AI chips – Page 2

Nvidia AI chips worth $1B smuggled to China after Trump export controls

AI, AI chips, Biz & IT, NVIDIA, syndication / Rejus Almole / July 25, 2025

Black market for US semiconductors operates despite efforts to curb Beijing’s high-tech ambitions.

Credit: VGG | Getty Images

At least $1 billion worth of Nvidia’s advanced artificial intelligence processors were shipped to China in the three months after Donald Trump tightened chip export controls, exposing the limits of Washington’s efforts to restrain Beijing’s high-tech ambitions.

A Financial Times analysis of dozens of sales contracts, company filings, and multiple people with direct knowledge of the deals reveals that Nvidia’s B200 has become the most sought-after—and widely available—chip in a rampant Chinese black market for American semiconductors.

The processor is widely used by US powerhouses such as OpenAI, Google, and Meta to train their latest AI systems, but banned for sale to China.

In May, multiple Chinese distributors started selling B200s to suppliers of data centers that serve Chinese AI groups, according to documents reviewed by the FT. This was shortly after the Trump administration moved to prevent sales of the H20—a less-powerful Nvidia chip tailored to comply with Joe Biden-era curbs.

It is legal to receive and sell restricted Nvidia chips in China, as long as relevant border tariffs are paid, according to lawyers familiar with the rules. Entities selling and sending them to China would be violating US regulations, however.

Last week, Nvidia chief Jensen Huang announced that the Trump administration would begin to allow the selling of its China-specific H20 chip once more.

In the three months beforehand, Chinese distributors from Guangdong, Zhejiang, and Anhui provinces sold Nvidia’s B200s, as well as other restricted processors such as the H100 and H200.

According to contracts reviewed by the FT and people with knowledge of the transactions, the total sales during this period are estimated to be more than $1 billion.

Nvidia has long insisted there is “no evidence of any AI chip diversion”. There is no evidence that the company is involved in, or has knowledge of, its restricted products being sold to China.

“Trying to cobble together data centers from smuggled products is a losing proposition, both technically and economically,” Nvidia told the FT. “Data centers require service and support, which we provide only to authorized Nvidia products.”

“The new century of a smart China”

One Anhui-based company, whose name translates to “Gate of the Era,” is one of the largest sellers of B200s, according to documents seen by the FT.

It was founded in February, as speculation mounted that Trump would stop H20 chip sales to China. The company is fully owned by a group with the same name based in Shanghai, registered on the same day, according to company filings.

The chips were sold in ready-built racks, each containing eight B200s as well as other components and software needed to plug straight into a data centre. Such a rack is about the size of a large suitcase and weighs close to 150 kg including packaging.

The current market price ranges between RMB 3 million to RMB 3.5 million ($489,000) per rack, down from more than RMB 4 million in mid-May when they first became available in China in large quantities. The current prices represent about a 50 percent premium from the average selling price of similar products in the US.

Since mid-May, Gate of the Era obtained at least two shipments of a few hundred B200 racks each, according to people with knowledge of the deals. They sold them directly—or indirectly via secondary distributors—to various data centre suppliers and other companies. Gate of the Era and its affiliates are estimated to have sold close to $400mn of such products.

Gate of the Era lists an AI solution provider China Century—or Huajiyuan in Chinese—as its largest shareholder, according to company registration files.

Also headquartered in Shanghai, China Century states on its website it has a lab in Silicon Valley as well as a supply chain centre in Singapore, with the company saying it uses data tools to build “the new century of a smart China.”

China Century claims to have more than 100 business partners and highlights AliCloud, ByteDance’s Huoshan Cloud, as well as Baidu Cloud as “trusted partners” on its website.

AliCloud and Baidu did not respond to requests for comment. Huoshan Cloud’s name was taken off China Century’s website after the FT approached them for comment. Huoshan Cloud said: “It is standard practice for any company to manage the unauthorized use of its logo.

“We have not procured Nvidia’s chips. We do not have any related [Nvidia chip] business,” said China Century, adding that it did “smart city work,”

The FT visited the registered headquarters of Gate of the Era at an office in a government-run industrial park dedicated to cryptography companies. No representative was available. The company had not yet moved into the office since changing its registration to the address in June.

The FT also visited its previous registered address, which was occupied by a real estate investment group that had been there for more than two years and claimed no connection. When reached on the phone, Gate of the Era declined to comment.

According to industry insiders, product specifications and pictures of packaging seen by the FT, many of the B200 racks sold by Gate of the Era, as well as other Chinese distributors, over the past months were originally from Supermicro, a US-based assembler that provides chip solutions to data centers.

There is no suggestion that Supermicro is involved in or has knowledge of its products being smuggled into China. Supermicro said it “complies with all US export control requirements on the sale and export of GPU systems.”

“Export controls will not prevent the most advanced Nvidia products from entering China,” said one Chinese data centre operator. “What it creates is just inefficiency and huge profits for the risk-taking middlemen.”

“It’s like a seafood market”

Some Chinese distributors openly market products such as Supermicro’s B200 racks on social media that show photos of packages with the company’s logo—although it has not been verified if the sales have been completed.

To showcase the “plug-and-use” nature of such racks, some vendors provide testing for buyers, according to those with knowledge of the practice and clips posted online. Transactions tend to happen on the spot, with buyers picking up the products after checking their legitimacy.

On social media, groups are created to match supply and demand from hundreds of traders and data centre suppliers.

