows others to more quickly build and distribute their own products using the same technologies.

Open source has been a cornerstone of the development of computer software, the internet and, now, artificial intelligence. The idea is that technology advances faster when its computer code is freely available for anyone to examine, use and improve upon.

China’s efforts could have enormous implications as A.I. technology continues to develop in the years to come. The technology could increase the productivity of workers, fuel future innovations and power a new wave of military technologies, including autonomous weapons.
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
57,570
Reputation
8,519
Daps
160,381
When OpenAI kicked off the A.I. boom in late 2022 with the release of the online chatbot ChatGPT, China struggled to compete with technologies emerging from American companies like OpenAI and Google. (The New York Times has sued OpenAI and its partner, Microsoft, claiming copyright infringement of news content related to A.I. systems.) But China’s progress is now accelerating.

Kuaishou released its video generator, Kling, in China more than a month ago and to users worldwide on Wednesday. Just before Kling’s arrival, 01.AI, a start-up co-founded by Kai-Fu Lee, an investor and technologist who helped build Chinese offices for both Google and Microsoft, released chatbot technology that scored nearly as well as the leading American technologies on common benchmark tests that rate the performance of the world’s chatbots.

Kai-Fu Lee smiling for a photo while wearing a tuxedo at a formal event. Other people pass by in the background.


Kai-Fu Lee, a co-founder of the start-up 01.AI. The company unveiled a new version of its technology this year that sits near the top of a leaderboard that ranks the world’s best technologies.Credit...Krista Schlueter for The New York Times

New technology from the Chinese tech giant Alibaba has also leaped to the top of a leaderboard that rates open-source A.I. systems. “We have disproved the commonplace belief that China doesn’t have the talent or the technology to compete with the U.S.,” Dr. Lee said. “That belief is simply wrong.”

In interviews, a dozen technologists and researchers at Chinese tech companies said open-source technologies were a key reason that China’s A.I. development has advanced so quickly. They saw open-source A.I. as an opportunity for the country to take a lead.

But that will not be easy. The United States remains at the forefront of A.I. research. And U.S. officials have resolved to keep it that way.

The White House has instituted a trade embargo designed to prevent Chinese companies from using the most powerful versions of computer chips that are essential to building artificial intelligence. A group of lawmakers has introduced a bill that would make it easier for the White House to control the export of A.I. software built in the United States. Others are trying to limit the progress of open-source technologies that have helped fuel the rise of similar systems in China.

Disclosure:

The New York Times Company has sued OpenAI and Microsoft, claiming copyright infringement of content related to artificial intelligence systems. The companies have sought to dismiss some of the claims. Times reporters have no involvement in the case and remain independent in their coverage.

The top American companies are also exploring new technologies that aim to eclipse the powers of today’s chatbots and video generators.

“Chinese companies are good at replicating and improving what the U.S. already has,” said Yiran Chen, a professor of electrical and computer engineering at Duke University. “They are not as good at inventing something completely new that will bypass the U.S. in five to 10 years.”

But many in China’s tech industry believe that open-source technology could help them grow despite those constraints. And if U.S. regulators stifle the progress of American open-source projects (as some lawmakers are discussing) China could gain a significant edge. If the best open-source technologies come from China, U.S. developers could end up building their systems atop Chinese technologies.

“Open-source A.I. is the foundation of A.I. development,” said Clément Delangue, chief executive of Hugging Face, a company that houses many of the world’s open-source A.I. projects. The U.S. built its leadership in A.I. through collaboration between companies and researchers, he said, “and it looks like China could do the same thing.”


Clément Delangue walking with a group of people outside the U.S. Capitol.


Clément Delangue, right, the chief executive of the A.I. company Hugging Face, said that open-source technology could help China make gains in the field of A.I.Credit...Kenny Holston/The New York Times

While anyone with a computer can change open-source software code, it takes a lot of data, skill and computing power to fundamentally alter an A.I. system. When it comes to A.I., open source typically means that a system’s building blocks serve as a foundation that allows others to build something new, said Fu Hongyu, the director of A.I. governance at Alibaba’s research institute, AliResearch.

