Japan Investment Corp (JIC), a Japanese state-backed fund, plans to sell chipmaking materials maker JSR to Mitsui Chemicals in a deal that could exceed 1.3 trillion yen ($8.2 billion), after buying it for some 900 billion yen two years ago, taking advantage of higher chip-firm valuations on the artificial intelligence investment boom, according to two people familiar with the matter who spoke to Reuters, in a text republished by DealStreetAsia. Founded in 1957, JSR is a leading manufacturer of photoresists used to transfer circuit patterns onto semiconductor wafers, and Nikkei Asia notes that the sale would be the biggest deal in the history of Japan’s chemical sector.
AMD recorded a $1.07 billion non-cash charge, or 66 cents a share, in the third quarter ended September 26, to account for warrants granted to OpenAI and Meta. The information comes from Reuters, in a text republished by StreetInsider. The company forecast faster-than-expected 2027 growth on surging data-center sales, and its shares rose 8% in extended trading. The warrants are part of a wave of equity incentives used by chipmakers to secure long-term purchases from a small group of AI customers: Cerebras granted OpenAI warrants against a $20 billion commitment, and Marvell issued warrants to Alphabet’s Google that give Google the right to buy a potential $12.2 billion stake.
Samsung Electronics will use its second semiconductor plant in Taylor, Texas, to produce base dies for high-bandwidth memory (HBM) chips for AMD and Nvidia, according to South Korea’s Maeil Business Newspaper, which cited industry sources, in a Reuters report published by WHTC. Construction is due to start formally this month, with the plant having secured orders for its entire output, most of it for AMD. The facility will use 2-nanometer technology and cover about 2.7 million square feet, roughly 1.5 times the size of the first plant, which has a contract with Tesla. Equipment is expected to be installed from February 2029, with mass production in 2030.
China rejected allegations of a Chinese espionage operation targeting Nvidia’s chief executive, Jensen Huang, describing the claims as groundless speculation, the BBC reported. Foreign ministry spokesman Guo Jiakun said at a regular media briefing in Beijing that the report was based on groundless speculation and was aimed purely at smearing and slandering China, the South China Morning Post reported: “The relevant report is based on groundless speculation and [is] purely aimed at smearing and slandering China.” Nvidia told the newspaper that “Megaspeed has been no different than many other regional cloud customers” and that “Jensen’s interactions with Ms. Huang are not unusual or unexpected”.
Nvidia pushed back on the report about American intelligence agencies surveilling a person close to CEO Jensen Huang as a possible Chinese intelligence asset, saying Singapore-based Megaspeed is just another regional cloud customer, that Huang's contacts with its executives are routine and that the chip-diversion allegations against the company were investigated and debunked last year, a company spokesperson said, as cited by CNBC.
Malaysia’s customs department seized 30 servers fitted with advanced artificial intelligence (AI) chips in a suspected export-control violation: the shipment, worth about 50 million ringgit (US$ 12.2 million), was intercepted in the free trade zone of the country’s main airport on Oct. 15 after being falsely declared as electrical components for re-export to another Asian country without the required permit. Malaysia introduced permit requirements for the export, transit and transshipment of AI chips of US origin under pressure from the US to keep sensitive technology from being rerouted to China through third countries.
Customs authorities found 30 server units containing advanced AI chips during an October 15 inspection in Kuala Lumpur International Airport’s free-trade zone, according to the South China Morning Post. The servers — falsely declared as electronic components and shipped through Malaysia so there would be no restrictions on export to the next destination, according to customs cited by the paper — were linked to a syndicate already involved this year in five cases with a total value exceeding US$ 24 million, officials cited by TRT World said.
A US export licence allows Firebird to ship 45,000 Nvidia Grace Blackwell graphics processing units (GPUs) to Kazakhstan, according to Nvidia. OilPrice reports that the chips are bound for Ekibastuz.