Apart from B200, various other restricted Nvidia chips such as H200, H100, and 5090 are being advertised openly on Chinese social media platforms such as Douyin and Xiaohongshu.

Packaging and installation pictures and videos seen by the FT show product logos of companies such as Supermicro, Dell, and Asus—infrastructure providers that assemble Nvidia’s chips into servers.

There is no suggestion that these companies are aware of the social media advertising or their products being sold in China.

Like Supermicro, Dell, and Asus said they maintained rigorous and strict compliance to all laws and regulations, including US export controls, and took action against partners who failed to comply.

“It’s like a seafood market,” said one distributor, “There’s no shortage.”

Racks for sale—with more smuggled stock to come

The B200 is in high demand given its performance, value, and relatively easy maintenance compared with the more complex Grace Blackwell series, according to industry insiders.

The GB200 AI rack, containing Nvidia’s most high-end products, also appear to be available in China despite US export controls.

One distributor claimed it had sold 10 racks of GB200 at close to RMB 40 million ($5.6 million) each. The FT could not independently verify this claim, while marketing information about GB200 from various distributors’ accounts on social media shows consistent pricing and stock status as “available for pick up onshore.”

Some Chinese distributors have even started advertising for their future stock of B300s, Nvidia’s upgrade from the B200 expected to enter mass production in the fourth quarter of this year.

US export controls have had some effect on the black market.

Given the nature of such products, leading Chinese AI players with global operations are not able to order them in a legally compliant way, install them in their own data centers, or receive Nvidia’s customer support.

This has led to third-party data centre operators becoming key buyers who then provide computing services. Other clients include smaller companies in tech, finance, and health care that do not have strong compliance requirements, as well as Chinese companies on the so-called US entity list that are not allowed to buy any Nvidia chips legally.

However, the scale of these projects is much smaller compared with mega clusters of data centers being built by tech giants around the world.

With H20 export controls having been lifted, many Chinese tech companies are expected to resume purchasing the compliant chips in large sums even though its performance is generations behind the still restricted products such as B200, according to people familiar with their plans.

Black market sales for B200s and other restricted Nvidia chips dropped noticeably after the relaxation of the H20 ban, according to multiple distributors.

“People are weighing their options now H20 is available again,” said one distributor. “But there will always be demand for the most cutting-edge stuff.”

The Southeast Asia stop off

Industry experts said that Southeast Asian countries have become markets where Chinese groups obtained restricted chips.

The US Department of Commerce is discussing adding more export controls on advanced AI products to countries such as Thailand as soon as September, according to two people familiar with the matter. This rule is mainly targeting Chinese intermediaries used to obtain advanced AI chips via these countries.

The US commerce department declined to comment. The Thai government did not respond to a request for comment.

Earlier this month, Malaysia introduced stricter export controls targeting advanced AI chip shipments from the country to other destinations, especially China.

The potential tightening of export controls on Southeast Asian countries has also contributed to buyers rushing to place orders before such rules take effect, according to people with knowledge of the matter.

Even if these avenues to obtain AI chips are closed, Chinese industry insiders said new shipping routes would be established. Supplies have already started arriving via European countries not on the restricted list.

“History has proven many times before that given the huge profit, arbitrators will always find a way,” said one Chinese distributor.

Additional reporting by Michael Acton, Demetri Sevastopulo, and Anantha Lakshmi.

Nvidia AI chips worth $1B smuggled to China after Trump export controls Read More »

OpenAI and partners are building a massive AI data center in Texas

AI, AI chips, AI GPU, AI infrastructure, Biz & IT, datacenters, Larry Ellison, machine learning, microsoft, NVIDIA, openai, oracle, sam altman, SoftBank, Stargate / Paul Patrick / July 24, 2025

Stargate moves forward despite early skepticism

When OpenAI announced Stargate in January, critics questioned whether the company could deliver on its ambitious $500 billion funding promise. Trump ally and frequent Altman foe Elon Musk wrote on X that “They don’t actually have the money,” claiming that “SoftBank has well under $10B secured.”

Tech writer and frequent OpenAI critic Ed Zitron raised concerns about OpenAI’s financial position, noting the company’s $5 billion in losses in 2024. “This company loses $5bn+ a year! So what, they raise $19bn for Stargate, then what, another $10bn just to be able to survive?” Zitron wrote on Bluesky at the time.

Six months later, OpenAI’s Abilene data center has moved from construction to partial operation. Oracle began delivering Nvidia GB200 racks to the facility last month, and OpenAI reports it has started running early training and inference workloads to support what it calls “next-generation frontier research.”

Despite the White House announcement with President Trump in January, the Stargate concept dates back to March 2024, when Microsoft and OpenAI partnered on a $100 billion supercomputer as part of a five-phase plan. Over time, the plan evolved into its current form as a partnership with Oracle, SoftBank, and CoreWeave.

“Stargate is an ambitious undertaking designed to meet the historic opportunity in front of us,” writes OpenAI in the press release announcing the latest deal. “That opportunity is now coming to life through strong support from partners, governments, and investors worldwide—including important leadership from the White House, which has recognized the critical role AI infrastructure will play in driving innovation, economic growth, and national competitiveness.”

