新云社 update: Google said its latest tensor processing units are the TPU 8TX and TPU 8I MAX, expanding its eighth-generation TPU portfolio. The company disclosed the two chip models without providing performance figures, pricing or a release date. The 8TX and 8I MAX join Google’s eighth-generation lineup as the company continues to build out its custom silicon for AI workloads.
新云社 reports: Alphabet shares rose 2.5% in premarket trading while Nvidia fell more than 6% after reports that Google will launch two new versions of its TPU systems in a challenge to Nvidia. The premarket moves pit Google’s custom silicon against Nvidia’s dominance in AI chips.
新云社 update: SWANCOR Advanced Materials (上纬新材) said its consumer-grade embodied intelligent robot business contributes only a small share of total revenue, after its shares rose more than 100% over a recent stretch. The company noted its latest price-to-earnings ratio is well above the 19.57x rolling average for the chemical raw materials and chemical products sector over the past month. It said the consumer robot industry is still in an early stage, with home-scenario applications nascent and sustainable end-market demand yet to be verified.
新云社 update: Apple has received approval to launch its AI features on iPad and MacBook devices in China, clearing a key regulatory hurdle for its Apple Intelligence rollout in the country. The approval covers AI functionality on the two device lines, though no launch date or specific feature set was disclosed. China’s approval process is a prerequisite for AI services operating in the market.
新云社 update: ByteDance on November 9 released Seedream 5.5, its Doubao image-creation model, which generates images substantially faster than the previous Seedream 5.0 Pro. The model is now live across Doubao, Doubao Work and Jimeng, with an API to follow on the Volcano Ark platform.
新云社 update: Nvidia’s board has formed a committee of independent directors and hired outside legal counsel to review matters concerning CEO Jensen Huang, while stating it still trusts Huang and his management team. Huang, who holds about 3.58% of shares and has led Nvidia for over 33 years, has no publicly named successor. The board’s 11 members include 10 independents; Nvidia reports earnings Nov. 17 and Huang faces a closed-door House briefing Nov. 18.
新云社 update: The Aerospace Information Research Institute of the Chinese Academy of Sciences and domestic tech companies unveiled the “Nebula Constellation Plan” on May 9 at the 2026 Space Cloud Industry Ecosystem Conference, marking a shift from technical validation to industrial deployment of China’s space-cloud ecosystem. The plan spans the full chain from domestic computing chips to satellite applications and will build a unified space-ground computing network serving natural resources, disaster response and low-altitude economy scenarios.
新云社 update: Nvidia said it plans to make a new investment in South Korean startups, with the first target being Korean AI company Point2. The company disclosed the plan without giving the size of the investment or a timeline. Point2 is described as an AI startup based in South Korea.
新云社 reports: Sakana AI plans to hold an initial public offering as early as autumn 2027. The Japanese AI developer has not disclosed a listing venue, target valuation or offering size. No further details of the IPO timetable were provided.
新云社 reports: SoftBank is reportedly reconsidering its NeoCloud business in the United States and may shift the operation to other regions. The reported review concerns the Japanese group’s US cloud infrastructure arm, with no decision or timeline disclosed. NeoCloud refers to next-generation cloud computing services. SoftBank has not publicly commented on the reported rethink.
新云社 reports: SoftBank is preparing for a possible slowdown in US data center demand. The move signals that the Japanese conglomerate is bracing for weaker growth in the buildout of AI infrastructure capacity in the United States. No further details on the scale or timing of the preparations were provided.
新云社 reports: Vietnam is seeking to deepen cooperation with Japan in artificial intelligence and semiconductors. The move signals Hanoi’s push to expand ties with Tokyo in two strategic technology sectors. No further details on specific agreements, timelines or participants were immediately available.
新云社 update: Yat Siu will become CEO of Animoca Brands while retaining his executive chairman role, the company said. Group president Evan Auyang will move to senior advisor to the board as part of a push into AI and institutional business. The leadership change was announced without a date for when Siu’s CEO appointment takes effect.
新云社 update: Qwen Office released version 2.0 on November 9, its largest upgrade since launching three months ago. The release introduces “Collaboration,” described as the industry’s first feature supporting multi-user and multi-agent teamwork, alongside Human-AI Co-creation, Material Space, Auto Tasks and Tri-device Connectivity. The update moves the Agent from serving individuals to serving teams and organizations, and all features are now fully available in the Qwen Office client.
新云社 update: China has become the first country to ratify the agreement establishing the World AI Cooperation Organization, Foreign Ministry spokesperson Guo Jiakun said at a regular press briefing on November 9. As the organization’s initiator and host country, China said the move reflects its sincere intent and firm commitment to advancing the body’s preparations. Beijing will push forward the organization’s formal operation and focus on practical cooperation and capacity-building for Global South countries.
新云社 update: Alibaba Cloud on Monday raised the monthly credit quotas on its Token Plan personal subscriptions by roughly 74% to 78% while keeping tier prices unchanged. The Lite tier rises from 11,500 to 20,000 credits at 39 yuan a month, Essential from 25,500 to 45,000 credits at 79 yuan, and Standard from 45,000 to 80,000 credits at 139 yuan.
新云社 update: Nvidia CEO Jensen Huang sent an apology letter to all employees on Nov. 8, acknowledging marital infidelity after reports linked him to a woman identified as Huang Le, a former Megaspeed executive. Three Republican House committee chairmen had written to Huang on Nov. 4 demanding he appear for a closed-door briefing by Nov. 18 on his contacts with Huang Le and Nvidia’s safeguards against foreign intelligence activity. Nvidia’s board has formed an independent committee and hired outside counsel to investigate.
新云社 reports: Evan Van Ness, founder of Ethereum Weekly, said in a recent interview that Pearl, a proof-of-useful-work L1 project, could surpass Bitcoin. Pearl mines through matrix multiplication, the same computation used in AI model training and inference, and can run it alongside inference at near-zero extra cost, he said. Van Ness cited Bitcoin’s early ossification and security-budget problems, arguing Pearl turns otherwise useless mining into useful computation that creates a flywheel lowering inference costs.
新云社 update: Elon Musk replied “I agree” to Sequoia Capital partner Shaun Maguire’s post calling Nvidia CEO Jensen Huang one of the greatest entrepreneurs of this generation. Maguire said Huang made an extremely foolish move that is bad for the US but likely without malice, adding that two mistakes do not cancel each other out and Huang remains irreplaceable to US AI leadership. Musk did not address the individual points, only expressing agreement.
新云社 update: Elon Musk said he held an in-depth review with the Grok Bot engineering team and that repair work is progressing, with users expected to see significant improvements by tomorrow. He had earlier said Grok Bot is being optimized to improve token-use efficiency and make it perform more intelligently.