The US Treasury Department proposed expanding the Outbound Investment Security Program, currently limited to China, including Hong Kong and Macau, to cover North Korea, Russia, Iran and Cuba, according to the Korea Herald. Certain investments by US persons in sensitive technologies would be prohibited, while other transactions would require notification. Covered sectors include semiconductors, microelectronics, artificial intelligence (AI), quantum information technologies, high-performance computing and supercomputing.
The computing capacity available to Chinese entities for artificial intelligence (AI) is expected to double in 2027, from about 4.6 million to 8.8 million H100-equivalents, a metric benchmarked against Nvidia's data-center graphics processing units, according to the South China Morning Post (SCMP). US-designed chips will account for about 65 per cent of that capacity. Remote access through offshore clouds is expected to swell from roughly 1 million this year to 2.8 million H100-equivalents by the end of 2027.
The US Department of Energy (DOE) announced that Energy Secretary Chris Wright has designated helium as a critical material under the Energy Act of 2020, stating that its unique properties make it indispensable to semiconductor production, with no viable substitutes in many critical applications. According to US News, the designation means more federal financing will be available for helium producers and the government will monitor it more closely.
Deepening “symbiosis” with artificial intelligence model start-up DeepSeek is helping Huawei Technologies catch up with Nvidia’s chip hardware and software ecosystem, according to a new SemiAnalysis report cited by the South China Morning Post. The report says DeepSeek is likely to rely “primarily” on Huawei’s “already mature and well developed” technology stack by 2027 or 2028, in a scenario that could mean Jensen Huang’s “nightmare is starting to come true”.
Chipmaker Taiwan Semiconductor Manufacturing Company (TSMC) is exploring alternatives to rare-earth components in factory equipment, according to three people familiar with the matter cited by Reuters, in a text republished by WHBL, aiming to reduce the risk that Beijing's export controls could disrupt operations at plants that make the world's most advanced chips.
The discussions remain at an early stage and have not yet resulted in a broad rollout of rare-earth-free equipment, according to two of those people.
Representative Jim Himes, the top Democrat on the House Intelligence Committee, said restrictions on advanced chips are Washington’s key lever to keep its lead in artificial intelligence over China, the South China Morning Post reported. He said he did not believe the US would hold that edge in frontier AI indefinitely. “And the one real lever we have there, of course, is just slowing the dispersion of advanced chips into China,” he said.
ByteDance, Alibaba and Tencent have ordered more than 3 million of Nvidia’s new RTX PRO 5500 graphics cards, with a collective sticker price of over $35 billion, according to two people familiar with the matter cited by Reuters in a text republished by The Business Times. Nvidia has told customers and partners it cannot guarantee that it can fill the orders because demand from Chinese technology companies far exceeds planned production.
Chinese chip and tech firms flagged as much as six-fold growth in unaudited third-quarter profits, according to the South China Morning Post (SCMP), on huge demand for home-grown processors and artificial intelligence infrastructure.
GlobalFoundries will build chip facilities under an investment deal between the United States and Japan, according to Kyodo News. Japan will invest about $21 billion in up to two chip factories in New York, jointly owned by the US government and Japanese interests and operated by GlobalFoundries, according to Reuters, in a text republished by WHTC. The first factory is expected to add annual US manufacturing capacity of about 900,000 300 mm wafers when fully operational, according to the same report.
ASML's sales to China surged 109 per cent in the third quarter compared with the previous three months, according to financial results released on Wednesday. The Dutch semiconductor equipment maker expects China to account for about 20 per cent of its total sales this year, after shipments to the market more than doubled in the previous quarter despite US pressure for tighter export restrictions on less advanced chipmaking machinery.
The US Department of Justice accused the chief executive of Earthmade Computer, 38-year-old Greg Lui, of using false paperwork to mask shipments of servers worth more than $300 million that he knew were destined for China, and of conspiring with freight-forwarding firms in countries such as Malaysia and Singapore to divert chip shipments into China, including Nvidia's A100 and H100 graphics processors, according to Ars Technica. He faces charges of conspiracy to violate US export controls, outbound smuggling and conspiracy to commit money laundering, according to Bloomberg, in a text republished by Yahoo.