OpenAI and partners are building a massive AI data center in Texas Read More »

Trump can’t keep China from getting AI chips, TSMC suggests

AI chips, Artificial Intelligence, china, Chips Act, Donald Trump, Huawei, NVIDIA, Policy, semiconductors, Taiwan Semiconductor Manufacturing Company, tariffs, TSMC, us-china trade war / Kris Guyer / April 22, 2025

“Despite TSMC’s best efforts to comply with all relevant export control and sanctions laws and regulations, there is no assurance that its business activities will not be found incompliant with export control laws and regulations,” TSMC said.

Further, “if TSMC or TSMC’s business partners fail to obtain appropriate import, export or re-export licenses or permits or are found to have violated applicable export control or sanctions laws, TSMC may also be adversely affected, through reputational harm as well as other negative consequences, including government investigations and penalties resulting from relevant legal proceedings,” TSMC warned.

Trump’s tariffs may end TSMC’s “tariff-proof” era

TSMC is thriving despite years of tariffs and export controls, its report said, with at least one analyst suggesting that, so far, the company appears “somewhat tariff-proof.” However, all of that could be changing fast, as “US President Donald Trump announced in 2025 an intention to impose more expansive tariffs on imports into the United States,” TSMC said.

“Any tariffs imposed on imports of semiconductors and products incorporating chips into the United States may result in increased costs for purchasing such products, which may, in turn, lead to decreased demand for TSMC’s products and services and adversely affect its business and future growth,” TSMC said.

And if TSMC’s business is rattled by escalations in the US-China trade war, TSMC warned, that risks disrupting the entire global semiconductor supply chain.

Trump’s semiconductor tariff plans remain uncertain. About a week ago, Trump claimed the rates would be unveiled “over the next week,” Reuters reported, which means they could be announced any day now.

Trump can’t keep China from getting AI chips, TSMC suggests Read More »

TSMC to invest $100B as Trump demands more US-made chips, report says

AI chips, Chips Act, chips fabs, Donald Trump, Policy, semiconductors, Taiwan Semiconductor Manufacturing Company, tariffs, TSMC / Tim Belzer / March 4, 2025

Currently, TSMC only builds its most advanced chips in Taiwan. But when the most advanced US fabs are operational, they’ll be prepared to manufacture “tens of millions of leading-edge chips” to “power products like 5G/6G smartphones, autonomous vehicles, and AI datacenter servers,” the Commerce Department said in 2024.

TSMC has not confirmed the WSJ’s report but provided a statement: “We’re pleased to have an opportunity to meet with the President and look forward to discussing our shared vision for innovation and growth in the semiconductor industry, as well as exploring ways to bolster the technology sector along with our customers.”

Trump threat of semiconductor tariffs still looms

Advanced chips are regarded as critical for AI innovation, which Trump has prioritized, as well as for national security.

Without a steady supply, the US risks substantial technological and economic losses as well as potential weakening of its military.

To avert that, Trump campaigned on imposing tariffs that he claimed would drive more semiconductor manufacturing into the US, while criticizing the CHIPS Act for costing the US billions. Following through on that promise, in February, he threatened a “25 percent or more tariff” on all semiconductor imports, the WSJ reported. According to CNBC, Trump suggested those tariffs could be in effect by April 2.

“We have to have chips made in this country,” Trump said last month. “Right now, everything is made in Taiwan, practically, almost all of it, a little bit in South Korea, but everything—almost all of it is made in Taiwan. And we want it to be made—we want those companies to come to our country, in all due respect.”

While it’s unclear if Trump plans to overtly kill the CHIPS Act, his government funding cuts could trigger a future where the CHIPS Act dies with no workers left to certify that companies meet requirements for ongoing award disbursements, a semiconductor industry consultant group, Semiconductor Advisors, warned in a statement last month.

“If I were running a chip company, I would not count on CHIPS Act funding, even if I had a signed contract,” SA’s statement said.

TSMC to invest $100B as Trump demands more US-made chips, report says Read More »

Trump should block Biden’s AI “gift” to China, Microsoft argues

AI chips, china, Donald Trump, Export controls, microsoft, national security, Policy / Rejus Almole / February 28, 2025

“Countries including Brazil, India, Israel, and the UAE are eminently capable of ramping up investments aimed at securing new ways to access increased computing capacity,” the Brookings Institute said. “Preventing companies in middle-tier countries from relying on the US to supply computing chips is a surefire way to push them into building non-US alliances that include stronger technology ties with China.”

The rule could also complicate the global AI landscape in ways the US may not anticipate, the Center for Strategic and International Studies (CSIS), a bipartisan, nonprofit policy research organization, forecast last week. It could “breed resentment, not cooperation” in tier-two countries that will likely “bristle at the fact that their AI ambitions depend on Washington’s goodwill and that they are being deliberately kept a generation behind the frontier,” CSIS wrote. And it could drive more open source AI like DeepSeek to be key to development in tier-two nations, perhaps further endangering US global leadership in AI, CSIS suggested.

“Ironically, the AI Diffusion Framework, meant to lock in American advantage, may instead midwife the very outcome it sought to prevent—an alternate AI stack and increased open-source development where China, as its most prolific contributor, emerges as the de facto leader,” CSIS reported.

China wooing countries targeted by rule, Microsoft says

But according to Smith, some parts of the rule should be preserved, like datacenter restrictions, including “qualitative provisions” that “ensure that AI technology components are deployed in certified, secure, and trusted datacenters.” That part of the rule, Smith suggested, “helps reduce the risk of chip diversion to China.”