As in other countries, in China there is an intense debate over whether the latest technological advances should be made accessible to anyone or kept as closely held company secrets. Some, like Robin Li, the chief executive of Baidu, one of the few companies in China building its own A.I. technology entirely from scratch, think the technology is most profitable and secure when it is closed-source — that is, in the hands of a limited few.

A.I. systems require enormous resources: talent, data and computing power. Beijing has made it clear that the benefits accruing from such investments should be shared. The Chinese government has poured money into A.I. projects and subsidized resources like computing centers.

But Chinese tech companies face a major constraint on the development of their A.I. systems: compliance with Beijing’s strict censorship regime, which extends to generative A.I. technologies.

Kuaishou’s new video generator Kling appears to have been trained to follow the rules. Text prompts with any mention of China’s president, Xi Jinping, or controversial topics like feminism and the country’s real estate crisis yielded error messages. An image prompt of this year’s National People’s Congress yielded a video of the delegates shifting in their seats.

Kuaishou did not respond to questions about what steps the company took to prevent Kling from creating harmful, fake or politically sensitive content.

By making their most advanced A.I. technologies freely available, China’s tech giants are demonstrating their willingness to contribute to the country’s overall technological advancement as Beijing has established that the power and profit of the tech industry should be channeled toward the goal of self sufficiency.

The concern for some in China is that the country will struggle to amass the computing chips it needs to build increasingly powerful technologies. But that has not yet prevented Chinese companies from building powerful new technologies that can compete with U.S. systems.

At the end of last year, Dr. Lee’s company, 01.AI, was ridiculed on social media when someone discovered that the company had built its A.I. system using open-source technology originally built by Meta, owner of Facebook and Instagram. Some saw it as a symbol of China’s dependence on American ingenuity.

Six months later, 01.AI unveiled a new version of its technology. It now sits near the top of the leaderboard that ranks the world’s best technologies. Around the same time, a team from Stanford University in California unveiled Llama 3-V, claiming it outperformed other leading models. But a Chinese researcher soon noticed that the model was based on an open-source system originally built in China.

It was the reverse of the controversy surrounding 01.AI last year: Rather than Chinese developers building atop U.S. technology, U.S. developers built atop Chinese technology.

If regulators limit open-source projects in the United States and Chinese open-source technologies become the gold standard, Mr. Delangue said, this kind of thing could become the norm.

“If the trend continues, it becomes more and more of a challenge for the U.S.,” he said.
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
57,570
Reputation
8,519
Daps
160,381

1/1
@rohanpaul_ai
Chinese LLMs catch up with US LLMs: Stepfun ranks higher than Gemini and Qwen ranks higher than 4o



GdVp2pUaQAAxBFc.png



To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196




1/2
@bensbitesdaily
Is China really behind in AI?

@deepseek_ai just replicated @OpenAI’s method of getting models to think before answering. The resulting models even beat the o1 models at some benchmarks.

What's going on here?

Chinese AI companies are shipping open-source models that match or outperform western counterparts (closed/open) despite GPU restrictions.

What does this mean?

DeepSeek's new R1-Lite model matches o1-preview's performance on AIME & MATH benchmarks—both tests of complex reasoning and mathematical ability. More importantly, like o1, it shows its thinking process in real-time, and performance improves as the model spends more time reasoning. Unlike o1 though, DeepSeek plans to make this completely open source.

Why should you care?

If you're using AI from web products like ChatGPT, Claude or using their cloud APIs, stick with it. The product/developer experience is worth it. But if you’re even thinking about local deployment, DeepSeek and Qwen models (from Alibaba) should be at the top of your mind.



2/2
@bensbitesdaily
Read our tutorial on using local models with LMStudio: Learn to use AI: How to run AI models locally on your PC | Ben's Bites




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196
 
Top