Taiwan’s deputy foreign minister, Francois Wu, said on Saturday that providing weapons to Taiwan to protect its freedom and provide semiconductors to the world is in the United States’ national interest, according to Reuters, in a text republished by The Straits Times. Xi Jinping told Donald Trump in Washington on Thursday that he hoped the United States would adhere to the “correct position” of opposing Taiwan independence. The United States ambassador to China, David Perdue, told CNBC on Friday that there has been no change in Trump’s Taiwan policy, including the 1979 Taiwan Relations Act, which mandates arms sales to the island.
T-Head, Alibaba’s semiconductor division, unveiled on Tuesday, September 22, 2026, at the Apsara conference in Hangzhou the Zhenwu V900 artificial intelligence (AI) accelerator, which the company describes as China’s most powerful AI chip by computing performance and which is scheduled for mass production in the first quarter of 2027, according to TechNode and TrendForce. According to the maker, the V900 has 216 GB of memory, 1,200 GB/s of chip-to-chip interconnect bandwidth and native FP8 and FP4 support, with a claimed performance three times that of its predecessor, the M890; clusters of up to 500,000 units, linked by T-Head’s own ICN Switch technology, would be able to train models of 5 trillion to 10 trillion parameters. Alibaba CEO Eddie Wu also set a target of lifting Alibaba Cloud’s global data centre capacity above 20 GW by 2032, backed by more than US$ 53 billion of investment over three years. The V900 figures are the maker’s own claims, with no independent benchmarking, and the capacity targets describe announced expansion, not contracted capacity.
Alibaba unveiled its new Zhenwu V900 artificial intelligence (AI) chip on Tuesday, September 22, 2026, describing it as the most powerful in China with three times the performance of its predecessor, and set a target of lifting Alibaba Cloud’s global data centre capacity above 20 gigawatts (GW) by 2032, according to Reuters and Mint, which cites the group’s official release. The announcement was made by company CEO Eddie Wu at the Apsara conference in Hangzhou, where Alibaba also laid out a plan to train Qwen models with 5 trillion to 10 trillion parameters on its path towards artificial superintelligence, and launched Qwen Intelligence for mobile devices, an end-to-end tool meant to run complex tasks on the handset itself. According to Mint, the chip, built by the group’s semiconductor arm, can be clustered at up to 500,000 units per cluster, as the company invests across three declared fronts: models, chips and cloud infrastructure.
Nvidia and other chipmakers are pressing Congress to strip from the National Defense Authorization Act (NDAA) provisions that would restrict China’s access to their most advanced chips, as the debate over artificial intelligence (AI) safety reshapes the politics of the issue in Washington, according to POLITICO. The Senate draft, which has not yet passed the chamber and is likely to go straight to negotiations with the House after the midterms, is expected to require tracking of exported chips, let Congress block exports of advanced AI chips to China and impose new controls on chipmaking equipment. “What Jensen argues is that the threat of China is greater if they have to develop their own chips, and that’s not a frivolous argument,” Senator Ted Cruz, who chairs the Senate Commerce Committee, told POLITICO. Mark Warner, the top Democrat on the Senate Intelligence Committee, said he is not sure he fully buys Huang’s argument that worldwide reliance on Nvidia chips is a good thing, and that the place where the company is gaining the most influence is the White House. A Senate Republican aide, granted anonymity to describe private conversations, said the administration’s slow pace on export controls is putting Republican senators under pressure to act and fill the vacuum in the defence package. Nvidia said in a recent earnings call that less than 1 percent of its data centre revenue came from certain high-end chips shipped to customers in China, a figure that has not reassured critics. Jensen Huang, the company’s chief executive, is expected to be a guest at the State Dinner in honour of Xi Jinping this week.
Huawei presented in this cycle HarmonyOS 7 reorganised around agents, the Atlas 960E supercomputer with optical interconnect and a map of “agentic” computing running from supernodes to clusters, according to OSCHINA and TudoCelular.
Taiwan broke ground on an advanced semiconductor packaging park anchored by TSMC, according to Reuters.