And other parts of the rule can be strengthened, Smith wrote, such as ensuring the Commerce Department has resources to enforce it and “expedite approvals” for any countries in the middle tier who may appeal to either move into the top tier or limit tier-two restrictions.

Trump should block Biden’s AI “gift” to China, Microsoft argues Read More »

OpenAI’s secret weapon against Nvidia dependence takes shape

AI, AI chips, Biz & IT, Broadcom, chatgpt, chatgtp, china, GPUs, machine learning, NVIDIA, openai, Reuters, Richard Ho, sam altman, Taiwan, TSMC, US government / Mike M. / February 10, 2025

OpenAI is entering the final stages of designing its long-rumored AI processor with the aim of decreasing the company’s dependence on Nvidia hardware, according to a Reuters report released Monday. The ChatGPT creator plans to send its chip designs to Taiwan Semiconductor Manufacturing Co. (TSMC) for fabrication within the next few months, but the chip has not yet been formally announced.

The OpenAI chip’s full capabilities, technical details, and exact timeline are still unknown, but the company reportedly intends to iterate on the design and improve it over time, giving it leverage in negotiations with chip suppliers—and potentially granting the company future independence with a chip design it controls outright.

In the past, we’ve seen other tech companies, such as Microsoft, Amazon, Google, and Meta, create their own AI acceleration chips for reasons that range from cost reduction to relieving shortages of AI chips supplied by Nvidia, which enjoys a near-market monopoly on high-powered GPUs (such as the Blackwell series) for data center use.

In October 2023, we covered a report about OpenAI’s intention to create its own AI accelerator chips for similar reasons, so OpenAI’s custom chip project has been in the works for some time. In early 2024, OpenAI CEO Sam Altman also began spending considerable time traveling around the world trying to raise up to a reported $7 trillion to increase world chip fabrication capacity.

OpenAI’s secret weapon against Nvidia dependence takes shape Read More »

US blocks China from foreign exports with even a single US-made chip

AI chips, applied materials, Chips Act, kla, lam research, Policy, samsung, semiconductor, us-china trade war / Rejus Almole / December 2, 2024

But while Commerce Secretary Gina Raimondo said that these new curbs would help prevent “China from advancing its domestic semiconductor manufacturing system” to modernize its military, analysts and “several US officials” told The Post that they pack “far less punch” than the prior two rounds of export controls.

Analysts told The Wall Street Journal that the US took too long to launch the controls, which were composed around June. As industry insiders weighed in on the restrictions, word got out about the US plans to expand controls. In the months since, analysts said, China had plenty of time to stockpile the now-restricted tech. Applied Materials, for example, saw an eye-popping 86 percent spike in net revenue from products shipped to China “in the nine months ending July 28,” the WSJ reported.

Because of this and other alleged flaws, it’s unclear how effectively Biden’s final attempts to block China from accessing the latest US technologies will work.

Beyond concerns that China had time to stockpile tech it anticipated would be restricted, Gregory Allen, the director at the Wadhwani AI Center at the Center for Strategic and International Studies, told the WSJ that these latest controls “left loopholes that Huawei and Chinese companies could exploit.”

Loopholes include failing to blacklist companies that Huawei regularly uses—with allies and American companies allegedly lobbying to exempt factories or fabs they like, such as ChangXin Memory Technologies Inc., “one of China’s largest memory chipmakers,” The Post noted. They also include failing to restrict older versions of the HBM chips and various chipmaking equipment that China may still be able to easily access, Allen said.

“These controls are weaker than what the United States should have done,” Allen told The Post. “You can make a halfway logical argument that says, ‘Sell everything to China.’ Then you can make a reasonable argument, ‘Sell very little to China.’ But the worst thing you can do is to dramatically signal your intention to cut off China’s access to tech but then have so many loopholes and such bungled implementation that you incur almost all of the costs of the policy with only a fraction of the benefits.”

US blocks China from foreign exports with even a single US-made chip Read More »

Workers demand more transparency after Intel secures $8B CHIPS funding

AI chips, Chips Act, chips fabs, chips plants, Donald Trump, intel, Policy, semiconductor / Mike M. / November 26, 2024

Intel awarded nearly $8B to “supercharge” US semiconductor innovation.

An aerial view from February 2024 shows construction progress at Intel’s Ohio One campus of nearly 1,000 acres in Licking County, Ohio. Credit: Intel Corporation

On Tuesday, the Biden-Harris administration finalized a CHIPS award of up to $7.865 billion to help fund the expansion of Intel’s commercial fabs in the US. By the end of the decade, these fabs are intended to decrease reliance on foreign adversaries and fill substantial gaps in America’s domestic semiconductor supply chain.

Initially, Intel was awarded $8.5 billion, but it was decreased after Intel won a $3 billion subsidy from the Pentagon to expand Department of Defense semiconductor manufacturing. In a press release, Secretary of Commerce Gina Raimondo boasted that the substantial award would set up “Intel to drive one of the most significant semiconductor manufacturing expansions in US history” and “supercharge American innovation” while making the US “more secure.”

For Intel, the CHIPS funding supports an expected investment of nearly $90 billion by 2030 to expand projects in Arizona, New Mexico, Ohio, and Oregon. Approximately 10,000 manufacturing jobs and 20,000 construction jobs will be created “across all four states,” the Commerce Department’s press release said. Additionally, Intel estimated that the funding will create “more than 50,000 indirect jobs with suppliers and supporting industries.”

According to the National Institute of Standards and Technology (NIST), which oversees CHIPS funding for manufacturing and research and development initiatives, the “funding will spur investment in leading-edge logic chip manufacturing, packaging, and R&D facilities.”

The sprawling effort includes the construction of two new fabs in Chandler, Arizona, the modernization of two fabs in Rio Rancho, New Mexico, building a new leading-edge logic fab in New Albany, Ohio, and creating a “premier hub of leading-edge research and development” in Hillsboro, Oregon. By the end, Intel expects to operate America’s largest advanced packaging facility in New Mexico and “one of only three locations in the world where leading-edge process technology is developed” in Oregon, NIST said.

Who’s enforcing worker safety commitments?

To succeed, Intel will need to build a talented workforce, so $65 million has been set aside to fund those efforts. The majority, $56 million, will “help train students and faculty at all education levels,” Intel said. Another $5 million will “help increase childcare availability near Intel’s facilities,” and the final $4 million will support efforts to recruit women and “economically disadvantaged individuals” as construction workers, Intel said.

Recruitment could be challenging if worker safety concerns are continually raised, though. Chips Communities United (CCU), a coalition of “labor, environmental, social justice, civil rights, and community organizations representing millions of workers and community members nationwide,” has been monitoring worker concerns at facilities receiving CHIPS funding. While the coalition fully supports Intel’s US expansion, they recently requested a full environmental impact statement at one of Intel’s Arizona fabs, detailing potential environmental and worker hazards, as well as mitigation plans.

As of August, CCU said that Ocotillo workers and communities had been given “insufficient detail on the use, storage, and release of hazardous substances, as well as other environmental impacts, to conclude that there are no significant environmental impacts.”

Workers have a bunch of questions. But perhaps most urgently, they need more information on how environmental safety commitments will be enforced, CCU suggested, because no one wants to work in constant fear of chemical exposure. Especially when Intel’s facilities in Oregon were revealed last year to have “accidentally turned off its air pollution control equipment for two months and underreported its CO₂ emissions.”

NIST noted that Intel is required to protect workers to receive CHIPS funding and has promised to meet regularly with workers and managers at each project facility to discuss worker safety concerns.

Intel could not immediately be reached for comment on whether it’s currently in discussions with workers impacted by CCU’s recent claims.

Weighing in on the Intel Community Impact Report that NIST released today, CCU applauded Intel’s commitments to bring workers to the table, adopt the “most protective health and safety standards for chemical exposure,” “segregate PFAS-containing waste for treatment and disposal,” and “make environmental compliance public when it comes to energy and water use,” CCU coalition director Judith Barish told Ars. But the enforceability of the promised workplace safety conditions remains a concern at Intel’s facilities.

“Protective workplace health and safety regulation” has “historically been missing in semiconductor production,” Barish told Ars. And it’s a big problem Intel’s current plan is to regulate the management of toxic chemicals following guidelines developed by industry—not government.

“Unlike government regulations, this standard is not easily available for public inspection since it is proprietary, copyrighted, and can only be inspected by purchasing it,” Barish told Ars. “Allowing a regulated entity to write the regulations that will be applied to it violates basic principles of good government.”

While segregating PFAS-containing waste sounds good, Barish said that workers need more transparency to understand how it “will be separated, stored, and treated and what the environmental impacts will be for nearby communities.”

It’s also unclear to workers what might happen if Intel fails to follow through on its commitments. The Commerce Department has emphasized that Intel’s funding will be disbursed “based on Intel’s completion of project milestones,” but workers “aren’t clear on the penalties or clawbacks the Commerce Dept. would impose if Intel failed to meet workforce, health and safety, or environmental milestones and metrics,” Barish said.

Intel only approved unionized workers at one site

For top talent to be attracted to Intel’s facilities, establishing the most protective safety protocols will be critical. But just as critical for workers—especially “economically disadvantaged” workers Intel is targeting for construction jobs—will be worker benefits.

Barish noted that Intel has only committed to employing unionized construction workers at one of four sites. The company may struggle to recruit workers, Barish suggested, without being clear about their rights to “join a union free from intimidation, captive audience meetings, exposure to anti-union consultants, threats of retaliation, and other obstacles to achieve bargaining.”

CCU plans to continue monitoring concerns at Intel’s fabs and others receiving CHIPS funding as the presidential administration potentially introduces CHIPS Act changes next year.

On the campaign trail, President-elect Donald Trump attacked the CHIPS Act, saying he was “not thrilled” with the price tag, CNBC reported. However, analysts told CNBC that any changes under Trump would likely be smaller rather than something drastic like repealing the law.

The Commerce Department continues to tout the CHIPS Act as a firmly bipartisan initiative. Intel CEO Pat Gelsinger, whose company’s large investment depends on bipartisan support for the CHIPS Act continuing for years to come, echoed that sentiment after the award was finalized.

“With Intel 3 already in high-volume production and Intel 18A set to follow next year, leading-edge semiconductors are once again being made on American soil,” Gelsinger said. “Strong bipartisan support for restoring American technology and manufacturing leadership is driving historic investments that are critical to the country’s long-term economic growth and national security. Intel is deeply committed to advancing these shared priorities as we further expand our US operations over the next several years.”

Ashley is a senior policy reporter for Ars Technica, dedicated to tracking social impacts of emerging policies and new technologies. She is a Chicago-based journalist with 20 years of experience.

Workers demand more transparency after Intel secures $8B CHIPS funding Read More »

US suspects TSMC helped Huawei skirt export controls, report says

5G, AI chips, Apple, Artificial Intelligence, china tariffs, Chips Act, Export controls, Huawei, NVIDIA, Policy, qualcomm, semiconductors, Taiwan Semiconductor Manufacturing Company, TSMC / Kris Guyer / October 18, 2024

In April, TSMC was provided with $6.6 billion in direct CHIPS Act funding to “support TSMC’s investment of more than $65 billion in three greenfield leading-edge fabs in Phoenix, Arizona, which will manufacture the world’s most advanced semiconductors,” the Department of Commerce said.

These investments are key to the Biden-Harris administration’s mission of strengthening “economic and national security by providing a reliable domestic supply of the chips that will underpin the future economy, powering the AI boom and other fast-growing industries like consumer electronics, automotive, Internet of Things, and high-performance computing,” the department noted. And in particular, the funding will help America “maintain our competitive edge” in artificial intelligence, the department said.

It likely wouldn’t make sense to prop TSMC up to help the US “onshore the critical hardware manufacturing capabilities that underpin AI’s deep language learning algorithms and inferencing techniques,” to then limit access to US-made tech. TSMC’s Arizona fabs are supposed to support companies like Apple, Nvidia, and Qualcomm and enable them to “compete effectively,” the Department of Commerce said.

Currently, it’s unclear where the US probe into TSMC will go or whether a damaging finding could potentially impact TSMC’s CHIPS funding.

Last fall, the Department of Commerce published a final rule, though, designed to “prevent CHIPS funds from being used to directly or indirectly benefit foreign countries of concern,” such as China.

If the US suspected that TSMC was aiding Huawei’s AI chip manufacturing, the company could be perceived as avoiding CHIPS guardrails prohibiting TSMC from “knowingly engaging in any joint research or technology licensing effort with a foreign entity of concern that relates to a technology or product that raises national security concerns.”

Violating this “technology clawback” provision of the final rule risks “the full amount” of CHIPS Act funding being “recovered” by the Department of Commerce. That outcome seems unlikely, though, given that TSMC has been awarded more funding than any other recipient apart from Intel.

The Department of Commerce declined Ars’ request to comment on whether TSMC’s CHIPS Act funding could be impacted by their reported probe.

US suspects TSMC helped Huawei skirt export controls, report says Read More »

DOJ subpoenas Nvidia in deepening AI antitrust probe, report says

AI chips, antitrust, Antitrust law, Artificial Intelligence, competition law, GPUs, Jensen Huang, monopoly, NVIDIA, Policy, run:ai / Mike M. / September 4, 2024

The Department of Justice is reportedly deepening its probe into Nvidia. Officials have moved on from merely questioning competitors to subpoenaing Nvidia and other tech companies for evidence that could substantiate allegations that Nvidia is abusing its “dominant position in AI computing,” Bloomberg reported.

When news of the DOJ’s probe into the trillion-dollar company was first reported in June, Fast Company reported that scrutiny was intensifying merely because Nvidia was estimated to control “as much as 90 percent of the market for chips” capable of powering AI models. Experts told Fast Company that the DOJ probe might even be good for Nvidia’s business, noting that the market barely moved when the probe was first announced.

But the market’s confidence seemed to be shaken a little more on Tuesday, when Nvidia lost a “record-setting $279 billion” in market value following Bloomberg’s report. Nvidia’s losses became “the biggest single-day market-cap decline on record,” TheStreet reported.

People close to the DOJ’s investigation told Bloomberg that the DOJ’s “legally binding requests” require competitors “to provide information” on Nvidia’s suspected anticompetitive behaviors as a “dominant provider of AI processors.”

One concern is that Nvidia may be giving “preferential supply and pricing to customers who use its technology exclusively or buy its complete systems,” sources told Bloomberg. The DOJ is also reportedly probing Nvidia’s acquisition of RunAI—suspecting the deal may lock RunAI customers into using Nvidia chips.

Bloomberg’s report builds on a report last month from The Information that said that Advanced Micro Devices Inc. (AMD) and other Nvidia rivals were questioned by the DOJ—as well as third parties who could shed light on whether Nvidia potentially abused its market dominance in AI chips to pressure customers into buying more products.

According to Bloomberg’s sources, the DOJ is worried that “Nvidia is making it harder to switch to other suppliers and penalizes buyers that don’t exclusively use its artificial intelligence chips.”

In a statement to Bloomberg, Nvidia insisted that “Nvidia wins on merit, as reflected in our benchmark results and value to customers, who can choose whatever solution is best for them.” Additionally, Bloomberg noted that following a chip shortage in 2022, Nvidia CEO Jensen Huang has said that his company strives to prevent stockpiling of Nvidia’s coveted AI chips by prioritizing customers “who can make use of his products in ready-to-go data centers.”

Potential threats to Nvidia’s dominance

Despite the slump in shares, Nvidia’s market dominance seems unlikely to wane any time soon after its stock more than doubled this year. In an SEC filing this year, Nvidia bragged that its “accelerated computing ecosystem is bringing AI to every enterprise” with an “ecosystem” spanning “nearly 5 million developers and 40,000 companies.” Nvidia specifically highlighted that “more than 1,600 generative AI companies are building on Nvidia,” and according to Bloomberg, Nvidia will close out 2024 with more profits than the total sales of its closest competitor, AMD.

After the DOJ’s most recent big win, which successfully proved that Google has a monopoly on search, the DOJ appears intent on getting ahead of any tech companies’ ambitions to seize monopoly power and essentially become the Google of the AI industry. In June, DOJ antitrust chief Jonathan Kanter confirmed to the Financial Times that the DOJ is examining “monopoly choke points and the competitive landscape” in AI beyond just scrutinizing Nvidia.

According to Kanter, the DOJ is scrutinizing all aspects of the AI industry—”everything from computing power and the data used to train large language models, to cloud service providers, engineering talent and access to essential hardware such as graphics processing unit chips.” But in particular, the DOJ appears concerned that GPUs like Nvidia’s advanced AI chips remain a “scarce resource.” Kanter told the Financial Times that an “intervention” in “real time” to block a potential monopoly could be “the most meaningful intervention” and the least “invasive” as the AI industry grows.

DOJ subpoenas Nvidia in deepening AI antitrust probe, report says Read More »

AI’s future in grave danger from Nvidia’s chokehold on chips, groups warn

AI chips, ai industry, Antitrust law, Artificial Intelligence, department of justice, intel, Jensen Huang, monopoly, NVIDIA, Policy / Kris Guyer / August 1, 2024

Controlling “the world’s computing destiny” —

Anti-monopoly groups want DOJ to probe Nvidia’s AI chip bundling, alleged price-fixing.

Ashley Belanger – Aug 1, 2024 6: 14 pm UTC

Sen. Elizabeth Warren (D-Mass.) has joined progressive groups—including Demand Progress, Open Markets Institute, and the Tech Oversight Project—pressuring the US Department of Justice to investigate Nvidia’s dominance in the AI chip market due to alleged antitrust concerns, Reuters reported.

In a letter to the DOJ’s chief antitrust enforcer, Jonathan Kanter, groups demanding more Big Tech oversight raised alarms that Nvidia’s top rivals apparently “are struggling to gain traction” because “Nvidia’s near-absolute dominance of the market is difficult to counter” and “funders are wary of backing its rivals.”

Nvidia is currently “the world’s most valuable public company,” their letter said, worth more than $3 trillion after taking near-total control of the high-performance AI chip market. Particularly “astonishing,” the letter said, was Nvidia’s dominance in the market for GPU accelerator chips, which are at the heart of today’s leading AI. Groups urged Kanter to probe Nvidia’s business practices to ensure that rivals aren’t permanently blocked from competing.

According to the advocacy groups that strongly oppose Big Tech monopolies, Nvidia “now holds an 80 percent overall global market share in GPU chips and a 98 percent share in the data center market.” This “puts it in a position to crowd out competitors and set global pricing and the terms of trade,” the letter warned.

Earlier this year, inside sources reported that the DOJ and the Federal Trade Commission reached a deal where the DOJ would probe Nvidia’s alleged anti-competitive behavior in the booming AI industry, and the FTC would probe OpenAI and Microsoft. But there has been no official Nvidia probe announced, prompting progressive groups to push harder for the DOJ to recognize what they view as a “dire danger to the open market” that “well deserves DOJ scrutiny.”

Ultimately, the advocacy groups told Kanter that they fear Nvidia wielding “control over the world’s computing destiny,” noting that Nvidia’s cloud computing data centers don’t just power “Big Tech’s consumer products” but also “underpin every aspect of contemporary society, including the financial system, logistics, healthcare, and defense.”

They claimed that Nvidia is “leveraging” its “scarce chips” to force customers to buy its “chips, networking, and programming software as a package.” Such bundling and “price-fixing,” their letter warned, appear to be “the same kinds of anti-competitive tactics that the courts, in response to actions brought by the Department of Justice against other companies, have found to be illegal” and could perhaps “stifle innovation.”

Although data from TechInsights suggested that Nvidia’s chip shortage and cost actually helped companies like AMD and Intel sell chips in 2023, both Nvidia rivals reported losses in market share earlier this year, Yahoo Finance reported.

Perhaps most closely monitoring Nvidia’s dominance, France antitrust authorities launched an investigation into Nvidia last month over antitrust concerns, the letter said, “making it the first enforcer to act against the computer chip maker,” Reuters reported.

Since then, the European Union and the United Kingdom, as well as the US, have heightened scrutiny, but their seeming lag to follow through with an official investigation may only embolden Nvidia, as the company allegedly “believes its market behavior is above the law,” the progressive groups wrote. Suspicious behavior includes allegations that “Nvidia has continued to sell chips to Chinese customers and provide them computing access” despite a “Department of Commerce ban on trading with Chinese companies due to national security and human rights concerns.”

“Its chips have been confirmed to be reaching blacklisted Chinese entities,” their letter warned, citing a Wall Street Journal report.

Nvidia’s dominance apparently impacts everyone involved with AI. According to the letter, Nvidia seemingly “determining who receives inventory from a limited supply, setting premium pricing, and contractually blocking customers from doing business with competitors” is “alarming” the entire AI industry. That includes “both small companies (who find their supply choked off) and the Big Tech AI giants.”

Kanter will likely be receptive to the letter. In June, Fast Company reported that Kanter told an audience at an AI conference that there are “structures and trends in AI that should give us pause.” He further suggested that any technology that “relies on massive amounts of data and computing power” can “give already dominant firms a substantial advantage,” according to Fast Company’s summary of his remarks.

AI’s future in grave danger from Nvidia’s chokehold on chips, groups warn Read More »

Intel’s “Gaudi 3” AI accelerator chip may give Nvidia’s H100 a run for its money

AI, AI chips, B200, Biz & IT, Blackwell, chatgpt, chatgtp, Gaudi 3, H100, H200, intel, Intel Gaudi, machine learning, NVIDIA, Nvidia Blackwell, openai / Paul Patrick / April 11, 2024

Adventures in Matrix Multiplication —

Intel claims 50% more speed when running AI language models vs. the market leader.

Benj Edwards – Apr 11, 2024 8: 56 pm UTC

Enlarge / An Intel handout photo of the Gaudi 3 AI accelerator.

On Tuesday, Intel revealed a new AI accelerator chip called Gaudi 3 at its Vision 2024 event in Phoenix. With strong claimed performance while running large language models (like those that power ChatGPT), the company has positioned Gaudi 3 as an alternative to Nvidia’s H100, a popular data center GPU that has been subject to shortages, though apparently that is easing somewhat.

Compared to Nvidia’s H100 chip, Intel projects a 50 percent faster training time on Gaudi 3 for both OpenAI’s GPT-3 175B LLM and the 7-billion parameter version of Meta’s Llama 2. In terms of inference (running the trained model to get outputs), Intel claims that its new AI chip delivers 50 percent faster performance than H100 for Llama 2 and Falcon 180B, which are both relatively popular open-weights models.

Intel is targeting the H100 because of its high market share, but the chip isn’t Nvidia’s most powerful AI accelerator chip in the pipeline. Announcements of the H200 and the Blackwell B200 have since surpassed the H100 on paper, but neither of those chips is out yet (the H200 is expected in the second quarter of 2024—basically any day now).

Meanwhile, the aforementioned H100 supply issues have been a major headache for tech companies and AI researchers who have to fight for access to any chips that can train AI models. This has led several tech companies like Microsoft, Meta, and OpenAI (rumor has it) to seek their own AI-accelerator chip designs, although that custom silicon is typically manufactured by either Intel or TSMC. Google has its own line of tensor processing units (TPUs) that it has been using internally since 2015.

Given those issues, Intel’s Gaudi 3 may be a potentially attractive alternative to the H100 if Intel can hit an ideal price (which Intel has not provided, but an H100 reportedly costs around $30,000–$40,000) and maintain adequate production. AMD also manufactures a competitive range of AI chips, such as the AMD Instinct MI300 Series, that sell for around $10,000–$15,000.

Gaudi 3 performance

Enlarge / An Intel handout featuring specifications of the Gaudi 3 AI accelerator.

Intel says the new chip builds upon the architecture of its predecessor, Gaudi 2, by featuring two identical silicon dies connected by a high-bandwidth connection. Each die contains a central cache memory of 48 megabytes, surrounded by four matrix multiplication engines and 32 programmable tensor processor cores, bringing the total cores to 64.

The chipmaking giant claims that Gaudi 3 delivers double the AI compute performance of Gaudi 2 using 8-bit floating-point infrastructure, which has become crucial for training transformer models. The chip also offers a fourfold boost for computations using the BFloat 16-number format. Gaudi 3 also features 128GB of the less expensive HBMe2 memory capacity (which may contribute to price competitiveness) and features 3.7TB of memory bandwidth.

Since data centers are well-known to be power hungry, Intel emphasizes the power efficiency of Gaudi 3, claiming 40 percent greater inference power-efficiency across Llama 7B and 70B parameters, and Falcon 180B parameter models compared to Nvidia’s H100. Eitan Medina, chief operating officer of Intel’s Habana Labs, attributes this advantage to Gaudi’s large-matrix math engines, which he claims require significantly less memory bandwidth compared to other architectures.

Gaudi vs. Blackwell

Last month, we covered the splashy launch of Nvidia’s Blackwell architecture, including the B200 GPU, which Nvidia claims will be the world’s most powerful AI chip. It seems natural, then, to compare what we know about Nvidia’s highest-performing AI chip to the best of what Intel can currently produce.

For starters, Gaudi 3 is being manufactured using TSMC’s N5 process technology, according to IEEE Spectrum, narrowing the gap between Intel and Nvidia in terms of semiconductor fabrication technology. The upcoming Nvidia Blackwell chip will use a custom N4P process, which reportedly offers modest performance and efficiency improvements over N5.

Gaudi 3’s use of HBM2e memory (as we mentioned above) is notable compared to the more expensive HBM3 or HBM3e used in competing chips, offering a balance of performance and cost-efficiency. This choice seems to emphasize Intel’s strategy to compete not only on performance but also on price.

As far as raw performance comparisons between Gaudi 3 and the B200, that can’t be known until the chips have been released and benchmarked by a third party.

As the race to power the tech industry’s thirst for AI computation heats up, IEEE Spectrum notes that the next generation of Intel’s Gaudi chip, code-named Falcon Shores, remains a point of interest. It also remains to be seen whether Intel will continue to rely on TSMC’s technology or leverage its own foundry business and upcoming nanosheet transistor technology to gain a competitive edge in the AI accelerator market.

Intel’s “Gaudi 3” AI accelerator chip may give Nvidia’s H100 a run for its money Read More »